Slack Export - #general

Tools for detecting wildlife in aerial images using active learning - microsoft/aerial_wildlife_detection

Original URL: https://github.com/microsoft/aerial_wildlife_detection

👍 Sara Beery, Mikey Tabak

Lukas Liebel (lukas.liebel@tum.de)

2020-06-10 04:23:22

Hi there, we recently finished a project for segmenting invasive plants from UAV RGB images with around 3.8 mm ground-sampling distance. A student of mine did a comprehensive study also on different networks. We didn't work with detection networks though. Still, let me know if you have any other questions. Maybe I can share some insights. We're also looking to publish this work in the next time. Code should be available soon as well!

🌿 Sara Beery

👍 Mikey Tabak

Mikey Tabak (tabakma@gmail.com)

2020-06-11 14:16:40

Thank you @Benjamin Kellenberger and @Lukas Liebel!

Mikey Tabak (tabakma@gmail.com)

2020-06-11 14:17:33

and @Ben Weinstein!

Mikey Tabak (tabakma@gmail.com)

2020-06-30 10:22:49

Another followup question on analyzing drone images: What type of packages/code have y'all used to take aerial photos that are georeferenced, crop them, detect objects (or segment pixels), and then put them back into a georeferenced format. This seems like a pretty big challenge that I assume others have addressed.

Lukas Liebel (lukas.liebel@tum.de)

2020-06-30 10:57:03

You mean, something like 'rasterio' to read and write georeferenced images with python?

👍 Mikey Tabak

Ben Weinstein (benweinstein2010@gmail.com)

2020-06-30 11:56:48

I’ve written a few gists around this idea, you’ll need to be a bit more specific, but this will give you the flavor https://gist.github.com/bw4sz/e2fff9c9df0ae26bd2bfa8953ec4a24c

👍 Mikey Tabak

Lukas Liebel (lukas.liebel@tum.de)

2020-06-30 12:15:36

As far as cropping patches for training from a bigger image or mosaic is concerned, windowed reading/writing using rasterio is very helpful. It also allows you to easily save prediction results/labels/whatever (basically everything that can be converted to a numpy array and, in the easiest case, is of the same shape as your image data) to georeferenced tiffs. A simple workflow could be:

open image as dataset

read metadata (shape, channels, georef)

read a patch from a specified position in the image (saves you from reading an enormous image to your poor RAM)

predict that patch

repeat these two steps until you accumulated all of your desired predictions in a numpy array (same size as the original image given by the image's metadata)

open a new dataset to output the predictions (e.g., a new GeoTIFF file)

copy the metadata from the image to this new dataset (you'll probably want to adjust the number of bands/channels) including the georeference

write the numpy array to this new dataset

done. You successfully generated a fully georeferenced prediction for your image. The only thing left to do is to develop your magic black box predictor 😛

👍 Mikey Tabak

Ben Weinstein (benweinstein2010@gmail.com)

2020-06-30 12:17:19

I just wrote some of this logic for wrapping into tensorflow records for nice fast training. beware large sizes on HDD. Much as @Lukas Liebel alludes to https://github.com/weecology/DeepTreeAttention/blob/6361e4c5eda907d83643cfd789c0bef78be3fe2a/DeepTreeAttention/generators/make_dataset.py#L53

GitHub

Implementation of Hang et al. 2020 "Hyperspectral Image Classification with Attention Aided CNNs" for tree species prediction - weecology/DeepTreeAttention

Original URL: https://github.com/weecology/DeepTreeAttention/blob/6361e4c5eda907d83643cfd789c0bef78be3fe2a/DeepTreeAttention/generators/make_dataset.py#L53

👍 Lukas Liebel, Sara Beery, Mikey Tabak

Mikey Tabak (tabakma@gmail.com)

2020-06-30 16:16:34

Thank you both for your responses! I was hoping to use rasterio (@Ben Weinstein it looks like you're using this too) but I haven't been able to get it installed. I'm (unfortunately) using Windows now and the installation is problematic. I'm hoping they'll respond to my question about it on their github page. My general plan is to get large aerial image files, split them up while preserving the georeferencing into chips, have technicians paint masks (for object segmentation) on some of the images, train a model on these annotated images, use the model on non-annotated images, and reassemble the chips into one georeferenced file, so I can have the locations of the objects in the images.

Mikey Tabak (tabakma@gmail.com)

2020-06-30 16:17:19

And @Lukas Liebel, yes I'll be doing this in Python

Lukas Liebel (lukas.liebel@tum.de)

2020-07-02 09:39:08

@Mikey Tabak Well, your plan definitely sounds doable 🙂 In general, GDAL might also help you (either the standalone command line tool or the python package). Especially when splitting the training images into tiles it may be more convenient than re-implementing your own routine in python. Both ways should work with minimal effort though.

Lukas Liebel (lukas.liebel@tum.de)

2020-07-02 09:42:14

@Mikey Tabak Regarding the installation of rasterio: In my experience (on Linux machines) it was always just a matter of a simple 'pip install rasterio' call. I'd suggest to use Anaconda. Chances are rasterio is also available as a conda package and Anaconda also comes with pip on Windows I guess. Another workaround could be using a Docker container. Even though this is kind of an extra mile to go, you may want to use it for training or other future tasks at some point anyway. I can help you with a minimal setup if you want to try that. It's dead simple for most of our daily applications but requires quite some Googling if you're not familiar with the terms and typical problems.

Mikey Tabak (tabakma@gmail.com)

2020-07-07 16:12:52

Thanks @Lukas Liebel I'm still working on getting rasterio installed and nothing is working. I might use R instead for this task. It interfaces really well gdal.

Mikey Tabak (tabakma@gmail.com)

2020-07-07 16:15:00

@Benjamin Kellenberger Do you have an estimate of when your AIDE software will support semantic segmentation? I don't need to use the software to run the AI model, I just want to be able to have technicians "paint" the images and then be able to access the annotations. The software is really impressive for its other functions!

➕ Sara Beery

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2020-07-08 02:36:43

@Mikey Tabak Thanks a lot for considering our software! 🙂 AIDE in its current version (v2) fully supports semantic segmentation and even has U-Net built-in! Annotation is still a bit manual (without tools like Photoshop’s magic wand built-in), but we’ll be adding those as we move along.
You can find a demo instance of it (v1) here.
Feel free to let me know if you have more questions, wishes for functionality, or if you find errors.

👍 Mikey Tabak, Sara Beery

Mikey Tabak (tabakma@gmail.com)

2020-07-08 11:06:26

Thanks @Benjamin Kellenberger! I got segmentation running on my project. I wasn't configuring it properly and that's why I didn't see the interface for the segmentation. This is the perfect tool for what I'm trying to do. I have some little questions about it, but I'm going to put these on the github page as others will likely have similar questions.

👍 Benjamin Kellenberger

Ben Weinstein (benweinstein2010@gmail.com)

2020-06-09 12:49:07

To keep the brainstorming + collaboration going (i’m really glad to see that this channel has gotten so active, thanks @Sara Beery and A14Earth for $). If anyone out there is wanting to work with me on hyperspectral image classification. I am creating an open-source implementation of this great paper (Hang et al. 2020 “Hyperspectral Image Classification with Attention Aided CNNs”) here: https://github.com/Weecology/DeepTreeAttention. Only 1 week in. All are welcome. I’m going to recreate their results using open source benchmarks.

GitHub

Cornell Birdcall Identification

Implementation of Hang et al. 2020 "Hyperspectral Image Classification with Attention Aided CNNs" for tree species prediction - weecology/DeepTreeAttention

Original URL: https://github.com/Weecology/DeepTreeAttention

🌳 Sara Beery, gvanhorn, Jonathan Granskog, Omiros Pantazis

Ben Weinstein (benweinstein2010@gmail.com)

2020-06-09 12:50:01

here is the model architecture

Holger Klinck (hk829@cornell.edu)

2020-06-16 07:16:35

Our Kaggle competition is live. If you have any questions about the competition, please let me know! https://www.kaggle.com/c/birdsong-recognition/overview

kaggle.com

Build tools for bird population monitoring

Original URL: https://www.kaggle.com/c/birdsong-recognition/overview

👍 Oisin Mac Aodha, gvanhorn, Sara Beery, Jonathan Granskog, David Rolnick, Ben Weinstein, Omiros Pantazis, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

😎 Jon Van Oast

Sara Beery (sara.beery@gmail.com)

2020-06-16 11:15:07

New funding opportunity, passed along by Serge Belongie: Citizen Science for Earth Systems Program (CSESP)

🎉 Jon Van Oast

Sara Beery (sara.beery@gmail.com)

2020-06-16 11:15:27

If you have questions, Questions concerning Citizen Science for Earth Systems Program may be directed to Kevin Murphy, who may be reached at kevin.j.murphy@nasa.gov and Gerald "Stinger" Guala, who may be reached at gerald.f.guala@nasa.gov.

ROSES-20 Amendment 32: A.41 Citizen Science for Earth Systems Program Final Text and Due Dates Released

The primary goal of the Citizen Science for Earth Systems Program (CSESP) is to develop and implement capabilities to augment and enhance NASA scientific data and capacity through voluntary observations, interpretations, or other direct participation by members of the general public to advance understanding of the Earth as a system. The program complements NASA's capability of observing Earth globally from space, air, land, and water by engaging the public in NASA's strategic goals in Earth Science (see https://science.nasa.gov/about-us/science-strategy).

ROSES-2020 Amendment 32 releases final text and due dates for A.41 Citizen Science for Earth Systems Program. Mandatory Notices of Intent are due August 4, 2020, and proposals are due September 11, 2020.

On or about June 12, 2020, this Amendment to the NASA Research Announcement "Research Opportunities in Space and Earth Sciences (ROSES) 2020" (NNH20ZDA001N) will be posted on the NASA research opportunity homepage at http://solicitation.nasaprs.com/ROSES2020 and will appear on SARA's ROSES blog at: https://science.nasa.gov/researchers/sara/grant-solicitations/roses-2020/.

Stefan Schneider (sschne01@uoguelph.ca)

2020-06-16 12:00:50

Hi all! Tomorrow, Wednesday June 17th at 1:00pm, I'll be defending my PhD on Deep Learning for Animal Re-ID. The presentation portion is open to the public and all interested individuals are encouraged to attend! It can be accessed at this Zoom link which you can open in any browser:

https://us02web.zoom.us/j/84071151383

The link opens at 12:45pm EST and entry will be denied after 12:55pm EST, so you gotta be on time. Look forward to having anyone that wants to check it out!

🎉 Sara Beery, gvanhorn, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Manish Rai, Malte Pedersen, Siyu Yang

🦁 Sara Beery

Malte Pedersen (mape@create.aau.dk)

2020-06-17 05:15:08

Hi all,
I hope it's okay that I advertise a bit for our new multiple object tracking challenge of zebrafish in 3D, which is now live on https://motchallenge.net/data/3D-ZeF20
Behavioral studies of zebrafish can help biologists understand the impact of harmful substances and materials, raising temperatures, and more in our marine environments and by publishing this dataset we hope that we can spark the interest in this relatively unknown field.
You can read more on our project page: vap.aau.dk/3d-zef
And in our CVPR2020 paper: 3D-ZeF: A 3D Zebrafish Tracking Benchmark Dataset

Cheers 🐟

motchallenge.net

MOT Challenge - Data

Multiple Object Tracking: Datasets, Benchmarks, Challenges and more.

Original URL: https://motchallenge.net/data/3D-ZeF20

👍 gvanhorn, Sara Beery

🐟 Sara Beery, Megan Cromp

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2020-06-19 18:06:27

XPRIZE has a $10mil rainforest competition going on: https://www.xprize.org/prizes/rainforest

Rainforests cover less than 10% of the earth's land surface, but they house approximately 50 million inhabitants and over 50% of the planet's biodiversity. Although they are the most biodiverse ecosystems, there is a limited knowledge of everything that lives in these iconic environments. The value of the standing trees are not fully understood and our ability to assess this value is restricted because the rainforest environment is dense, vast, and complex.

The winning team will develop novel technologies to rapidly and comprehensively survey rainforest biodiversity and use that data to improve our understanding of this complex ecosystem.

XPRIZE

Rainforest XPRIZE)

The $10M Rainforest XPRIZE is a five-year competition to enhance our understanding of the rainforest ecosystem.

Original URL: https://www.xprize.org/prizes/rainforest

🌴 Sara Beery, Omiros Pantazis, David, Ankita Shukla

Hannah Kerner (hkerner@umd.edu)

2020-06-24 16:56:18

Hi all, we are looking for a postdoc in the NASA Harvest group at UMD College Park. Research topic is ML and EO applications for cropland & crop production forecasting in smallholder agriculture (field to national scales) focusing on Sub-saharan Africa. Please apply or share with potential applicants! Thank you! https://ejobs.umd.edu/postings/78351

ejobs.umd.edu

Post-Doctoral Associate

The University of Maryland's Center on Global Agricultural Monitoring Research is seeking an outstanding researcher at the Post-doctoral associate or Assistant Research Professor with strong interest in machine learning and agriculture to join a diverse team working on satellite remote sensing applications for agricultural monitoring and food security, within the framework of the NASA Harvest Program and an on-going NASA SERVIR Applied Science Team project, led by UMD’s Center for Global Agricultural Monitoring. Harvest is NASA’s Food Security and Agriculture program, focused on advancing the use of earth observations applications for food security and stable agricultural markets, with a diverse set of over 40 partners from academia, government, private, NGO and humanitarian sectors (Program Website: <http://www.nasaharvest.org>). The successful candidate will work on research related to machine learning applications for crop production forecasting for smallholder agricultural systems at the field to national scales with focus on Sub-saharan Africa. This will involve developing models to map cropland, crop types, forecast crop yields and alert of impending crop shortfalls, to name a few, in order to inform key agricultural and food security decisions by a range of public and private stakeholders and develop training materials . This research will be carried out through the use of a wide range of satellite data, unique ground collected data-sets, global archives of diverse socio-economic data and statistics. The successful applicant should hold a PhD in computer science, remote sensing, agricultural sciences, physics, engineering, mathematics, or related fields. A strong programming background (especially Python, R, IDL, or C++) and an interest in agriculture and food security research and applications is required and experience with working the Google Earth Engine is a plus. The candidate will be expected to work well within a diverse team to design and lead projects that will contribute to the overall aim of the Harvest Program as well as work on ongoing activities. Applications from women and minorities are particularly sought. The University of Maryland is an Equal Opportunity Affirmative Action Employer.

Original URL: https://ejobs.umd.edu/postings/78351

🌍 Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, gvanhorn, Laurel Hopkins

Sara Beery (sara.beery@gmail.com)

2020-06-24 17:04:44

@Caleb Robinson, @Elijah Cole (Deactivated) you guys might know someone?

Caleb Robinson (calebrob6@gmail.com)

2020-06-24 17:54:54

Hi @Hannah Kerner! I don't know of anyone with an EO background looking for postdoc unfortunately

Caleb Robinson (calebrob6@gmail.com)

2020-06-24 17:55:39

I do know someone looking to apply to a CS PhD program

👍 Sara Beery

Caleb Robinson (calebrob6@gmail.com)

2020-06-24 17:56:26

I'll check back with them in 4-6 years 👍

😅 Sara Beery, Hannah Kerner

Hannah Kerner (hkerner@umd.edu)

2020-06-25 09:22:28

Candidates can also have an ML background but no EO background yet, as long as that's a direction they're interested in!

Ben Weinstein (benweinstein2010@gmail.com)

2020-06-28 10:47:34

@Hannah Kerner this just popped up on my scholar feed if useful (https://www.frontiersin.org/articles/10.3389/fenvs.2020.00085/full)

Frontiers

The Potential of Sentinel-2 for Crop Production Estimation in a Smallholder Agroforestry Landscape, Burkina Faso

Crop production statistics at the field scale are scarce in African countries, limiting potential research on yield gaps as well as monitoring related to food security. This paper examines the potential of using Sentinel-2 time series data to derive spatially explicit estimates of crop production in an agroforestry parkland in central Burkina Faso. This type of landscape is characterized by agricultural fields where cereals (millet and sorghum) and legumes (cowpea) are intercropped under a relatively dense tree canopy. We measured total above ground biomass (AGB) and grain yield in 22 field plots at the end of two growing seasons (2017 and 2018) that differed in rainfall timing and amount. Linear regression models were developed using the in situ crop production estimates and temporal metrics derived from Sentinel-2 time series. We studied several important aspects of satellite-based crop production estimation, including (i) choice of vegetation indices, (ii) effectiveness of different time periods for image acquisition and temporal metrics, (iii) consistency of the method between years, and (iv) influence of intercropping and trees on accuracy of the estimates. Our results show that Sentinel-2 data were able to explain between 41 and 80% of the variation in the in situ crop production measurements, with relative root mean square error for AGB estimates ranging between 31 and 63% in 2017 and 2018, respectively, depending on temporal metric used as estimator. Neither interc...

Original URL: https://www.frontiersin.org/articles/10.3389/fenvs.2020.00085/full

Hannah Kerner (hkerner@umd.edu)

2020-06-29 09:00:27

thanks for sharing, definitely relevant!

Sara Beery (sara.beery@gmail.com)

2020-06-25 10:13:52

I'm giving a WILDLABS "Tech Tutorial" this morning (8am PST) on getting started using ML for camera traps. It's geared more towards an ecology/non-ML audience so I didn't think to post it here, but if anyone is interested you may still be able to register! https://www.eventbrite.co.uk/e/tech-tutors-how-do-i-get-started-using-ml-for-my-camera-traps-tickets-109658752280?ref=estw

If you can't register, the recording will be posted on YouTube tomorrow :)

Eventbrite

Tech Tutors: How do I get started using ML for my Camera Traps?

The seond episode of WILDLABS Tech Tutors will stream at 11 AM ET/ 4 PM BST on Thursday, June 25th

Original URL: https://www.eventbrite.co.uk/e/tech-tutors-how-do-i-get-started-using-ml-for-my-camera-traps-tickets-109658752280?ref=estw

👍 Oisin Mac Aodha, Nathaniel Rindlaub, Elijah Cole (Deactivated), Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Tee R. (taylornroberts2009@gmail.com)

2020-06-25 13:12:29

Hey all! Newbie here, wanting to learn more about wildlife and AI!

❤️ Sara Beery

👍 Oisin Mac Aodha, Benjamin Kellenberger

👋 Jon Van Oast, Siyu Yang

Sara Beery (sara.beery@gmail.com)

2020-06-25 13:12:57

Welcome!

Jon Van Oast (jon@wildme.org)

2020-06-25 14:30:25

glad to have you here

Sara Beery (sara.beery@gmail.com)

2020-06-26 13:46:51

Context R-CNN is on the Google AI blog today! https://ai.googleblog.com/2020/06/leveraging-temporal-context-for-object.html

Google AI Blog

Leveraging Temporal Context for Object Detection

Posted by Sara Beery, Student Researcher, and Jonathan Huang, Research Scientist, Google Research Ecological monitoring helps researcher...

Original URL: https://ai.googleblog.com/2020/06/leveraging-temporal-context-for-object.html

👍 gvanhorn, Benjamin Kellenberger, Elizabeth Bondi, Hannah Kerner, Zac Winzurk, Chris Yeh

😍 Chris Yeh, Jon Van Oast

Drew Gray (drewjgray@gmail.com)

2020-06-30 18:12:22

Hey everyone! I just joined, thanks for the invite @Sara Beery. I have just started a non-profit called Aqualink. We are building a global ocean monitoring system with connected buoys measuring the temperature at coral reefs. A few of our flagship sites will have live streaming camera.. coral cams 🙂 Once we are up and running we are looking to build some models for coral reef health analysis through computer vision and machine learning. Let me know if you would like to learn more or get involved! We will be open-sourcing all of our data and models as well.

🐚 Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, David Healey, Elizabeth Bondi

🐙 Lukas Liebel, Tee R.

👍 Carly Batist

Ben Weinstein (benweinstein2010@gmail.com)

2020-06-30 20:43:26

Hi @Drew Gray, perhaps you know them already but https://www.nature.com/articles/srep23166 the scripps, san diego groups did alot of work here in the last decade, it would be good to hear from them.

Scientific Reports

Improving Automated Annotation of Benthic Survey Images Using Wide-ban

Improving Automated Annotation of Benthic Survey Images Using Wide-band Fluorescence

Original URL: https://www.nature.com/articles/srep23166

👍 Sara Beery

Drew Gray (drewjgray@gmail.com)

2020-07-01 12:04:42

Thanks @Ben Weinstein!

Bob Zak (robert.zak@comcast.net)

2020-07-01 19:42:51

Hi, all. Another new join -- thanks @Sara Beery ^2. Post retirement, I'm looking at an "encore" career in outreach and education for wildlife conservation based on trail cameras with my wife. Nothing too big, yet -- a blog and Janet's book -- but with with aspirations for using AI technology to improve usefulness of cameras in capturing compelling photos and videos of wildlife behavior.

🦊 Sara Beery, Elizabeth Bondi, Lukas Liebel, Omiros Pantazis, Jonathan Granskog

👍 Alasdair Davies

Sara Beery (sara.beery@gmail.com)

2020-07-01 19:57:05

@Sam Kelly I talked to Bob earlier, and he's interested in edge-based solutions. You two should talk!

Bob Zak (robert.zak@comcast.net)

2020-07-01 20:10:43

I am. I emailed @Sara Beery about porting Megadetector to TFLite for embedded SOCs -- something that appears a overly ambitious for the current state of the art -- still, we have what's left of Moore's Law 🙂 Another area I've thought about is whether we can improve the PIR sensor accuracy using AI -- perhaps leveraging new Si aimed at "always on " audio processing. Very interested to hear what you are up to.

Sara Beery (sara.beery@gmail.com)

2020-07-01 20:12:56

I think lightweight detection is totally possible and even useful! It's just not as accurate, which is why we kept the MegaDetector a larger model.

Bob Zak (robert.zak@comcast.net)

2020-07-01 20:15:49

Oops -- I misunderstood. I need to understand accuracy tradeoffs more. In app space I have in mind, camera location independence is key. In any case, I should start small to limit the amount of trouble I get into

Sara Beery (sara.beery@gmail.com)

2020-07-01 20:17:22

I think the PIR sensor accuracy seems really interesting too!

Sam Kelly (sam@conservationxlabs.org)

2020-07-02 10:47:33

Hi @Bob Zak! Would love to chat to you about this! We have been thinking about some of this for a while at Conservation X Labs! Definitely some things I would love to discuss with you!

❤️ Sara Beery

Bob Zak (robert.zak@comcast.net)

2020-07-02 12:56:59

@Sam Kelly Great! I just updated my profile with a skype address; (updated) phone in profile also works. I'm around now through the afternoon.

Jonathan Granskog (jonathan.granskog@gmail.com)

2020-07-06 14:28:57

I love this idea and it’s something I’ve been thinking about too. Maybe you could look into using NVIDIA’s Jetson Nano or Xavier NX as well.

Bob Zak (robert.zak@comcast.net)

2020-07-06 15:40:45

Yeah -- there seem like there might be a lot of hardware options. But (as a newb to the details) I have basic questions about MD model size; extent that it's been optimized to reduce said size; tolerance to thinks like int8 mapping; etc. I guess another option would be to start with same training set, but target a smaller (more edge friendly) model to start with. So many questions...

👍 Sam Kelly, Jonathan Granskog

Sam Kelly (sam@conservationxlabs.org)

2020-07-06 15:51:45

lots of awesome innovation happening in this space -
On the hardware front: https://github.com/basicmi/AI-Chip
We are attempting what you are talking about with a YOLO-tiny / Darknet (or similar) base model with limited classes and transfer learning approach with pruning/quantization included. Will be sure to share how it goes

GitHub

basicmi/AI-Chip

A list of ICs and IPs for AI, Machine Learning and Deep Learning. - basicmi/AI-Chip

Original URL: https://github.com/basicmi/AI-Chip

Bob Zak (robert.zak@comcast.net)

2020-07-06 16:28:19

So someone has already concluded that running current Inception-Resnetv2 MD (tuned, optimized, int8, etc.) doesn't work on one of these edge devices?

Ed Miller (ed@hypraptive.com)

2020-08-01 19:55:21

@Bob Zak and @Sam Kelly I am also interested in intelligent camera traps. There are SoCs that can run high accuracy models and those the can run really low power. The trick is to find the middle ground that makes sense. You can run Inception-Resnetv2 at the edge, but the power might be more than you bargin for. But that also depends on how frequently you run a full inference.

Bob Zak (robert.zak@comcast.net)

2020-08-01 21:00:05

On further investigation, and after conversation with @Sam Kelly, I'm convinced that the "intelligent camera" needs to start somewhere other than Megadetector. MD is solving a much harder problem -- locating any animal in any arbitrary trail camera photo, with no a priori context. Vs. at the edge, you have all sorts of context -- e.g. you know what the (animal-less) scene is, under any lighting condition. Intuitively, it seems like this should really simplify the detection problem. Which I think is critical -- even with low power hardware, my sense is that deep networks like MD are orders of magnitude more energy intensive than can be supported on batteries. Even worse, the most interesting edge camera application (I think) is likely video, and here you need to keep up with frame rate.

Ed Miller (ed@hypraptive.com)

2020-08-02 21:55:24

I agree MD, as it is, is too heavy for frame by frame detection at the edge. I would think you would have a cascade of models, starting with "something is out there", perhaps using motion detector or "smart PIR" as you mentioned in another thread. The fixed context of the camera, as you mentioned, should help with verifying there is an object of interest without having to run a complex model. Once you have a region of interest, you can run though a further cascade of models as needed:

You may want to classify at a coarse level, similar to MD: animal/person/vehicle. This would be beneficial if the MD-"lite" model is lighter weight than the species classifier, or if you want fast alerts for human/vehicle hits (possible poachers, etc.).

If you want species level classification, you can have models trained on the "common" species for the camera area.

If you want individual classification (like we do for the BearID Project) you only need to run when there's a species match.

You shouldn't need to classify every frame at each level. You can use some computer vision based object tracking algorithms (again the known background should make this easier) to follow the object through frames and only run the classifier on a few frames to make sure there is agreement. You could also run some of the models with lowered resolution to save cycles/energy. Also, not everything has to run "live".

You probably don't need the bits that say something interesting is there to run on every fame. Maybe you run a few at the beginning to make sure we want to record this, then rely on the object tracking to keep the keep recording until the object leaves the scene. You can run some of the other models in the background from the recording when nothing else is happening.

Any thoughts on a data backchannel such as LoRA or DASH7?

Bob Zak (robert.zak@comcast.net)

2020-08-03 07:38:41

A hierarchy of models definitely makes sense (although, depending on the size of each of the models, the energy cost of swapping the models around could be significant). In addition to the usual ai worries (training, accuracy, size, etc), I think you also have to worry about “value proposition” for the usage model relative to other solutions, especially considering energy costs. E.g., storage capacity, and energy per image is relatively cheap — if you can afford to wait for results, then doing all the processing offline makes a lot of sense. If you need the data urgently (e.g. for poaching detection, or “real-time” animal localization) then you have to consider the the alternative of sending thumbnail images via radio for backend processing — or, more generally, which models in the hierarchy to run locally, vs which to run in the backend. Two overall pieces of scaffolding would be good to have: an “energy budget” model (positing energy costs for acquisition, storage, processing, and wireless rx/tx). The other would be a list of usage cases and their key figures of merit. I’ve not looked at Lora or dash7, though I did poke around at some of the satellite startups. I also looked at rficient (https://www.iis.fraunhofer.de/content/dam/iis/en/doc/il/ics/ic-design/Datenblaetter/Factsheet_WakeUp_v4.pdf) as a potential technology for remote sensors (promising, still waiting on sample availability)

👍 Sam Kelly, Ed Miller, Ben Weinstein, Sara Beery

Sam Kelly (sam@conservationxlabs.org)

2020-08-03 11:02:12

@Bob Zak after our discussion, we actually tried to compress MD for on-device deployment, and unfortunately it is not technically possible with the current available libraries (some of the CNN operations are not available for tensorflow lite). The Yolo algorithms and EfficientNet seem to be the go-tos in this area. On the comms/backhaul - would be very interested to hear about this, and the animal tracking/tag world may be a good analogy - LTE/LoRa/Iridium/ARGOS/VHF. I know @Alasdair Davies has a module for Argos that I'm interested in.
@Ed Miller 100% agree with the cascade idea and keeping algorithms tailored to predicted (or study related) classes. I would be very interested to hear experts thoughts on MD-lite (animal/no animal) vs robust study-specific species detectors.

😎 Ed Miller

Alasdair Davies (alasdair@shuttleworthfoundation.org)

2020-08-03 11:30:49

@Bob Zak & @Sam Kelly @Ed Miller - super interesting thread here. I think we are all in the same realm of thought regarding what energy is used in the field on the edge, suitable models and realistic constraints. Backhaul to send data (photos or otherwise) to the cloud for MD continues to interest me. I have done it over Iridium rudics (slowly) with thumbnails and ZSL are working on Instant Detect 2 to advance this (gateway has the modem, lower power sensors send to the gateway for processing in the field). Currently looking at an EfficentDet model for edge thermal image detection with Leptons for an initial detection to wake something beefier (still edge) only if it's "worth" waking in the camera space to get more confidence and to spend more power

😎 Ed Miller, Sara Beery

Bob Zak (robert.zak@comcast.net)

2020-08-03 13:12:12

A lot to unpack, there. Has anyone looked at the impact of resolution and spectral content on animal target detection? E.g. MD is severely downsampled from the sensor data, but still ~300x300 rgb. At the other extreme, the PIR sensor is a effectively single pixel (maybe two, depending on how you think of it) in a single wavelength.

Bob Zak (robert.zak@comcast.net)

2020-08-03 13:13:25

https://openaccess.thecvf.com/content_cvpr_2017/papers/Huang_SpeedAccuracy_Trade-Offs_for_CVPR_2017_paper.pdf

Looks at (among many other salient options) the impact of image resolution -- testing 300x300 and 600x600. They find:

The effect of adjusting image size. It has been observed by other
authors that input resolution can significantly impact detection
accuracy. From our experiments, we observe that decreasing resolution
by a factor of two in both dimensions consistently lowers accuracy (by
15.88% on average) but also reduces inference time by a relative
factor of 27.4% on average. One reason for this effect is that high
resolution inputs allow for small objects to be resolved. Figure 4(b),
which compares detector performance on large objects against that on
small objects, confirms that high resolution models lead to
significantly better mAP results on small objects (by a factor of 2 in
many cases) and somewhat better mAP results on large objects as
well. We also see that strong performance on small objects implies
strong performance on large objects in our models, (but not vice-versa
as SSD models do well on large objects but not small).

I don't believe this paper explores the effect of narrowing the spectral input (e.g. of processing a black and white image). Is MD more accurate on daytime (color) photos, than on nightime, IR illuminated B&W photos?

Sara Beery (sara.beery@gmail.com)

2020-08-03 13:15:44

I've looked at the effect of performance vs input image resolution, and it doesn't change performance on big animals much but drastically effects performance on small animals or animals that are farther away (totally intuitive, you're shrinking an already-small object so it loses information). Also, MDv4 takes in much bigger than 300x300, it uses a keep-aspect-ratio resizer with a min dimension of 600

Sara Beery (sara.beery@gmail.com)

2020-08-03 13:17:24

performance on day vs night depends on species, there's a bigger effect on performance when you see a species outside of it's normal time (ie most raccoon images are taken at night, so it struggles with raccoons during the day)

Ed Miller (ed@hypraptive.com)

2020-08-03 16:41:58

Anecdotally, I have seen similar performance even with downscaling with large animals (as noted by Sara). I would also point out that if the downscale catches the animal most of the time, then you could use it by default, any only run a larger image occasionally for sanity check (or if you really think something should be there, but the downscaled version didn't find it).

Sara Beery (sara.beery@gmail.com)

2020-08-03 16:46:33

To some extent the real question is what are you trying to do on the edge. If you're trying to detect elephants it's a very different question then trying to detect rodents using an off-the-shelf camera with a PIR sensor optimized for large mammals. I think most small mammals don't trigger the camera at all, and then if they do they're harder to detect using CV because they're small and camouflaged. I think edge systems will need to be open-source and adaptable so that the elephant/bear/big cat/large mammal folks and the rodent/bird/herp folks can both optimize them for their use cases.

👍 Ed Miller, Alasdair Davies

Bob Zak (robert.zak@comcast.net)

2020-08-04 12:02:15

(Back to B&W performance) I guess the real test would be train a single channel version of MD on a dataset of photos converted to B&W. I wouldn't expect it to improve accuracy 🙂, but I bet it would be a net win "accuracy per Joule". Similar for reducing the precision of each channel. Or, more generally -- within a fixed computation budget, what is most efficient allocation of resources towards: model depth; number (and peak special sensitivity) of color channels; and precision/encoding of each channel? (I'll look, but if someone already knows) Are there references that cover this topic?

Alasdair Davies (alasdair@shuttleworthfoundation.org)

2020-08-04 14:24:34

Did someone mention elephants 🐘 @Sara Beery 😄

🐘 Sara Beery, Elizabeth Bondi, Oisin Mac Aodha

❤️ Sara Beery, Bob Zak

😎 Ed Miller

Oisin Mac Aodha (macaodha@caltech.edu)

2020-08-06 18:03:54

Increasing image resolution helps classification performance on the iNaturalist 2017 dataset:
https://arxiv.org/pdf/1806.06193.pdf
(Table 3. this is not a detection task)

👍 Sara Beery

Oisin Mac Aodha (macaodha@caltech.edu)

2020-08-06 18:07:07

@Bob Zak Maybe I'm not fully understanding the B&W model you suggest, but isn't there a danger that any savings from reducing from a 3 channel to 1 channel input will be dwarfed by all the later computation which will not be impacted by changing the input dimensionality.

Bob Zak (robert.zak@comcast.net)

2020-08-06 20:03:00

Yeah -- good point. My ignorance of model structure and resource allocation showing 😞 Certainly, the deeper the model, the less important the size of the input layer. Thanks for pointer to iNat paper!

Ed Miller (ed@hypraptive.com)

2020-08-06 20:34:48

@Oisin Mac Aodha Thanks for the pointer to the FGVC paper. The paper is focused on the classification step, and only considered resolutions from 299x299 (89k pixels) to 560x560 (314k pixels). The 4x image size difference will have some compute/energy impact, but it does also provide a 15% improvement on the Top-1 error (29.93% -> 25.37%). This could be very useful for the BearID Project face classifier, as we currently only use a 150x150 pixel face image.

Some of the discussion earlier on this thread was referring to the image resolution for an Object Detection network. For example the BearID Project's bear face detector. We have tried this on input images from 15M pixels down to 307k pixels. My completely unquantified observation shows a huge reduction in compute with little effect on accuracy (at least for my single class detector). Perhaps we could run some sweeps to quantify this.

@Sara Beery and @Siyu Yang: has anyone run compute vs accuracy comparisons for Mega Detector?

Has anyone run such comparisons for other object detectors with more classes (like those trained on COCO, etc.)?

Oisin Mac Aodha (macaodha@caltech.edu)

2020-08-07 04:36:27

Maybe one pointer would be the EfficientDet paper which explores models with different input resolution (but note that the model size is changing too).
https://arxiv.org/abs/1911.09070

😎 Ed Miller

Bob Zak (robert.zak@comcast.net)

2020-07-01 19:55:44

(yikes -- slack newb -- was looking for something more subtle with these links 😞 )

Sara Beery (sara.beery@gmail.com)

2020-07-01 19:56:22

No worries!

Sara Beery (sara.beery@gmail.com)

2020-07-07 20:22:53

New job posting for a data scientist at a startup working on carbon removal & climate solutions: https://apply.workable.com/carbonplan/j/E6E55611B9/

apply.workable.com

Software engineer / Data scientist - CarbonPlan

CarbonPlan is a non-profit working on the data and science of carbon removal and climate solutions.We aim to ensure the quality, scientific integrity, and transparency of climate solutions, and we work across a range of approaches — forests, soil,...

Original URL: https://apply.workable.com/carbonplan/j/E6E55611B9/

👍 Jonathan Granskog, Suzanne Stathatos, Siyu Yang

Sara Beery (sara.beery@gmail.com)

2020-07-08 15:17:41

WILDLABS is running their 3rd annual Conservation Tech Survey! Let's make sure that CVML researchers have a voice in the survey: https://mailchi.mp/wildlabs/community-survey-2020?e=7b9d646075

mailchi.mp

Have your say in our #Tech4Wildlife Community Survey 2020

In our third annual survey, WILDLABS is evaluating the State of Conservation Technology alongside researchers at Colorado State University. Take the survey to contribute your thoughts on the conservation tech community’s progress, needs, and opportunities for growth, and to help us shape the future of the WILDLABS community of practice.

Original URL: https://mailchi.mp/wildlabs/community-survey-2020?e=7b9d646075

👍 Carly Batist, gvanhorn, Elizabeth Bondi, Talia Speaker

Jędrzej Świeżewski (jedrzej@appsilon.com)

2020-07-10 04:02:42

Hi all 👋,
glad to be joining. At Appsilon we have an AI for Good initiative, which led me into the conservation space. Curious to see what you are up to and will be glad to share our results here.

Appsilon Data Science | End to End Data Science Solutions

Appsilon Data Science | Data Analytics & ML Solutions

Appsilon is a world class data science consultancy. We build advanced Shiny dashboards, predictive ML models, and cutting edge computer vision solutions.

Original URL: https://appsilon.com/

🌍 Sara Beery, Thijs, Björn Lütjens

Carly Batist (cbatist@gradcenter.cuny.edu)

2020-07-10 08:22:51

Welcome! Loved reading the blogs about your work with camera traps in Gabon. Excited to see what comes it. Have you worked at all with passive acoustic devices/data? I am currently trying to figure out the best way to detect lemur calls in my continuous audio recordings from Madagascar.

Jędrzej Świeżewski (jedrzej@appsilon.com)

2020-07-13 03:27:14

Hi Carly,
I don't have much experience with the analysis of acoustic data, but would be interested to read up on it. Can you recommend papers/blogs?

Carly Batist (cbatist@gradcenter.cuny.edu)

2020-07-13 12:23:49

I can definitely go down the rabbit hole with that, but let me pull together some of the reviews and notable case studies, at least for terrestrial use. The marine mammal community uses acoustic monitoring via hydrophones extensively, but I am more familiar with the literature for terrestrial animals like my study species

Carly Batist (cbatist@gradcenter.cuny.edu)

2020-07-13 12:38:27

Marques et al (2013)-Estimating animal population density using passive acoustics.
Heinicke et al (2015)-Assessing the performance of a semi-automated acoustic monitoring system for primates.
Sánchez-Gendriz et al (2017)-A methodology for analyzing biological choruses from long-term passive acoustic monitoring in natural areas.
Gibb et al (2018)-Emerging opportunities and challenges for passive acoustics in ecological assessment and monitoring.
Kershenbaum et al (2019)-Tracking cryptic animals using acoustic multilateration: A system for long-range wolf detection.
Wood et al (2019): Detecting small changes in populations at landscape scales: A bioacoustic site-occupancy framework.
Sethi et al (2020): Characterizing soundscapes across diverse ecosystems using a universal acoustic feature set.
Kvsn et al (2020): Bioacoustics data analysis-A taxonomy, survey and open challenges

👏 Sara Beery, Thijs, Jędrzej Świeżewski

Jędrzej Świeżewski (jedrzej@appsilon.com)

2020-07-14 04:08:44

Thanks @Carly Batist! That is a helpful compilation!

Thijs (thijs@q42.nl)

2020-07-13 03:14:24

Hi all! I work at https://hack-the-planet.io/ where we are working on creating a AI-camera-trap addon, that allows for realtime classification of camera trap images. Results are pushed to a backend via a satellite uplink 📡

This smart-camera-trap project is part of our https://www.hackthepoacher.com/ initiative. We are also building a GSM detection system to track down poachers.

Thijs (thijs@q42.nl)

2020-07-13 03:15:06

I'm working with @Jędrzej Świeżewski on this project en we are planning to rollout prototypes this year.
If anyone has thoughts about this project (especially about inferencing at the edge) please let me know, open for suggestions! 🙌

👍 Ankita Shukla, Siyu Yang, Jonathan Granskog, Jędrzej Świeżewski, gvanhorn, Elizabeth Bondi, Carly Batist, Ed Miller, Lloyd Hughes

Ed Miller (ed@hypraptive.com)

2020-08-01 19:58:53

Have you already selected your hardware platform?

Tee R. (taylornroberts2009@gmail.com)

2020-07-13 21:11:45

Hey all, does anyone have any advice for someone looking to get into the field? I got a bachelor's in history about ten years ago, and have been self learning AI and ML. I would love to be able to use those skills for environmental/natural sciences. Do I need coursework in biology and things like that? Math? (Which I admit is not my strong suit.) Any advice would be much appreciated!.

Sara Beery (sara.beery@gmail.com)

2020-07-14 11:07:41

I'm sure there are many ways to get started, but I might recommend reaching out to a local conservation group and offer your help? Many small conservation orgs don't have the bandwidth or resources to work with cutting edge tech, so it could be a nice way to get your hands dirty on a real problem.

👍 Björn Lütjens

Tee R. (taylornroberts2009@gmail.com)

2020-07-14 13:20:50

okay!

Jonathan Granskog (jonathan.granskog@gmail.com)

2020-07-14 11:47:10

This seems pretty cool: https://www.kickstarter.com/projects/opencv/opencv-ai-kit

Kickstarter

OpenCV AI Kit

A tiny, powerful, open source Spatial AI system

Original URL: https://www.kickstarter.com/projects/opencv/opencv-ai-kit

🙌 Sara Beery

👍 Ștefan Istrate

John Payne (drjohnpayne@gmail.com)

2020-07-14 16:24:01

I’m wondering how others have dealt with the problem of inevitable category creep: you examine 100,000 photos and discover that you have a new category that you hadn’t included when you started training the model, 200 hours ago. It would be nice not to have to retrain the model from start. I can imagine you might a) create some lumped categories (“other”, “other_animal”, whatever) in the beginning to anticipate that, b) or perhaps even add layers to the CNN…but that could get complicated with a complex R-CNN model with ROI heads that also need to be trained…

Sara Beery (sara.beery@gmail.com)

2020-07-14 16:26:16

Definitely tricky. One option (of many) would just be to add the category and fine-tune using your previous model as a starting point? It might converge faster.

John Payne (drjohnpayne@gmail.com)

2020-07-14 16:27:08

Yes…so I guess the trick would be getting the weights from the smaller model to load correctly.

Sara Beery (sara.beery@gmail.com)

2020-07-14 16:27:47

just load everything except the final layer(s) where you map to the categories?

John Payne (drjohnpayne@gmail.com)

2020-07-14 16:30:27

It’s been a while since I looked at the code for loading weights and I can’t remember how hard it would be to just load a subset of the layers, but I think that approach makes sense. Thanks.

Sara Beery (sara.beery@gmail.com)

2020-07-14 16:32:52

Usually you can exclude layers by name

John Payne (drjohnpayne@gmail.com)

2020-07-14 16:36:28

Are you referring to Pytorch, I hope? Also, different topic but since I have your attention for a moment, is the code for your Context R-CNN model online yet? I’m really interested in that — such a smart idea.

Sara Beery (sara.beery@gmail.com)

2020-07-14 16:38:53

I believe tf and pytorch allow for excluding layers while loading. Context R-CNN is online in the tensorflow object detection API (sorry, this one's not pytorch!) https://github.com/tensorflow/models/blob/master/research/object_detection/README.md#context-r-cnn

GitHub

tensorflow/models

Models and examples built with TensorFlow. Contribute to tensorflow/models development by creating an account on GitHub.

Original URL: https://github.com/tensorflow/models/blob/master/research/object_detection/README.md#context-r-cnn

John Payne (drjohnpayne@gmail.com)

2020-07-14 16:39:52

Great, thanks for your help.

👍 Sara Beery

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2020-07-15 02:32:48

I just recently investigated this. There are not many solutions, mainly because training a CNN with only few examples (there are not many once a new class needs to be added) does not really work. Suggestions like few-shot and meta learning don’t work either, as they rely on many new classes/tasks with few examples.

The most recent version of AIDE solves this as follows: every time a new class is discovered, the model adds new output neurons to the classification head (of RetinaNet, ResNet, etc.) and initializes their weights by copying over some other classes’ values. This way, although the initial performance on the new class is not great, it does not alter the accuracy of the other classes at all. Thanks to active learning, the accuracy will improve over time, though.
I have ideas on how to improve this mode in future releases, but for the time being, the software does support adding new classes on-the-fly, at least for the built-in models.

❤️ Sara Beery

John Payne (drjohnpayne@gmail.com)

2020-07-17 00:42:57

Thanks Beni. I’m not sure what you mean by “adding new output neurons,” but I can imagine two ways to add a new class: Method 1) Add a new linear layer; for example if the current output layer dimensions are (bs x c) where bs is batch size and c is number of classes, then you add a new linear layer sized (c x c+1) on top of it, making your final output (bs x c+1). That means your model grows each time you add new classes. Method 2, which I think is what Sara was suggesting, is to replace the final layer (bs x c) with a new layer (bs x c+1) and simply re-learn the weights of that final layer. Out of curiosity, am I correct that you are using Method 1 in AIDE? To me, Method 2 seems a little easier to think about, but I can’t think of any major reason to prefer one over the other.

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2020-07-17 02:25:09

Hi John,
What I meant is the following: the final layer of a CNN maps from D to C, where D is the number of features of the penultimate layer (e.g., 512 for ResNet-18), and C is the number of classes. Effectively, this last layer contains weights of size DxC.
Whenever there is a new class present, the models in AIDE expand this matrix and append a Dx1 vector (for 1 class) to it, which is initialized from a combination of existing weights. Effectively, there is no layer replacement or additional layer to add, but instead the final classification layer gets expanded to Dx(C+1). The advantage is that the weights for the existing classes can be retained, and the new class added without it confusing the model. I hope this helps!

John Payne (drjohnpayne@gmail.com)

2020-07-17 15:53:09

OK thanks Beni; yes that’s what I was referring to as Method 2 except that you are also transferring the old weights. Makes sense.

Sara Beery (sara.beery@gmail.com)

2020-07-21 18:29:09

Hey @Benjamin Kellenberger do you have any direct comparisons of this initialization vs random init when you increase the output layer? It makes total sense, and I'm curious if you get increased accuracy, increased training efficiency, or both? And when you do this do you retrain over your whole training set + the new examples? Do you freeze everything except the last layer or allow all weights to be updated?

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2020-07-22 14:09:09

Hi @Sara Beery
I have only run very quick tests, nothing substantial yet. I just trained a classification model (ResNet-18) on a remote sensing dataset with half of the classes (and images) removed. I then added the output neurons for a variable number of new classes and checked performance. Nothing was frozen; I always trained the full model end-to-end.
With random initialization, the accuracy of the entire model went to zero; it basically put the whole model out of balance. Adding weights by copying other classes’ values resulted in still perfect accuracy for the trained classes, and (obviously) chance guess for the new class. I had to re-train on the new class for about the same number of epochs again as the initial model to get it up to speed.

I originally wanted to investigate this further and try out some more intelligent initialization schemes, but then dropped the idea. I might still do it just for AIDE.

Sara Beery (sara.beery@gmail.com)

2020-07-22 14:12:10

Gotcha. So to get good performance on the new classes you still need to train about the same amount of time over all the data. Thats kinda what I expected, but I agree it would be interesting to test further. Essentially this sounds ideal specifically for an active learning scenario? Where you'd still want the model to work "in the interim" while you're adding new training data or training on new data, before training is complete.

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2020-07-22 14:14:03

Precisely. People have tried to address this with meta learning, attention models, etc., all of which sounds like an overkill or has hidden costs (meta learning requires many tasks with few examples).
It would basically solve the one major problem I have with AIDE in that it allows getting models to provide somewhat decent predictions at the start instead of randomness.

Sara Beery (sara.beery@gmail.com)

2020-07-22 14:21:17

Makes total sense! Probably if you plan to retrain completely before eval though you can just take the simple path of increasing the final layer and randomly initializing all of it? If it gets to the same accuracy in the same amount of time. Just because it's easier to code up 🙂

John Payne (drjohnpayne@gmail.com)

2020-07-22 15:10:23

@Benjamin Kellenberger I would think that you could reduce your training time for new layers by progressively un-freezing the model. As you probably know, the standard approach in transfer learning is to start with all but the new top layer frozen (sometimes BatchNorm layers should also be unfrozen), and gradually unfreeze further and further down as the training progresses. The reason for that is that the lower-level layers in a CNN are doing simple things like identifying edges and curves, …and somewhat higher up, layers are identifying shapes, textures and so on, all of which are likely to still be useful for feeding into the topmost layers that are specific to your particular classification problem. But maybe I misunderstood when you said “Nothing was frozen; I always trained the full model end-to-end.”

Sara Beery (sara.beery@gmail.com)

2020-07-22 15:30:18

I've not really found it to make much difference in practice to unfreeze all right away or unfreeze them progressively.

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2020-07-23 06:45:19

Same here to be honest; I found the learning signal from the few new examples to be so little that there was hardly any change to the early layers.
Batch norm is another story… I had terrible experience with it and since then always replace it with instance norm. Yielded the same result and is much more stable.

John Payne (drjohnpayne@gmail.com)

2020-07-23 16:44:55

Interesting observations, thanks. I guess I had just accepted the orthodoxy without testing it.

John Brandt (John.Brandt@wri.org)

2020-07-24 09:22:07

Hi all -- I'm a data scientist at the World Resources Institute (https://www.wri.org) where I work on a number of deep learning / AI projects. Currently working on remote sensing change detection for tree planting and other restoration projects. Excited to see what everyone else in this channel is working on!

World Resources Institute

World Resources Institute | Making Big Ideas Happen

A global research organization working on six critical goals that the world must achieve this decade in order to secure a sustainable future:climate, energy, food, forests,water, cities & transport.

Original URL: https://www.wri.org

👍 Oisin Mac Aodha, Sara Beery, Tee R., Carly Batist, Björn Lütjens

😎 Jon Van Oast

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-24 09:28:34

Welcome @John Brandt, @Hannah Kerner I think you and John have alot to discuss on agricultural landcover mapping in sub-Saharan africa, also @Tony Chang on multiple temporal fusion of satellite data for tree cover estimation.

🎉 Hannah Kerner, Sara Beery

Hannah Kerner (hkerner@umd.edu)

2020-07-24 09:30:23

Yes! @John Brandt I read your paper recently about crop type classification with attention, would be great to talk more about that

John Brandt (John.Brandt@wri.org)

2020-07-24 10:10:08

Hi @Hannah Kerner - Ben told me great things about your work. Nothing ever came of the attention paper for crop type classification, though I'd love to revisit it. I'm currently working on doing trees outside of forests monitoring with Sentinel. We're working in Ghana, Niger, Cameroon, and Malawi to develop jurisdictional maps on agroforestry and restoration implementation, and (separately) working with local governments across south and Central America to develop monitoring systems for restoration implementation. The preprint of the paper is currently updating to a new version on arxiv and hasn't been announced (since it had R&R), so I've attached it here. Would love to connect about your work and see if there is any overlap (since we're primarily focused on ag landscapes!)

John Brandt (John.Brandt@wri.org)

2020-07-24 10:11:12

@Tony Chang similarly interested in chatting with you! We are doing multi temporal image fusion to help mitigate changes in growing season / leafing seasons so that we can use one global model, would love to chat

Tony Chang (tony@csp-inc.org)

2020-07-31 22:59:23

Definitely John! Welcome to the group! Yeah, the global model challenge is one research that I'm exploring at the moment. Pretty interesting question regarding change detection too. Please feel free to connect with me. My email is tony@csp-inc.org

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-24 16:50:53

I’m wondering if anyone with more formal knowledge on the math of backpropagation/optimization can give me some insight. I am recreating a deep hyperspectral CNN with weak spectral attention. Architecture looks like this. The authors make a point of training the two arms of the network independently. Within each arm, each attention layer gets a softmax layer and then they sum the cross-entropy loss weighted by the depth of the CNN. Shallower layers are down-weighted. Fine, i get that. Then they train both arms together using standard cross-entropy. What is the virtue of this two step strategy? Why not train end-to-end and sum all losses at the same time? I’ve coded it and it provides a modest boost (2% val acc), but it takes a long time to train, and its cumbersome. Repo is here: https://github.com/weecology/DeepTreeAttention , paper is here: https://arxiv.org/abs/2005.11977

John Brandt (John.Brandt@wri.org)

2020-07-27 10:04:55

This is really really interesting. My expectation would be that the virtue is similar to that derived from stochastic weight averaging (SWA) (https://arxiv.org/pdf/1803.05407.pdf) where the weights are frozen after epochs near the end of training, and the prediction is the average of the prediction of these late-stage epochs. The authors in the above article discuss in depth how averaging these good predictions results in an even better prediction, based on theoretical analyses of what the cross-entropy loss surface should look like. Based on this, starting with a summation of the two arms being already trained is somewhat similar to SWA where combining the two trained weights identifies a lower optima. The benefit of the paper you linked is that this lower optima is then the start of a new training cycle, where there are new patterns to learn between the overall network, that can then even find lower optima. I would expect that if you train the overall network end-to-end, you don't get the benefit of finding two optima, combining them to find a new even lower optima, and then having new spaces to explore

👍 Ben Weinstein

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-27 10:55:32

By that logic I should (strongly) decrease the learning rate at the 2nd stage of learning to not hop out of any minimum I just found in the first stage of learning. I’m still playing with the right parameters for learning rate decay.

John Brandt (John.Brandt@wri.org)

2020-07-27 13:02:17

ya! do you use adabound?

John Brandt (John.Brandt@wri.org)

2020-07-27 13:02:31

if you use adabound you could probably just keep the learning rate as it would have converged to SGD by then

John Brandt (John.Brandt@wri.org)

2020-07-27 13:02:43

https://arxiv.org/pdf/1902.09843.pdf

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-27 13:03:43

no, i was using ADAM. looking now.

John Brandt (John.Brandt@wri.org)

2020-07-27 13:04:08

basically just smoothly switches Adam to become SGD since SGD is known to generalize better than Adam but Adam is more stable early on

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-31 17:26:26

Hi everyone, long set of posts. I’m giving a talk at the Florida Museum of Natural History on an overview for a Computer Vision for Ecology. September 18th, i’ll post zoom link. What are the largest obstacles to applying AI to your projects? How would they be solved? What are our grand challenges? Let’s start a discussion here. Here is my outline so far, all are welcome to comment there too. https://docs.google.com/document/d/1j79muPrdUxo6-GyYR4ML39apwfUqK5qRtxg6gg7sRkw/edit?usp=sharing

❤️ Sara Beery, Tee R., Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

👍 Talia Speaker, Ed Miller

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-31 17:26:55

Why do we need a computer vision for ecology?

Fieldwork is expensive and laborious

Human presence reduces detection probabilities

Ecological models are data hungry (e.g N-mixture, Markov, bayesian latent states)

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-31 17:29:15

What are the major obstacles?

Not enough data

Large barrier to entry

Cross disciplinary collaboration is difficult

Sara Beery (sara.beery@gmail.com)

2020-07-31 17:36:53

Bias in the data we do have, natural world is long-tailed... both probably fall under "not enough data" one way or another

👍 David Healey

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-31 17:37:53

added. “Ecology is often focused on rare classes with large geographic variation.”

👍 Sara Beery, Tee R.

Sara Beery (sara.beery@gmail.com)

2020-07-31 17:49:28

In you data sharing section, it would be nice to talk about the opportunity to leverage bycatch when data is open

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-31 17:51:18

‘A role for museums as curators of digital information: many studies are focused on a single set of species, but record many incidental sightings that can be used as training data.’

Sara Beery (sara.beery@gmail.com)

2020-07-31 17:51:56

training data or just useful observations that have previously been left to rot on hard drives

👍 Ben Weinstein, Tee R.

Carly Batist (cbatist@gradcenter.cuny.edu)

2020-07-31 19:52:30

I would add something to acknowledge the diff toolkits ecologists and computer scientists are trained to use (and therefore stick with). e.g., every intro biostats class I've heard of uses R, as does most of the ecology world, whereas computer scientists breathe Python. Not saying one is better than the other, but it's hard to get either side to invest in learning a new programming language. If you wanted to really get at the core of the problem, you'd likely have to get undergrad programs to change the number or structure of required courses.

👍 Ben Weinstein, Sara Beery, Tee R., Björn Lütjens

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-31 19:54:14

with a nod to RopenSci, reticulate and https://keras.rstudio.com/

👍 Carly Batist

Carly Batist (cbatist@gradcenter.cuny.edu)

2020-07-31 19:58:48

And for data/model sharing, http://lila.science/ is a good example! Also Wildlife Insights, Xeno-Canto (which the warbleR package can pull directly from), MobySound/OrcaSound, the Macaulay library.

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-31 19:59:41

shows how much outreach is needed, I’m a contributor to several of those things, but never heard of the warbleR package.

Carly Batist (cbatist@gradcenter.cuny.edu)

2020-07-31 20:06:25

Bioacousticians love that package. Also gibbonR, monitoR (all 3 build upon seewave and tuneR). And for web apps, ARBIMON/Sieve Analytics is good, but has a limit to storage/computing before you have to pay (they've also just merged with Rainforest Connection).

arbimon.sieve-analytics.com

Arbimon II - Home

Arbimon II - Bio-acoustics analyzer

Original URL: https://arbimon.sieve-analytics.com/home

Ben Koger (benkoger@gmail.com)

2020-08-01 14:47:49

From a behavioral ecology point of view, people were traditionally forced to choose between working in artificial lab environments to get good high resolution behavioral data, or working in natural environments and making do with much coarser behavioral information. Computer vision finally lets us get high quality behavioral datasets in truly natural environments.

Ben Weinstein (benweinstein2010@gmail.com)

2020-08-01 18:05:10

What is the next step for behavioral ecology in natural settings using these tools. What is the grand goal/obstacle?

Amrita Gupta (agupta375@gatech.edu)

2020-08-04 12:45:19

RE: cross-disciplinary collaboration difficulties. As a CS PhD student explicitly working in the "AI for sustainability" space for the past several years, I think it is still challenging to initiate collaborations between CS researchers and ecologists. Attending events like NACCB or US-IALE to learn about domain challenges can help, but is outside of typical venues for which AI researchers can get support to attend (in line with what Sara was saying about what is considered "CS research"). Even when we do get to participate in these events, I believe whether or not a collaboration comes out of it still comes down to knowing an "advocate" who will connect people on both sides.

❤️ Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-31 17:29:51

Better representation, how do we make AI more inclusive?

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-31 17:47:31

There is a nice debate here that parallels the discussion of double blind paper review versus active support for underrepresented groups that I think is really relevant to our community. https://www.nytimes.com/2020/07/16/arts/music/blind-auditions-orchestras-race.html?searchResultPosition=1

The New York Times

By Anthony Tommasini

To Make Orchestras More Diverse, End Blind Auditions

If ensembles are to reflect the communities they serve, the audition process should take into account race, gender and other factors.

Original URL: https://www.nytimes.com/2020/07/16/arts/music/blind-auditions-orchestras-race.html?searchResultPosition=1

❤️ Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Carly Batist (cbatist@gradcenter.cuny.edu)

2020-07-31 19:40:45

Just a small note-I would use the term 'global south' rather than 'developing countries,' just to move away from the colonial terminology.

👍 Ben Weinstein, Sara Beery, Tee R.

Carly Batist (cbatist@gradcenter.cuny.edu)

2020-07-31 19:43:55

And I'd add something about making actually equitable collaborations between researchers from the global south and north. Westerners going down and telling them what to do rather than listening and working with them also reinforces this neocolonial thinking. We need to be building in-situ capacity and long-term infrastructure. As it is now, someone gets a grant and goes down for 6 months to collect data and then leaves, which is 1) exploitative (often field assistants and local collaborators/org's are not included in pubs or even told the results) and 2) not sustainable.

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-31 19:47:32

What are the measures that would help actualize this? I gave a talk virtually in Ecuador last week, and a lot of the discussion was focused on hardware resources, that just having access to the correct kinds of computational resources would be a really large step forward for a lot of field teams that I’ve been working with in Ecuador. I think there’s a trade off in terms of how much development should be focused on the cloud, which helps reduce the barrier to entry, since less coding is needed, versus getting data into the cloud, which can be really hard for my collaborators in Latin America.

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-31 19:52:34

@Lily Xu - OLD ACCOUNT, DO NOT MESSAGE any thoughts here from some of your recent organizing.

Carly Batist (cbatist@gradcenter.cuny.edu)

2020-07-31 20:13:39

Yes the hardware is definitely big, particularly because it facilitates in-country analyses rather than just data collection. But obsolescence is also an issue, especially with how fast tech is moving now. I work in Madagascar and the computers the research station got are really old/slow/broken/etc now and shipping is basically non-existent so it's a matter of someone flying new stuff in. I'd also say that whichever side of the trade-off you want, equitable partnerships are still integral and unfortunately in many places, not equitable at all.

👍 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Carly Batist (cbatist@gradcenter.cuny.edu)

2020-07-31 20:14:29

I by no means am trying to suggest I have answers to any of this! Just giving my experience being on the ecology side of things in the name of cross disciplinary collaboration! 🙂

👍 Ben Weinstein, Sara Beery

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2020-08-03 14:05:15

Hi Ben, just seeing this discussion now! I'm excited for your talk in September.

That's a great point connecting with the orchestras and faults of blind auditions/reviewing — many arguments for affirmative action apply here, too. I've done some reviewing for workshops recently (AI for Good and Mechanism Design for Social Good) which aren't blind, and during reviewer discussions I see people being a lot more mindful to authors and institutions from historically underrepresented areas, including Latin America.

In some ways, I think virtual conferences/events definitely help serve as an equalizer: conferences that formerly cost $2000+ to attend now cost $5–10. (Ariel Procaccia wrote an op-ed about the success of virtual EC)

I also think we need to, as Sara and others have mentioned here, reconsider what are "valuable" contributions to AI. For example, producing NLP datasets that are in languages other than English and other projects that help combat inequality throughout the AI pipeline, from data gathering to labeling to designing algorithms.

Bloomberg.com

Conferences Work Better When Everyone Stays Home

I expected the coronavirus to ruin a scientific gathering earlier this month. Then I virtually showed up.

Original URL: https://www.bloomberg.com/opinion/articles/2020-07-21/coronavirus-scientific-conferences-work-better-online

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-31 17:32:20

i’ll start calling on people in the next few days with specific questions.

Tee R. (taylornroberts2009@gmail.com)

2020-07-31 22:09:35

I think the large barrier to entry is an interesting point, I am not sure if this is quite what you're looking for but knowledge, either in AI or the topic of the studies can be a bit of a barrier. It can be overwhelming for someone approaching from the outside, either as a newbie or from a more technical background. I think looking at that might also encourage more grass roots help in some of the data tasks that consume so much time and effort.

Tee R. (taylornroberts2009@gmail.com)

2020-07-31 22:10:44

This may be somewhat more centered around the US experience, though I think I have seen some more grass roots connections in other places regarding ecology and natural studies...I don't know.

👍 Ben Weinstein, Sara Beery

Bistra Dilkina (dilkina@usc.edu)

2020-07-31 22:18:13

@Ben Weinstein this is a great topic. Re: inclusion/diversity in AI, I would say that interdisciplinary research projects such as AI+Ecology can play a key role in improving diversity in AI, and also in installing more inclusivity, awareness and open-mindedness in AI researchers (like similar combinations of AI with social and life sciences) . BUT it is important to discuss the barriers to entry for AI/CS people to this subfield, as well as the barriers for ecologists to it. I often get approached by CS students with passion for the environment, and environment eng/ecology students with interest in computational methods, but there are no clear resources on how to successfully build a research or professional path

👆 Tee R., Sara Beery, Carly Batist, Elizabeth Bondi, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Björn Lütjens

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-31 22:35:34

Thanks @Bistra Dilkina for these thoughts, we often see this just from the ecology side. Also, i’m hoping to point to some of the park ranger optimization work under the heading of “Integrating measures of uncertainty in machine learning and ecology”.

Bistra Dilkina (dilkina@usc.edu)

2020-07-31 22:38:00

@Ben Weinstein are you going to also talk about remote sensing AI for ecology?

Ben Weinstein (benweinstein2010@gmail.com)

2020-07-31 22:38:52

thats where most of my research is, so just as an example of how we can use weak supervision on unlabeled data to tackle the ‘not enough data’ obstacle.

Sara Beery (sara.beery@gmail.com)

2020-08-01 01:03:16

As someone who is fighting to figure out how to keep doing this as a career, I can really second the lack of clear resources. Another issue, from the academic CS side, is the balancing act between impactful for conservation and publishable in CS that any CS grad student in this space has to consider when deciding what projects to focus on.

👍 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Siyu Yang, Ankita Shukla, Björn Lütjens

Tee R. (taylornroberts2009@gmail.com)

2020-07-31 22:22:32

I 100% agree with this, this is where I'm at. Learning the tech side but wanting to help out on the eco side too...

❤️ Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Tee R. (taylornroberts2009@gmail.com)

2020-08-02 18:35:45

You guys, I made a random viz for anyone who wants to see: https://public.tableau.com/profile/tee8225#!/vizhome/TexasBirdsProject/TexasBirdBiodiversity?publish=yes I was using the eBird data to do a few visualization projects. One fun thing I did learn is that blue herons have a far wider range than I thought. I just moved to Texas so may use this data to see if I can catch sight of a few of them, haha! 🙂

public.tableau.com

T. Roberts - Profile | Tableau Public

Currently exploring data science and machine learning, I'm a creative with a passion for striking visuals and making an impact.

Original URL: https://public.tableau.com/profile/tee8225#!/vizhome/TexasBirdsProject/TexasBirdBiodiversity?publish=yes

🐦 Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2020-08-03 12:12:43

very cool. Did you download the csv or is live updating through the app?

👍 Tee R.

Tee R. (taylornroberts2009@gmail.com)

2020-08-03 20:22:50

I downloaded the file

karen bakker (karen.bakker@ubc.ca)

2020-08-07 12:08:57

Hi everyone - just wanted to introduce myself as a new member. I'm a Professor at the University of British Columbia, and I've been working on digital tech and environmental governance innovation for the past several years (my original training was in physics and environmental sciences). I'm currently working on a book on this topic to be published with Princeton University Press next year, and I also maintain a database of innovative digital conservation technologies (3,000+ and growing fast). Some of you might be interested in this meta-review of the academic literature on digital tech/environmental governance: Bakker, Karen, and Max Ritts. "Smart Earth: A meta-review and implications for environmental governance." Global environmental change 52 (2018): 201-211. https://www.sciencedirect.com/science/article/pii/S0959378017313730

sciencedirect.com

Smart Earth: A meta-review and implications for environmental governance

Environmental governance has the potential to be significantly transformed by Smart Earth technologies, which deploy enhanced environmental monitoring…

Original URL: https://www.sciencedirect.com/science/article/pii/S0959378017313730

❤️ Sara Beery, Carly Batist, Tee R., Siyu Yang, Riccardo de Lutio, Elizabeth Bondi, David

👍 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Megan Cromp

Carly Batist (cbatist@gradcenter.cuny.edu)

2020-08-07 13:58:15

Awesome compilation!! Can't believe I didn't catch this paper when it first came out, thanks for sharing here!

🙂 karen bakker

karen bakker (karen.bakker@ubc.ca)

2020-08-07 12:14:15

I also collaborate with the United Nations Environment Program on their global initiatives on digital tech and environmental data. Here's a recent article I co-authored with David Jensen (UNEP) that outlines some of the ideas that are being debated by international organizations: https://medium.com/@davidedjensen_99356/digital-planet-20-priorities-3778bf1dbc27

Medium

Are these the 20 top priorities in 2020 for a digital ecosystem for Earth?

What are concrete targets in 2020 for realizing a robust and equitable digital ecosystem for the planet?

Reading time

29 min read

Original URL: https://medium.com/@davidedjensen_99356/digital-planet-20-priorities-3778bf1dbc27

🙌 Sara Beery, Siyu Yang, Elizabeth Bondi

👍 Megan Cromp, Bistra Dilkina

Chris Yeh (chrisyeh96@gmail.com)

2020-08-10 14:37:56

Passing along a conference opportunity from a different Slack group:

We're happy to announce that CompSust DC 2020 has just opened up submissions! This year's DC will be held October 17-18, 2020 virtually. More info and application in the post below, and online: http://www.compsust.net/compsust-2020/

👏 Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Oisin Mac Aodha, Björn Lütjens

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2020-08-10 14:44:56

woohoo! 🙂

😍 Chris Yeh

Chris Yeh (chrisyeh96@gmail.com)

2020-08-10 14:38:58

[ MESSAGE TEXT EMPTY ]

Tony Chang (tony@csp-inc.org)

2020-08-11 15:15:49

Dear AI for Conservation Group!

For the past eight years, Conservation Science Partners has been a leader in the fields of conservation biology and landscape ecology through our use of advanced geospatial, remote sensing, and ecological modeling approaches. In recent years, the intersection of environmental sciences and computer sciences has become more apparent, and CSP is now seeking to integrate talented and diverse individuals with skills in data sciences to solve some of the grand challenges in conservation. As we launch a new data science and analytics initiative, we are excited to announce two new positions: Front End Developer and Data Engineer. Please see the attached hiring announcements and circulate among your network. Also, if there are any questions regarding the position, please feel free to contact me at tony@csp-inc.org.

Sincerely,
Tony Chang

👍 gvanhorn, Sara Beery, Carly Batist, Talia Speaker, Björn Lütjens, Amrita Gupta

Carly Batist (cbatist@gradcenter.cuny.edu)

2020-08-11 15:20:09

You should also post to Wildlabs if you haven't already!

Wildlabs.net

Great Zebra and Giraffe Count and ID - LILA BC

Original URL: https://www.wildlabs.net/

👍 Talia Speaker, Björn Lütjens

Bistra Dilkina (dilkina@usc.edu)

2020-08-11 15:59:39

@Amrita Gupta

Tony Chang (tony@csp-inc.org)

2020-08-11 16:04:24

Thanks @Carly Batist I'll check it out!

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2020-08-13 18:01:20

The MD4SG working group on environment explores environmental challenges through the lens of computation and economics, with a focus on issues that disproportionally impact already-vulnerable populations. Our biweekly group this fall will focus on land use, climate, and energy. 🌲🌍☀️

Fill out this short survey by August 21 6pm ET if you're interested in joining our group.

Wanyi Li and I began this group in the spring and are excited for the upcoming semester!

😍 Sara Beery, Elizabeth Bondi, Siyu Yang

Björn Lütjens (bjoern.luetjens@gmail.com)

2020-08-19 18:19:00

The final list of all NeurIPS 2020 workshops is now available at: https://nips.cc/Conferences/2020/Schedule?type=Workshop Some relevant ones for this group might be: AI for Earth Sciences and Tackling Climate Change with ML.

nips.cc

NeurIPS 2020 Schedule

NeurIPS Website

Original URL: https://nips.cc/Conferences/2020/Schedule?type=Workshop

👏 Sara Beery, David

David Rolnick (dsrolnick@gmail.com)

2020-08-20 07:20:09

Weighing in from the climate change workshop, we'd love to have submissions from this group! If you do submit, remember to make explicit how your submission is relevant to climate change. Not every conservation-related problem is a climate change problem, but there can be overlap - for example if monitoring populations that are specifically under threat from climate change.

More info about the workshop here:
https://www.climatechange.ai/events/neurips2020

🌍 David, Björn Lütjens

Mikey Tabak (tabakma@gmail.com)

2020-08-20 06:40:46

How do you computer vision folks keep up with the latest and most effective methods? I'm finding that there's so much new stuff published in so many places that it's hard to know which approach to use. (My background is in ecology and the field doesn't move quite as quickly, and everything is published in the peer-reviewed literature.) I'm trying to do object segmentation on really small objects (animal carcasses) from really high up (175 m drone). I've tried a few methods, but I keep finding newer stuff. I also keep maxing out my insufficient server, so I don't have unlimited computing power to try new things currently.

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2020-08-20 09:39:44

There’s two different kinds of recommendations I can give here.
The problem of keeping up with the latest publications can be solved easily in computer vision by checking the following resources:

arXiv: https://arxiv.org/list/cs.CV/recent

CVF (all CV conferences): http://openaccess.thecvf.com/

As for picking the right pipeline: in all honesty, I deliberately don’t use the latest method. Oftentimes, they are not widely spread, very exhaustive to implement, and just an incremental improvement over existing methods. Instead, I tend to stick to the models that have been shown to be good. They may not yield the very best performance, but they are solid and can be adapted to one’s needs. For me, these include:

Image classification: I always use ResNet, even though there’s plenty of newer, more powerful methods.

Segmentation: I am a big fan of U-Net; it is easy to understand and performs like a champ. DeepLab is also great, though.

Object detection: I have been using RetinaNet and YOLO, if needed. Most of the time I could do away with bounding boxes and just predict points, in which case I use a ResNet without the average pooling layer. Honestly, this is what I would recommend for you. I may be biased, but I have been working with tiny objects from drones, and this outperforms other models in speed, accuracy and simplicity.

👍 Mikey Tabak, Björn Lütjens, Sara Beery, Siyu Yang

Ben Weinstein (benweinstein2010@gmail.com)

2020-08-20 10:06:20

I agree with this, I also think that its worth directing your focus away from architectures and more towards the entire pipeline. In ecology we almost always have limited data, getting familiar with fine-tuning strategies, transfer learning strategies generating synthetic data and understanding model parameters such as loss weights and learning rates often makes a HUGE difference as compared to the actual architecture.

👍 Mikey Tabak, Björn Lütjens, Sara Beery, Siyu Yang

Mikey Tabak (tabakma@gmail.com)

2020-08-20 10:06:44

@Benjamin Kellenberger Thank you so much for your guidance! This is really helpful. For this project I have been using mostly DeepLabV3 and I was next going to try U-Net. I had not thought about using just points for these little carcasses instead of segmentation, but I'll give this a shot. Also thanks for providing these links! I'll keep an eye on them to stay current on the latest methods. I really appreciate you sharing your wisdom here.

👍 Benjamin Kellenberger

Mikey Tabak (tabakma@gmail.com)

2020-08-20 10:14:24

Thank you @Ben Weinstein! (I was actually just reading your recent paper in Ecological Informatics this morning.) I have been doing a lot of fine tuning with adjusting hyperparameters and transfer learning. I currently cannot use an HPC, so adjusting parameters and then re-training a model takes a really long time on the small server we have. Do you have readings you recommend for strategies to adjust the pipeline? I've worked through some examples in the past, but I'm certainly not an expert and a lot of my adjustments are arbitrary decisions that I make in the process.

Ben Weinstein (benweinstein2010@gmail.com)

2020-08-20 11:55:28

usually testing a small portion of your model is good practice, it has the added benefit of revealing any overfitting when you go from your small dataset to the full dataset.

👍 Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2020-08-20 11:55:44

Keras has a few optomizers, but I also use the ones from comet.ml

Ben Weinstein (benweinstein2010@gmail.com)

2020-08-20 11:56:02

that helps with parameter searches.

Ben Weinstein (benweinstein2010@gmail.com)

2020-08-20 11:56:56

more broadly the focus on data generation is so important. I’ve seen transfer learning among very disparate classes work well, we used that tree model to start and trained for drone work on birds and it was hugely beneficial.

👍 Mikey Tabak

Ben Weinstein (benweinstein2010@gmail.com)

2020-08-20 11:57:26

there is only so much information you can squeeze out of small ecological datasets.

Ben Weinstein (benweinstein2010@gmail.com)

2020-08-20 11:58:29

I think the camera trap competitions that @Sara Beery ran also showed that grabbing imagery from other locations can be useful, even in the extreme situation when you don’t have real information for a target class (see their kaggle competition going from CA to Idaho)

👍 Sara Beery, Mikey Tabak

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2020-08-20 12:01:56

For parameter optimisation, I second @Ben Weinstein — trying on a decently sized subset usually reveals enough as to what settings one needs.

👍 Mikey Tabak

Sara Beery (sara.beery@gmail.com)

2020-08-20 12:04:33

Agreed with all of the above. I think it's good to start with something robust that has been proven out over time, and make sure you've really maxed that out by exploring data augmentation, hyperparameters, and cotraining or pretraining from other relevant public datasets. Once you have an idea of what you can do with a well-proven model it makes it easier to determine whether a newfangled method is worth trying to incorporate. I completely agree with @Benjamin Kellenberger that frequently the newest methods can be unreasonably complicated to implement and finicky to train for very small gain.

👍 Mikey Tabak

Ben Weinstein (benweinstein2010@gmail.com)

2020-08-20 12:08:36

besides our methods, right @Sara Beery

😅 Sara Beery, Mikey Tabak

Ben Weinstein (benweinstein2010@gmail.com)

2020-08-20 12:08:56

those are always worth adopting.

Ben Weinstein (benweinstein2010@gmail.com)

2020-08-20 18:22:27

a follow up here from today. Forgot to apply per image normalization and centering. Added per image and performance increased 20%. Alot more important than any architecture difference.

👍 Sara Beery, Benjamin Kellenberger, Mikey Tabak

Mikey Tabak (tabakma@gmail.com)

2020-09-03 13:32:25

@Benjamin Kellenberger @Ben Weinstein @Sara Beery Thank you all so much for your guidance! I fiddled with hyperparameters, augmentation, optimizers, and loss functions and now I have good model (it's actually better than our technicians at finding objects. Now I'm getting started on another project and have another question, so I'll start a new thread.

🙌 Sara Beery

👍 Benjamin Kellenberger

Dan Morris (agentmorris@gmail.com)

2020-08-27 14:54:25

Two new individual ID datasets available on lila.science thanks to the good folks at Wild Me:

http://lila.science/datasets/great-zebra-giraffe-id
http://lila.science/datasets/whale-shark-id

Everyone go identify some animals with AI! (Bonus: there are lots of fantastic baby zebra pictures.)

LILA BC

Overview This dataset contains images of plains zebra (Equus quagga) and Masai giraffe (Giraffa tippelskirchi) with bounding boxes and individual animal identifications. Images are taken from a two-day census of Nairobi National Park, located just south of the airport in Nairobi, Kenya. The “Great Zebra and Giraffe Count” (GZGC) photographic census was organized on February...

Original URL: http://lila.science/datasets/great-zebra-giraffe-id

LILA BC

Whale Shark ID - LILA BC

Overview This data set contains images of whale sharks (Rhincodon typus) with bounding boxes and individual animal identifications. This dataset represents a collaborative effort based on the data collection and population modeling efforts conducted at Ningaloo Marine Park in Western Australia from 1995-2008. Photos (7888) and metadata from 2441 whale shark encounters were collected from...

Original URL: http://lila.science/datasets/whale-shark-id

🎉 Sara Beery, Elijah Cole (Deactivated), Ben Weinstein, Oisin Mac Aodha, Tee R., Jonathan Granskog, Zac Winzurk, David Healey, gvanhorn, Björn Lütjens, Srishti, Omiros Pantazis

😍 Jon Van Oast, Chris Yeh, Riccardo Pressiani, Ankita Shukla

👍 Mikey Tabak, Siyu Yang

Oisin Mac Aodha (macaodha@caltech.edu)

2020-08-27 15:08:09

Looks very cool! Would it be possible to have some example images on the landing page for each? Not necessarily hard examples, but just to see the variety. I like how @Sara Beery did it for her Caltech Camera Trap dataset:
https://beerys.github.io/CaltechCameraTraps/

beerys.github.io

CaltechCameraTraps

Caltech Camera Trap dataset page

Original URL: https://beerys.github.io/CaltechCameraTraps/

👍 Jon Van Oast

Dan Morris (agentmorris@gmail.com)

2020-08-27 15:40:04

I do like that idea, in fact I like the idea of doing that across all LILA datasets, but then again I'm also lazy, so let's call it a crowdsourcing effort. 🙂 If anyone sends me a bunch of images from any dataset on LILA that demonstrate the breadth of the data in an interesting way, I'll find a way to either put them right on the main page for that dataset, or - more likely - host a separate HTML file somewhere that shows those images, and link to that from the main page.

🎉 Jon Van Oast, Oisin Mac Aodha, Sara Beery, Carly Batist

Matt Ziegler (mattzig@cs.washington.edu)

2020-08-31 03:28:39

Hi there! I've been lurking for a while but haven't introduced myself yet. I'm a PhD student in U Washington's ICTD Lab, mostly focused on technologies for the community/social/political aspects of conservation.

It's not really AI, but @Kennedy Murrithi, @William Njoroge and I have a new paper about basic mobile phone services for engaging with Ol Pejeta's surrounding villages: opening communication channels to improve service delivery and more fairly share the costs/benefits of conservation, and building social capacity to work together better when issues arise. (To toot my own horn, we won best paper at ACM Computing and Sustainable Societies this year!) Check it out:

The paper: https://mattziegler.net/papers/Ol-Pejeta-Phones-COMPASS2020.pdf
15-minute presentation: https://www.youtube.com/watch?v=hEUJq97n1-E
3-minute version: https://www.youtube.com/watch?v=gRryk_xUMkk

YouTube

Matt Ziegler (https://www.youtube.com/channel/UCiiiyXNSDdl3-bNw7gT-QPQ)

Mobile Phones and Conservancy Communities

Original URL: https://www.youtube.com/watch?v=gRryk_xUMkk

👍 Oisin Mac Aodha, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Carly Batist, Sara Beery, David Healey, Jon Van Oast, Srishti, Mikey Tabak

Jon Van Oast (jon@wildme.org)

2020-09-01 12:18:16

looking forward to reading this paper and watching these videos. i was lucky enough to stay at ol pejeta for a bit, working with zebra and giraffe conservation (photo id software project we work on).

👍 Sara Beery, Matt Ziegler

Matt Ziegler (mattzig@cs.washington.edu)

2020-09-01 14:57:12

Nice! I've definitely heard of you. Maybe catch you there again sometime!

🎉 Jon Van Oast

Mikey Tabak (tabakma@gmail.com)

2020-09-03 13:41:38

What software are folks using to annotate videos for training computer vision models? It looks like there are some options out there, but it's hard to tell what will work well for this application. I'm planning to do segmentation (or bounding box) on the videos and for one project I'll be trying to find volant animals at 175 meters above the camera, so the technicians will need to be able to zoom in. @Benjamin Kellenberger developed the great AIDE software for still images, and I'm wondering if there is anything of similar quality for videos.

Sara Beery (sara.beery@gmail.com)

2020-09-03 13:42:51

I recently annotated bounding boxes in video using vatic, which is open source, but I'm not sure if it supports zooming. http://www.cs.columbia.edu/~vondrick/vatic/

👍 Mikey Tabak, Jonathan Granskog

➕ Srishti

Sara Beery (sara.beery@gmail.com)

2020-09-03 13:46:02

I just hosted it locally, and didn't do a large amount of data.

Ben Weinstein (benweinstein2010@gmail.com)

2020-09-03 13:46:10

i heavily use rectlabel for Mac

👍 Mikey Tabak, Zac Winzurk, Srishti, Jonathan Granskog, Sara Beery

Ed Miller (ed@hypraptive.com)

2020-09-03 21:55:45

Anyone have experience with VoTT? Is this what is used for the Azure labeling tool?

GitHub

microsoft/VoTT

Visual Object Tagging Tool: An electron app for building end to end Object Detection Models from Images and Videos. - microsoft/VoTT

Original URL: https://github.com/microsoft/VoTT

👍 Mikey Tabak, Mari Reeves

Utkarsh Goel (ugoel@connect.hku.hk)

2020-09-04 07:15:15

There’s an open source project called labelme, and I’ve used it for a project earlier. Does video annotation well, but I will have to check if it has a feature for zooming in.

https://github.com/wkentaro/labelme

GitHub

wkentaro/labelme

Image Polygonal Annotation with Python (polygon, rectangle, circle, line, point and image-level flag annotation). - wkentaro/labelme

Original URL: https://github.com/wkentaro/labelme

👍 Mikey Tabak

Ben Koger (benkoger@gmail.com)

2020-09-14 05:13:56

I've started using CVAT in the past few months and have really liked it. Developed by intel but open source has nice interface and supports most standard types of annotation on images and videos. Used to be a little unstable but much better now. Also easy to run on a server so many people can annotate same project. Allows zooming. https://github.com/openvinotoolkit/cvat

GitHub

openvinotoolkit/cvat

Powerful and efficient Computer Vision Annotation Tool (CVAT) - openvinotoolkit/cvat

Original URL: https://github.com/openvinotoolkit/cvat

Mari Reeves (mari_reeves@fws.gov)

2020-12-11 18:13:57

After looking into this question, I ended up using VoTT for ease of use and the ability to export records to different formats.

👍 Sara Beery, Mikey Tabak

Ben Weinstein (benweinstein2010@gmail.com)

2020-09-16 11:18:31

I’m seeing alot more people here in general than in the jobs channel. See that channel for a new position in Switzerland that my collaborators are hiring for deep learning for ecological object detection.

👍 Sara Beery, gvanhorn, Mikey Tabak, Mike C

Ben Weinstein (benweinstein2010@gmail.com)

2020-09-18 13:41:58

I’m giving a talk on “A Computer Vision for Ecology” at Florida Museum National History (+ iDigBio) at 3pm ET today. Zoom webinar: https://ufl.zoom.us/j/98489487554?pwd=dW1ULzZyTnFoQVYrczJja1JKMDV3dz09
Meeting ID: 984 8948 7554 Password: 671049. I was asked to do a wide ranging talk, so @Siyu Yang @Sara Beery @gvanhorn, @Holger Klinck and others get a shoutout for recent work. Thanks everyone who helped with the slides and address those big questions. Slides are here if interested. https://www.dropbox.com/t/GqyTNrallTfy7G9v

Dropbox

Ben Weinstein sent you 1 item

ComputerVisionforEcology.pptx

Original URL: https://www.dropbox.com/t/GqyTNrallTfy7G9v

👍 Oisin Mac Aodha, Sara Beery, Elijah Cole (Deactivated), Dan Morris

Ben Weinstein (benweinstein2010@gmail.com)

2020-09-25 16:41:17

If anyone actually wanted to see my talk on computer vision for ecology. Here is the recording https://ufl.zoom.us/rec/share/OCn_21P2mqZa2jAqy3xNJ8xXRbqgXuArqekIssggyTsn2Zvn5ccv9z1QrifnamkM.eqhoPnR_XS57QOMJ

🌳 Sara Beery

Holger Klinck (hk829@cornell.edu)

2020-09-18 13:42:31

Cool. Thanks Ben!

Ben Weinstein (benweinstein2010@gmail.com)

2020-09-18 13:44:41

Definitely. Do you happen have a slide that shows/plays any audio and birdnet predicts it? I just show a screenshot and describe it.

Holger Klinck (hk829@cornell.edu)

2020-09-18 13:49:14

Thread Reply:

Holger Klinck (hk829@cornell.edu)

2020-09-18 13:49:30

Feel free to modify as you see fit!

👍 Ben Weinstein

Sara Beery (sara.beery@gmail.com)

2020-09-24 16:52:38

This is cool, lots of tropical forest data about to become public: https://www.planet.com/pulse/planet-ksat-and-airbus-awarded-first-ever-global-contract-to-combat-deforestation/?fbclid=IwAR3BwzSlsuKrFZbdDwJ3t68nRxX6fibSjFVaW7QWUsc39wNGx9_kLwVwJDc

planet.com

Planet, KSAT and Airbus Awarded First-Ever Global Contract t...

Planet | Planet, KSAT and Airbus Awarded First-Ever Global Contract to Combat Deforestation

Original URL: https://www.planet.com/pulse/planet-ksat-and-airbus-awarded-first-ever-global-contract-to-combat-deforestation/?fbclid=IwAR3BwzSlsuKrFZbdDwJ3t68nRxX6fibSjFVaW7QWUsc39wNGx9_kLwVwJDc

🎉 Jon Van Oast, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, David, Mike C

👍 Srishti, Ankita Shukla, David, Siyu Yang

Ben Weinstein (benweinstein2010@gmail.com)

2020-09-24 16:56:01

I had a meeting with Global Forest Watch not so long ago and didn’t hear about this @John Brandt will you at WRI have access to these data? Global Forest Watch is explicitely mentioned in the announcement. Will there be a web portal or an agreement with WRI? What can we expect in terms of data access for the scientific community? Like if, for example, wanted to host a server performing deepforest predictions for ecuador in realtime.

😍 Sara Beery

Sara Beery (sara.beery@gmail.com)

2020-09-24 16:59:15

@Ben Weinstein that sounds amazing

John Brandt (John.Brandt@wri.org)

2020-09-25 09:15:07

We have a call with Planet and NICFI this afternoon to discuss the terms of the data release. We know the contract is for 2 years, with an option of extension up to 4 years, and we know that data will be released one month after the mosaic period (so November data will be released in December). We don't yet know what the process will be like for getting access to the data - we believe it will be locked behind some sort of application process, but it will be free for non profits / academics to use if they go thru the process. I don't know when they are releasing this to everyone, though I know it is planned. For the moment they're working with us at WRI and people at FAO to user test the data access

Ben Weinstein (benweinstein2010@gmail.com)

2020-09-25 10:05:17

thanks. Keep up informed. Let me know if I can be of use.

Sara Beery (sara.beery@gmail.com)

2020-09-25 13:08:47

Deadline to apply for this year's Geo for Good Workshop is tonight! https://sites.google.com/corp/earthoutreach.org/geoforgood20/home

👍 Jon Van Oast, Benjamin Hoffman, Srishti

Sara Beery (sara.beery@gmail.com)

2020-10-01 21:59:03

We are putting together a proposal for the Fine-Grained Visual Categorization (FGVC) workshop at CVPR 2021, with a deadline of Oct. 16. (See https://ai.googleblog.com/2020/05/announcing-7th-fine-grained-visual.html for a recap of the most recent FGVC.) We’re looking for a o-organizer for the 3rd edition of the Herbarium challenge, with data provided by New York Botanical Garden. Kiat Chuan Tan served in this role the past two years, and kindly offered to give pointers to/share his codebase with the next co-organizer.

If this might be of interest to you, drop me a line and I’d be happy to tell you more about what it entails!

Google AI Blog

Announcing the 7th Fine-Grained Visual Categorization Workshop

Posted by Christine Kaeser-Chen, Software Engineer and Serge Belongie, Visiting Faculty, Google Research Fine-grained visual categorizat...

Original URL: https://ai.googleblog.com/2020/05/announcing-7th-fine-grained-visual.html

🎉 Jon Van Oast, Oisin Mac Aodha

👍 Srishti, Subhransu Maji

Carly Batist (cbatist@gradcenter.cuny.edu)

2020-10-02 09:08:28

Rainforest Connection has updated the ARBIMON platform (releasing Oct 15), now with UNLIMITED free (!!!!) storage and access to their template matching model (they're testing out random forest and neural network functionality as well). Join the listserv if you're interested in hearing about more updates.

👍 gvanhorn, Sara Beery, Sam Kelly

gvanhorn (grv22@cornell.edu)

2020-10-02 09:11:26

very cool!

Ben Weinstein (benweinstein2010@gmail.com)

2020-10-03 23:09:30

I’m looking for papers that do multi-sensor ensemble learning. No majority voting or regression, but actually using the features generated from two data streams to make a jointly learned prediction. I’m working on RGB + Hyperspectral tree species prediction and cannot find an ensemble that outperforms hyperspectral alone. I am currently trying to concat the dense features before the softmax layer and then learn a new joint representation. No luck. Maybe @Patrick Gray @Hannah Kerner?

👍 Sara Beery

David Healey (david.w.healey@gmail.com)

2020-10-03 23:25:19

Do the separate RGB and hyperspectral models have a lot of overlap in their errors on test sets? Are the hyperspectral errors basically a subset of the RGB errors?

Ben Weinstein (benweinstein2010@gmail.com)

2020-10-03 23:26:40

not yet. We have a third set of validation data. First trying on the same holdout set used to train each member of the ensemble. It should ATLEAST overfit and improve, even if we later find out it doesn’t generalize to a new test set.

Ben Weinstein (benweinstein2010@gmail.com)

2020-10-03 23:27:39

I’ve tried for example, batchnorm before concat because maybe the networks have learned features of different magnitudes.

Ben Weinstein (benweinstein2010@gmail.com)

2020-10-03 23:28:11

i’ve tried multiplying them instead of concat.

Ben Weinstein (benweinstein2010@gmail.com)

2020-10-03 23:29:11

https://github.com/weecology/DeepTreeAttention/blob/9b48d277df8024286a41e86100d6ad5e180ca156/DeepTreeAttention/models/Hang2020_geographic.py#L105 to make it literal.

GitHub

marireeves/WetlandsMappingHawaii

Implementation of Hang et al. 2020 "Hyperspectral Image Classification with Attention Aided CNNs" for tree species prediction - weecology/DeepTreeAttention

Original URL: https://github.com/weecology/DeepTreeAttention/blob/9b48d277df8024286a41e86100d6ad5e180ca156/DeepTreeAttention/models/Hang2020_geographic.py#L105

Ben Weinstein (benweinstein2010@gmail.com)

2020-10-03 23:29:49

i’ve tried freezing the layers versus finetuning with the new joint loss.

David Healey (david.w.healey@gmail.com)

2020-10-03 23:30:22

Yeah you're right to just be looking at the holdout sets first that you used to train them both. My question is: with that holdout set, are the sets of examples that the individual models get wrong subsets of each other?

David Healey (david.w.healey@gmail.com)

2020-10-03 23:31:35

are there a large number of specific examples that RGB gets right than hyperspectral gets wrong?

Ben Weinstein (benweinstein2010@gmail.com)

2020-10-03 23:31:43

don’t know yet, i’ll look, i’m sure they aren’t perfectly nested, but I hear you. RGB data (3 channels) + infrared (369 channels), so atleast they should be reasonably diverse.

Ben Weinstein (benweinstein2010@gmail.com)

2020-10-03 23:32:43

its funny because I assumed I’d get a modest improvement and the real goal was to study joint connections at differing layers so that the two data sources could teach each other features. I haven’t anticipated this going so badly.

David Healey (david.w.healey@gmail.com)

2020-10-03 23:32:45

I've had this problem before in other contexts and it just turned out that they were not orthogonal enough

👍 Ben Weinstein, Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2020-10-03 23:32:58

fascinating.

David Healey (david.w.healey@gmail.com)

2020-10-03 23:35:03

Another thing you can do is basically freeze both pretrained models, and initialize the final weights on the hyperspectral half of the concat to exactly what it is on the pretrained model, and the RGB half to 0. So you're literally starting with exactly the same individual hyperspectral model

👍 Ben Weinstein

David Healey (david.w.healey@gmail.com)

2020-10-03 23:36:06

Then train the last layer and see if it improves at all in the first few batches

David Healey (david.w.healey@gmail.com)

2020-10-03 23:38:54

Another thing might be to see the relationship between the number of channels you include in the hyperspectral model and the performance. With hyperspectral imaging its always possible there is so much overlap that having more than a couple dozen channels basically doesn't add anything new, and that could extend to RGB also.

Bistra Dilkina (dilkina@usc.edu)

2020-10-16 03:52:32

@Caleb Robinson

Mari Reeves (mari_reeves@fws.gov)

2020-12-11 18:18:17

I was doing a wetlands mapping project a few years ago where I took a different approach to this question. I had access to 8 band orthophotos, but also to several GIS layers, which had information about soils, vegetation type, and land cover classification. I ended up using more standard machine learning models (Random Forests, Boosted Regression, MARS) and created a training set of points across the landscape, which I classified as "wet" or "not wet" in roughly equal proportions. Then I used these points to "drill down" through all the different layers and ran those as predictors for my model. The hyperspectral layers were still the most informative predictors for water, and if you were going to use something like "soil type" as a predictor, then you need to make sure you fit the model using all classes you want to predict back to (otherwise the model chokes on new classes it hasn't seen yet when it goes to predict - why the numerical inputs are better than categorial) anyway, just figured I'd share, because the models actually did a really good job of predicting wetlands on the landscape ..even finding those covered with vegetation and these very small anchiline pools (few meters wide only). It was pretty compute intensive, especially when it came to the prediction step (for all Hawaiian islands on 5m grid) but certainly manageable with modern and cloud resources. The project is here, on github if you want to check it out. My code was pretty clunky back then, but you'll get the idea.

GitHub

This project aims to use multispectral satellite imagery and other local datasets to map wetlands in Hawaii - marireeves/WetlandsMappingHawaii

Original URL: https://github.com/marireeves/WetlandsMappingHawaii/blob/master/WetlandsMappingHNLDataSciNotes.pdf

Sara Beery (sara.beery@gmail.com)

2020-10-15 22:05:19

Awesome new Research Associate position opening at University of Minnesota at the intersection of computer vision and community science, collaborating with Wildlife Insights and Zooniverse! https://hr.myu.umn.edu/jobs/ext/337848

🎉 Jon Van Oast, Ben Weinstein, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Oisin Mac Aodha, Talia Speaker, Carly Batist, Björn Lütjens, Jason Holmberg (Wild Me)

Jon Van Oast (jon@wildme.org)

2020-10-15 22:14:45

whoa, cool.

Sara Beery (sara.beery@gmail.com)

2020-10-19 12:01:55

We've had a bunch of new people join in the last week! Want to introduce yourselves?

🎉 Jon Van Oast, gvanhorn

Ixchel Meza (ixchel.meza.ch@gmail.com)

2020-10-19 12:26:51

Hi, I’m Ixchel Meza. I’ve worked with satellite imagery to generate land cover and land cover change. Now I am working with camera trap images for classification and next I also will join the team that works with audio. I work at CONABIO (Mexico). 👋

🌍 Sara Beery, Oisin Mac Aodha, Omiros Pantazis, Björn Lütjens

🌿 Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

👋 Jon Van Oast, Sara Beery, gvanhorn, Siyu Yang, Jonathan Granskog, Mikey Tabak

Ben Weinstein (benweinstein2010@gmail.com)

2020-10-19 12:41:29

Welcome! What lab at CONABIO works on these issues? I work with researchers throughout latin america, mostly ecuador and colombia, and we don’t have any partners in mexico yet.

Ixchel Meza (ixchel.meza.ch@gmail.com)

2020-10-19 12:46:56

hi, thanks!. I work at the Dirección General de Proyectos Interinstitucionales, where do you work?

Ben Weinstein (benweinstein2010@gmail.com)

2020-10-19 12:50:28

I’m at University of Florida, but work closely with non-profits on biodiversity surveys. https://deepforest.readthedocs.io/, http://benweinstein.weebly.com/deepmeerkat.html. I’d be interested in hearing more about what the status of these kind of work is in mexico. The audio work is for bird detection?

Dr. Ben Weinstein

DeepMeerkat

Machine learning for video-based biodiversity surveys.

Original URL: http://benweinstein.weebly.com/deepmeerkat.html

Bistra Dilkina (dilkina@usc.edu)

2020-10-19 12:52:21

Hi, Ixchel. Welcome to this wonderful channel. I will also be very interested in learning more on work on these topics in Mexico, as my research group also works on similar topics at the USC Center for AI in Society.

Ixchel Meza (ixchel.meza.ch@gmail.com)

2020-10-19 14:03:59

We have a project called Sipecam that involves monitoring techniques to evaluate the effect of defaunation. If you want, we could have a meeting to talk about it a little bit more with other part of the team. https://sipecamdata.conabio.gob.mx/

Cindy Vargas (cvarga16@asu.edu)

2020-10-19 21:13:22

Hi, my name is Cindy Vargas! I am a first-year PhD student at Arizona State University working on marine conservation research. My research is going to focus on how we can develop and test a camera monitoring system that uses machine learning to assess fishing catch composition in small-scale fishers in Baja California Sur, Mexico.

🐟 Sara Beery, Oisin Mac Aodha, Omiros Pantazis, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

😎 Jon Van Oast, Stefan Schneider, Mikey Tabak

Omiros Pantazis (omiros.pantazis.16@ucl.ac.uk)

2020-10-20 06:43:57

Hi all, just realised I have been hanging around without having introduced myself! I am Omi, a 1st year PhD student at UCL, working on Biodiversity monitoring with Deep learning. I am also part of the Biome Health Project (https://www.biomehealthproject.com/) that tries to figure out the effects anthropogenic pressure has on biodiversity. Currently working on data coming from camera traps but planning to work with passive acoustic monitoring as well.

👋 Oisin Mac Aodha, Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, gvanhorn, Björn Lütjens, Mikey Tabak

😎 Jon Van Oast

Clara Panchaud (clarapasu@gmail.com)

2020-10-29 09:16:48

Hi, my name is Clara Panchaud, also new here! I am a master student at NTNU in Norway. I study statistics and I am interested in applications in ecology and conservation. Right now I am working on my master thesis on step selection functions and looking for a PhD for next year 😁

👋 gvanhorn, Benjamin Kellenberger, Sara Beery, Alex Borowicz, Jonathan Granskog, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Omiros Pantazis, Björn Lütjens, Mikey Tabak

Ben Weinstein (benweinstein2010@gmail.com)

2020-10-30 12:22:18

Any suggestions for hyperspectral data augmentations and class imbalance? @Patrick Gray what did you use for your land cover work? I am randomly flipping and rotate 369 band inputs, oversampling the rarest classes to about 20% of the max class rate. I’ve tried equal oversampling/undersampling and weighting losses.

👍 Mikey Tabak

Sara Beery (sara.beery@gmail.com)

2020-10-30 12:55:37

@Elijah Cole (Deactivated) do you have any tips here?

Bistra Dilkina (dilkina@usc.edu)

2020-10-30 14:17:52

@Caleb Robinson

❤️ Sara Beery

Caleb Robinson (calebrob6@gmail.com)

2020-10-30 19:27:12

I've never worked with hyperspectral imagery so this is total guess work...

Have you tried determining feature importance at the band level (then filtering out some bands) -- 369 dimensional inputs to a CNN seems too large. Are the spectral ranges of the bands adjacent to one another, maybe you could merge groups? Is spectral attention over input bands a thing @Elijah Cole (Deactivated)?

You could apply dropout over channels between input and first conv layer (Similar motivation to cutout augmentation, but channel wise)

Ben Weinstein (benweinstein2010@gmail.com)

2020-10-30 19:34:17

thanks, following a recent spectral/spatial HSI paper on architecture. https://arxiv.org/abs/2005.11977 still trying to find the right place for class imbalance. I will be explore band reduction too. Trying to debate whether its worth going down of the road of synthesizing images using a GAN. https://arxiv.org/pdf/1903.05580.pdf

arXiv.org

Hyperspectral Image Classification with Attention Aided CNNs

Convolutional neural networks (CNNs) have been widely used for hyperspectral image classification. As a common process, small cubes are firstly cropped from the hyperspectral image and then fed...

Original URL: https://arxiv.org/abs/2005.11977

👍 Caleb Robinson

Caleb Robinson (calebrob6@gmail.com)

2020-10-31 03:23:44

Relevant (and very recent) paper that I just stumbled on https://arxiv.org/pdf/2010.12337.pdf

👍 Sara Beery

John Brandt (John.Brandt@wri.org)

2020-11-02 09:49:59

I find that the effective number of samples works better than normal loss weighting methods: https://arxiv.org/pdf/1901.05555.pdf

👍 Sara Beery, Omiros Pantazis

Björn Lütjens (bjoern.luetjens@gmail.com)

2020-11-10 10:26:31

Would it make sense to do gaussian blur augmentations, i.e. run a gaussian filter over the spatial domain? Thinking about this, there might be some sophisticated augmentation that simulates atmospheric noise? This wouldn't address class imbalance, but rather "uniform" data augmentation.

Björn Lütjens (bjoern.luetjens@gmail.com)

2020-11-10 10:37:17

A friend just recommended ATRAN (https://atran.arc.nasa.gov/cgi-bin/atran/atran.cgi) to sim. atmospheric noise. Might be most relevant for satellite rather than aircraft data, depending on the height that you took your data.

👍 Sara Beery

Petar Gyurov (pgyurov93@gmail.com)

2020-11-10 12:23:03

Hey guys, I'm new here 👋 I've just started helping out the New Zealand DoC on a volunteer basis with some camera trap tooling. I'm a backend engineer by trade but have exposure to ML and data science. I studied theoretical physics at university - now I wish I had done computer vision instead!
Anyway, I'm new to the conservation field and I was pleasantly surprised to see how active it is; I'm excited to contribute and learn! See you guys around.

🐅 Sara Beery, Oisin Mac Aodha, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Björn Lütjens

👍 Mikey Tabak

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2020-11-10 12:32:50

A warm welcome to the Kiwiland!
Having a different background than CV is more helpful than it ﬁrst seems—it’s the combination and transgression of disciplines that makes conservation ﬂy in my opinion.

Ben Weinstein (benweinstein2010@gmail.com)

2020-11-10 12:36:08

honestly what we need is more backend engineers. We’ve got great models. how to share them, make them reproducible, create competent UI interfaces for less experienced programmers. Good data storage and metadata management.

👍 Sara Beery, Benjamin Kellenberger, Björn Lütjens

Ben Weinstein (benweinstein2010@gmail.com)

2020-11-10 12:36:21

Those are the things we need.

Sara Beery (sara.beery@gmail.com)

2020-11-10 12:36:38

Completely agree

Ben Weinstein (benweinstein2010@gmail.com)

2020-11-10 12:38:23

on my todo list is to have a look at https://www.biorxiv.org/content/10.1101/2020.10.02.323329v1 and see about all the above things

bioRxiv

U-Infuse: Democratization of Customizable AI for Object Detection

1. Image data is one of the primary sources of ecological data used in biodiversity conservation and management worldwide. However, classifying and interpreting large numbers of images is time and resource expensive, particularly in the context of camera trapping. Deep learning models have been used to achieve this task but are often not suited to specific applications due to their inability to generalise to new environments and inconsistent performance. Models need to be developed for specific species cohorts and environments, but the technical skills required to achieve this are a key barrier to the accessibility of this technology to ecologists. There is a strong need to democratise access to deep learning technologies by providing an easy to use software application allowing non-technical users to custom train custom object detectors. 2. U-Infuse addresses this issue by putting the power of AI into the hands of ecologists. U-Infuse provides ecologists with the ability to train customised models using publicly available images and/or their own camera trap images, without the constraints of annotating and pre-processing large numbers of images, or specific technical expertise. U-Infuse is a free and open-source software solution that supports both multiclass and single class training and inference, allowing ecologists to access state of the art AI on their own device, customised to their application without sharing IP or sensitive data. 3. U-Infuse provides ecological practitioners with the ability to (i) easily achieve camera trap object detection within a user-friendly GUI, generating a species distribution report, and other useful statistics, (ii) custom train deep learning models using publicly available and custom training data, (iii) achieve supervised auto-annotation of images for further training, with the benefit of editing annotations to ensure quality datasets. 4. Broad adoption of U-Infuse by ecological practitioners will improve camera trap image analysis and processing by allowing significantly more image data to be processed with minimal expenditure of time and resources. Ease of training and reliance on transfer learning means domain-specific models can be trained rapidly, and frequently updated without the need for computer science expertise, or data sharing, protecting intellectual property and privacy. ### Competing Interest Statement The authors have declared no competing interest.

Original URL: https://www.biorxiv.org/content/10.1101/2020.10.02.323329v1

👍 Sara Beery, Petar Gyurov

Petar Gyurov (pgyurov93@gmail.com)

2020-11-10 12:51:55

@Ben Weinstein That's really encouraging to hear. I have plenty of experience with that kind of stuff and will be happy to help. I will read through the paper you sent over as it's actually something I have been thinking about already.

Mikey Tabak (tabakma@gmail.com)

2020-11-18 09:57:27

Welcome @Petar Gyurov. This is a great place to learn from the experts. I'm an ecologist by training and I'm very fortunate to get great advice from these brilliant computer scientists.

Petar Gyurov (pgyurov93@gmail.com)

2020-11-18 10:21:25

Thanks @Mikey Tabak I've been taking a look at some of your work (MLWIC) -- great stuff 👍

Björn Lütjens (bjoern.luetjens@gmail.com)

2020-11-10 13:41:58

Hello Everybody,
Does somebody know of a pointer to a land cover/land use segmentation model in pytorch, that's ideally pretrained on pure-RGB NAIP imagery and/or Sentinel-2 imagery? Maybe @Ixchel Meza or @Patrick Gray? Thank you so much 🙂 🙂

Ixchel Meza (ixchel.meza.ch@gmail.com)

2020-11-10 14:19:22

No, sorry. For segmentation we used a commercial product http://www.imageseg.com/

biscloud

BIS Cloud

Automated geospatial image segmentation and feature extraction. Cloud service or Python API.

Original URL: http://www.imageseg.com/

👍 Björn Lütjens

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2020-11-10 15:15:06

@Caleb Robinson has some good work on land cover mapping... his CVPR 2019 paper has released code. It's in tensorflow, not pytorch, but may still be useful!

paper: https://openaccess.thecvf.com/content_CVPR_2019/papers/Robinson_Large_Scale_High-Resolution_Land_Cover_Mapping_With_Multi-Resolution_Data_CVPR_2019_paper.pdf

code: https://github.com/calebrob6/land-cover

GitHub

calebrob6/land-cover

Code for training and testing deep learning based land cover models. - calebrob6/land-cover

Original URL: https://github.com/calebrob6/land-cover

👍 Björn Lütjens, Sara Beery, Mikey Tabak, Aaron Ferber

Björn Lütjens (bjoern.luetjens@gmail.com)

2020-11-10 17:27:18

oh true! That repo is really amazing. Unfortunately, our whole code base is in pytorch and I'd need to access the weights of the model at different layers. So it seems quite tricky to integrate. But i'll definitely give it another look, thanks Lily!

Thanks for the link Ixchel! This might actually be interesting for another project/idea. Seems like I can upload an image to test the quality.

😊 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Sara Beery

☺️ Ixchel Meza

Patrick Gray (patrick.c.gray@duke.edu)

2020-11-17 10:09:37

Hmm all my stuff is in Keras and nothing pretrained on RGB imagery. Though @Ben Weinstein may know of something good!

👍 Björn Lütjens

Caleb Robinson (calebrob6@gmail.com)

2020-11-17 14:40:11

Thanks for the vote of confidence @Björn Lütjens! I'm actively working on a similar repo in PyTorch with which I'll train similar LC models -- I'll ping you in a few weeks when this is done.

🤗 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

😄 Björn Lütjens

Björn Lütjens (bjoern.luetjens@gmail.com)

2020-11-18 14:37:58

that sounds amazing! thank you so much @Caleb Robinson

Caleb Robinson (calebrob6@gmail.com)

2021-02-10 13:52:33

@Björn Lütjens I just noticed this thread in my thread list -- forgot all about it, sorry! The repo in PyTorch is here https://github.com/calebrob6/dfc2021-msd-baseline. We use it for training baseline models for the data fusion contest, however it works "as is" for training with high-res labels too

calebrob6/dfc2021-msd-baseline

A baseline for the "Multitemporal Semantic Change Detection" track of the 2021 IEEE GRSS Data Fusion Competition.

Stars

14

Language

Jupyter Notebook

🎉 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Björn Lütjens

Björn Lütjens (bjoern.luetjens@gmail.com)

2021-02-11 09:29:47

amazing; thank you so much Caleb!! :D:D

👍 Caleb Robinson

Mikey Tabak (tabakma@gmail.com)

2020-11-18 09:55:57

Hi folks, for those of you using aerial imagery to detect animals, have you ever combined overhead and oblique imagery in the same model? And if so, do you have any suggestions? I'll be working with several datasets to detect and count (and delineate species and sex for) a few bird species. My plan is to use object segmentation, but I'm not sure this will be enough to account for the fact that some images will be from overhead (drones pointed down) and some will be oblique (from out the window of a fixed wing aircraft). I don't have any of the images yet, just planning in advance. Thank you

Oisin Mac Aodha (macaodha@caltech.edu)

2020-11-18 09:58:38

Perhaps of interest when using "top" and "side" views: https://www.cv-foundation.org/openaccess/content_cvpr_2016/papers/Wegner_Cataloging_Public_Objects_CVPR_2016_paper.pdf

👍 Mikey Tabak, Sara Beery

Mikey Tabak (tabakma@gmail.com)

2020-11-18 10:03:21

Thank you @Oisin Mac Aodha, I'm checking this out now

👍 Oisin Mac Aodha

Björn Lütjens (bjoern.luetjens@gmail.com)

2020-11-18 14:47:34

How big are the birds in each image stream? If the size is very different it might be useful to have a pipeline of (1) two different bird detection algorithms for each image stream, (2) crop and resize, s.t., the birds in each small bird tile are same size, (3) use bird classification algorithm trained on both image streams; maybe even combined with ImageNet etc.

I'm totally not an expert here, so just brainstorming 🙂

👍 Mikey Tabak

Mikey Tabak (tabakma@gmail.com)

2020-11-18 15:21:45

Thanks @Björn Lütjens. I'm not sure how large (how many pixels) the birds will be (I should be receiving some images soon). But another challenge is that of the overhead images, they will be taken from different elevations, so the birds will be of different size. I like the idea of having a preliminary detection model to pull out the bird, standardize the size, and then run a classification algorithm. My major concern with combining the two types of datasets, though, is that birds are going to look different from overhead vs from the side, so I might need a different classification model for each type of view also.

Björn Lütjens (bjoern.luetjens@gmail.com)

2020-11-18 15:25:16

agreed! do you have sample images? 🙂 🐦

Mikey Tabak (tabakma@gmail.com)

2020-11-26 11:23:44

Unfortunately no sample images yet. I'm also not sure I'll get to share them when I do

Sara Beery (sara.beery@gmail.com)

2020-11-23 10:35:42

Vaquita Hacks hackathon is looking for mentors!

"The Vaquita is a critically endangered species, and time is running out to save it. Join Earth Hacks, The Conservation Project International, The Countering Wildlife Trafficking Institute, Earth League International, and many more for Vaquita Hacks on Dec 12-13, the first hackathon dedicated to vaquita conservation! Projects from the hackathon will be going directly to organizations working on vaquita conservation."

They are looking for mentors to help student participants in this program. Folks who are interested in mentoring should email Sanjana Paul <sanjana@earthhacks.io>.

Applications for student participation are open (now extended until November 23).

T-CPI

The Conservation Project International

The Conservation Project International (T-CPI) is a platform dedicated to supporting and mentoring young conservationists and future leaders. Our community-based platform offers a dedicated programme of support to help young people find valuable conservation work experience and volunteer roles.

Original URL: https://www.tcproject.co.uk/

Mysite

Countering Wildlife Trafficking Institute

Geospatial Analysis for Countering Wildlife Trafficking

Original URL: https://www.cwtinstitute.org/

Earth League International

Home - Earth League International

Original URL: https://earthleagueinternational.org/

👍 Mikey Tabak

Mikey Tabak (tabakma@gmail.com)

2020-11-26 11:42:15

Anyone have ideas for finding a small object in video that does not look distinct from the background in any one frame? I'm searching for video clips (around 3 seconds) that contain a bat. The only way for humans to tell that it is a bat is by looking at the whole video. Segmentation on images doesn't work because often the bat looks like a cloud in an image. Classification of videos doesn't work because the bats are so small and only in the video for a very short time (I also have a very small sample size). I've tried tracking but haven't had much success. I was thinking of combining a segmentation model (like DeepLabV3) with a video classification model (3D-CNN); although I'm not really sure what this looks like. I don't actually need segmentation, though, because I don't care where it is in the video, I really just want to classify the videos. Thanks for any ideas in advance and a Happy Thanksgiving to the folks in USA!

Elijah Cole (Deactivated) (ecole@caltech.edu)

2020-11-26 15:40:34

@Sara Beery has some work that might be useful: https://arxiv.org/abs/1912.03538

arXiv.org

Context R-CNN: Long Term Temporal Context for Per-Camera Object Detection

In static monitoring cameras, useful contextual information can stretch far beyond the few seconds typical video understanding models might see: subjects may exhibit similar behavior over multiple...

Original URL: https://arxiv.org/abs/1912.03538

👍 Sara Beery, Mikey Tabak

Sara Beery (sara.beery@gmail.com)

2020-11-26 16:00:45

Happy to chat and look at some data and see if something like our attention- based approach or a more traditional detection/tracking method might make sense!

👍 Mikey Tabak

Mikey Tabak (tabakma@gmail.com)

2020-11-27 06:07:22

Thank you @Sara Beery and @Elijah Cole (Deactivated)! I hadn't thought about using Sara's Context R-CNN, but this might be the best strategy. I'm going to keep trying to develop a pipeline combining deeplab and a 3d-cnn as I described above, but I don't think this will work. I appreciate your willingness to chat Sara. I'll reach out next week.

Mari Reeves (mari_reeves@fws.gov)

2020-12-11 17:43:50

Hi all - I'm new here, but am interested in this conversation for a few reasons. We have a thermal videography study funded for small bats (Pacific Sheath Tails) on Aguiguan in the Marianas. I think my colleagues at USGS in Colorado have been using a differencing algorithm to detect bats in video, and can catch up with them to see where they are at with that, per your question. But I also just got a project monitoring Boobies in Hawaii with a long series of mostly empty camera trap data, where there are intermittent Boobies, but fixed decoys in the images, so I think anything I train to pick out a booby is going to also pick out the decoys, so I need a way to pull them out and only select photos with the real boobies in them. I am interested to read Sara et al.'s paper and have a flexible schedule if there's any way you'd be willing to let me sit in on your conversation?

👍 Mikey Tabak

Sara Beery (sara.beery@gmail.com)

2020-12-11 17:47:02

We already talked, but re: the boobies...if the cameras are static can you just label where in pixel space the decoys are and throw out any detections with high IoU with the decoy locations?

Sara Beery (sara.beery@gmail.com)

2020-12-11 17:48:05

And I'm sure @Mikey Tabak would love to chat with your colleagues at USGS, that sounds like a very similar challenge!

Dan Morris (agentmorris@gmail.com)

2020-12-01 09:51:35

New aerial image dataset (~21k point annotations on seabirds) added to LILA:

http://lila.science/datasets/aerial-seabirds-west-africa/

LILA BC

Aerial Seabirds West Africa - LILA BC

Overview This dataset contains high-resolution aerial RGB imagery obtained from a census of breeding seabirds in West Africa in 2019. The data was collected as part of a census of breeding seabirds along the West African coast where UAVs were used to map colonies. The dataset consists of an orthomosaic covering a sandy island with...

Written by

lilawp

Est. reading time

1 minute

Original URL: http://lila.science/datasets/aerial-seabirds-west-africa/

👍 gvanhorn, Sara Beery, Subhransu Maji, Ritwik

🐦 gvanhorn, Elijah Cole (Deactivated), Elizabeth Bondi, Sara Beery

Jenna James (jennaj@vulcan.com)

2020-12-01 14:29:45

Hi there, my name is Jenna. I am a User Experience Designer at Vulcan Inc. supporting a project on dolphin signature whistle annotation. I am interested in gathering more information and gaining a better understanding about the marine mammal bioacoustics community, specifically acoustic dataset processes and annotations.

All information collected will be used for research purposes to help inform how we can best support this community. The survey should take around 10 minutes.

Thank you for taking the time to fill out this survey. Your input is greatly appreciated! Please feel free to share this survey.

https://forms.gle/jvRXh3kjYHKPqUZ78

Cheers,
Jenna, User Experience Designer at Vulcan Inc.

Google Docs

Research: Bioacoustics Workflows for Marine Mammal Researchers

Hello! I am a User Experience Designer at Vulcan Inc. supporting a project on dolphin signature whistle annotation. I am interested in gathering more information and gaining a better understanding about the marine mammal bioacoustics community, specifically acoustic dataset processes and annotations. All information collected will be used for research purposes to help inform how we can best support this community. The survey should take around 10 minutes. Thank you for taking the time to fill out this survey. Your input is greatly appreciated! Cheers, Jenna, User Experience Designer at Vulcan Inc. If you have any questions, please contact me at <a href="mailto:jennaj@vulcan.com">jennaj@vulcan.com</a>

Original URL: https://forms.gle/jvRXh3kjYHKPqUZ78

🐬 Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Zac Winzurk

Sara Beery (sara.beery@gmail.com)

2020-12-01 14:30:43

@Holger Klinck

Mari Reeves (mari_reeves@fws.gov)

2020-12-11 17:53:37

Hi Everyone - I'm new here, I work on threatened and endangered species with the US Fish and Wildlife Service in the Pacific Islands. I have several projects involving camera trap and drone imagery for monitoring wildlife behavior and performing counts. In the process of labeling a dataset for training an object detector model, the question came up... how should you handle uncertainty about objects in your images. I have a collaborator using VoTT to tag images of breeding birds on the island of Lehua. There are birds that are pretty obvious, but there are also alot of rock outcrops, that look similar to the targets of interest in the pictures. One perspective is that you only want to tag targets you are certain about, but I am planning to use a YOLO object detector, and I think this will confuse the model and lead to poor predictions because this model does learn from context in the empty parts of pictures. So @Dan Morris had the idea to assign two separate tags and weight them differently in the modeling process (one for "definitely bird", and another for "maybe a rock" kind of thing), but he also suggested I throw this question out to the group here. I can't seem to find much discussion about the experimental design aspects of annotation, it's more nuts and bolts. Your feedback is appreciated. I've attached one of our tagged drone photos, so you can see the objects of interest. They are the white dots in this image.

🕊️ Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Sara Beery (sara.beery@gmail.com)

2020-12-11 17:58:28

@gvanhorn this reminds me a bit of some of what you were talking about this morning about hard/unlabeled stuff in your audio data, any thoughts?

To me I guess an important first thing to think about is what types of errors would be worse in this scenario (overpredicting birds vs. missing birds), and if you can have human experts address the challenging cases?

Mari Reeves (mari_reeves@fws.gov)

2020-12-11 18:00:22

I think it's worse to miss birds, since the ultimate objective is to develop models to identify them to species level if possible and count them, and my other fear is that they look so similar, it will actually confuse the model (resulting in worse performance) as it tries to distinguish between the two.

Mari Reeves (mari_reeves@fws.gov)

2020-12-11 18:02:41

I also was a little concerned when I read that YOLO models sometimes don't perform as well on small targets (which is obviously what we have here), so if anyone has used these models specifically to do a similar project, I am interested in your results.

Sara Beery (sara.beery@gmail.com)

2020-12-11 18:04:12

This is a drone image, do you have a camera trap example? Are you thinking of training a joint model or separate models for the camera traps and the drones?

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-11 19:06:31

In my experience with trees, its always better to include everything, or else you risk training your detector to ignore potential objects that look similar to the objects you are trying to predict. Depends a bit whether recall or precision is more important for your question.

👍 Sara Beery, Mari Reeves

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-11 19:06:59

I am also working on data preprocessing using deep autoencoders where you first screen your data for potential annotation anomalies.

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-11 19:07:27

so my recommendation is include, since you can always work to throw out later. You can never recover annotations you don't mark.

Ed Miller (ed@hypraptive.com)

2020-12-11 20:05:34

In the dlib labelling tool, ImgLab, you can mark regions to ignore. The dlib object detector training (MMOD) will not use these boxes for training, and it will also not use these areas for negative examples. I'm not sure if Yolo ot other object detectors have a similar mechanism.

Mari Reeves (mari_reeves@fws.gov)

2020-12-11 20:37:31

@Sara Beery the camera traps and drone images are different projects. I am not aware that we have any co-collected data of these different types.

👍 Sara Beery

Mari Reeves (mari_reeves@fws.gov)

2020-12-11 20:38:54

@Ed Miller and @Ben Weinstein thanks for the feedback, interesting.

John Payne (drjohnpayne@gmail.com)

2020-12-12 01:39:38

It may depend on what your purpose is. I’m working on object recognition for aerial survey images, the purpose being to derive population estimates from counts of animals on transects. Since we need to assess the bias, precision and accuracy of our population estimates, there is no substitute in our case for hand-checking a large set of randomly-selected images so that we can estimate both a false-positive and a false-negative (missed animal) rate for the neural network model. For jobs where it isn’t as important to understand bias, precision and accuracy, you may be able to get away without manual checking. Multi-model comparisons, Bayesian updating, or other clever automated methods may be able to replace hand-checking in some cases, but human observers are still the gold standard when you’re on the wild frontier and the abilities of models haven’t been thoroughly assessed.

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-12 10:54:11

@John Payne what are you working on? We should chat, sounds basically like the project we have in the everglades on wading bird population dynamics from drone surveys. http://tree.westus.cloudapp.azure.com/everglades/

John Payne (drjohnpayne@gmail.com)

2020-12-12 13:27:01

Hi @Ben Weinstein, I have two survey projects; one with Howard Frederick on aerial surveys in Tanzania and another with a colleague who is using drones in Kazakhstan. I look forward to talking — it would be great to get a conversation going around these issues.

Mari Reeves (mari_reeves@fws.gov)

2020-12-14 14:01:23

@Ben Weinstein @John Payne I would love to chat with you both about methods and study systems. Although we are working towards using drone surveys of breeding birds on Lehua, our refuges program has a drone and there is just alot of interest in using this technology to conduct surveys. An intro meeting would be great.

Sara Beery (sara.beery@gmail.com)

2020-12-14 14:03:40

@Benjamin Kellenberger is my go-to expert on animal surveys in drone data, I'd recommend bringing him in to the meeting as well!

John Payne (drjohnpayne@gmail.com)

2020-12-14 14:16:45

@Mari Reeves I look forward to hearing about your project. @Sara Beery That’s a good idea. We’re in communication with Beni and have been experimenting with his AIDE program. I’ll send out an email to organize a call.

Nathaniel Rindlaub (nathaniel.rindlaub@tnc.org)

2020-12-14 18:01:28

Hey all - wow great timing. I work for The Nature Conservancy and we’re about to embark on an aerial data labeling project (sea bird census around Palmyra). We’re currently trying to evaluate tools for processing and labeling large geoTIFF orthomosaics; I hopped on this channel to see if anyone had any recs for workflows/software to assist and this conversation has already been super helpful. I didn’t know about AIDE or ImgLab. Does anyone have any other recommendations for drone data labeling tools we should check out?

Howard L Frederick (simbamangu@gmail.com)

2020-12-16 02:57:55

@Nathaniel Rindlaub are you looking for automated processing or manual or a mix thereof?

Nathaniel Rindlaub (nathaniel.rindlaub@tnc.org)

2020-12-16 10:45:35

@Howard L Frederick we're looking to develop a model that will assist in a manual review process, so a mix I suppose. However we're just in the early stages of this and are trying to figure out a workflow for building up a training dataset at the moment.

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:16:33

We have a private Zooniverse project up and running for everglades wading birds.

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:16:37

Works great and is easy.

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:19:32

also, btw, starting from a pretrain model is super useful for transfer learning. For the birds we start from a tree model. Shameless plug for https://deepforest.readthedocs.io/en/latest/

👍 Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:21:32

and once the first paper comes out i can share the egret label model, which should be useful for other nesting birds.

Nathaniel Rindlaub (nathaniel.rindlaub@tnc.org)

2020-12-16 19:50:00

@Ben Weinstein cool! Very helpful - I'll be sure to check out Zooniverse, and I'd love to try out the egret label model when it's ready. Also interesting that a model trained to detect trees is a good place to start for finding birds sitting in them!

I am just getting up to speed on our project (@Vienna Saccomanno can speak to all of this in more detail) but I believe we are trying to work towards a (1) a detection model that can pick out birds both roosting in palm trees and floating on the water... and then (2) likely a classifier that can classify them as either blue or red footed boobies.

Vienna Saccomanno (v.r.saccomanno@tnc.org)

2020-12-16 20:21:33

@Ben Weinstein this is incredibly helpful and interesting. For our work on Palmyra to census four different species of seabirds, we are mostly focused on the birds in the canopy (mostly Pisonia Heliotropium) and I am intrigued at your use of DeepForest. We've been concerned about animals partially covered by the canopy, so I'm very much looking forward to your first paper. Automated processing is the goal, but we are aware that some level of manual review might be necessary.

Sara Beery (sara.beery@gmail.com)

2020-12-16 20:29:57

I'm a pretty strong believer that all AI for conservation projects should have some built in, ongoing manual review, if nothing else than to verify difficult data and do quality control! My hope is that process would be much more lightweight than reviewing all the data, but it would help catch any obvious issues and make sure your model is maintaining accuracy over time

Vienna Saccomanno (v.r.saccomanno@tnc.org)

2020-12-16 20:42:24

Thanks for that, @Sara Beery. I think that is right and looking forward to learning more.

Howard L Frederick (simbamangu@gmail.com)

2020-12-17 00:13:13

@Nathaniel Rindlaub I processed a few big orthos not long ago using ImageMagick to cut the image into tiles then giving annotators folders of images in labelImg; I’d probably do it in CVAT if I repeat the process since you can manage multiple users, and overlap images between individual annotators to check consistency.

tzutalin/labelImg

🖍️ LabelImg is a graphical image annotation tool and label object bounding boxes in images

Website

<a href="https://youtu.be/p0nR2YsCY_U">https://youtu.be/p0nR2YsCY_U</a>

Stars

12791

Nathaniel Rindlaub (nathaniel.rindlaub@tnc.org)

2020-12-17 10:48:00

@Howard L Frederick amazing! Thanks for your recs! I'll definitely check them out.

Sara Beery (sara.beery@gmail.com)

2020-12-17 10:52:38

@gvanhorn has some awesome work on how to intelligently & efficiently combine annotations (including bboxes) from sets of annotators. I used it when building the Caltech Camera Traps dataset. https://openaccess.thecvf.com/content_cvpr_2018/html/Van_Horn_Lean_Multiclass_Crowdsourcing_CVPR_2018_paper.html

Howard L Frederick (simbamangu@gmail.com)

2020-12-17 11:55:22

@Nathaniel Rindlaub there are some reviews of annotation software linked here:

https://medium.com/tektorch-ai/best-image-labeling-tools-for-computer-vision-393e256be0a0

https://lionbridge.ai/articles/image-annotation-tools-for-computer-vision/

https://hackernoon.com/the-best-image-annotation-platforms-for-computer-vision-an-honest-review-of-each-dac7f565fea

Medium

Best Image Labeling Tools For Computer Vision

Image labeling or image annotation is the process of identifying or recognizing different units in an image. This process helps us to make…

Reading time

5 min read

Original URL: https://medium.com/tektorch-ai/best-image-labeling-tools-for-computer-vision-393e256be0a0

Lionbridge AI

24 Best Image Annotation Tools for Computer Vision | Lionbridge AI

Introducing the best image annotation tools that you can use to quickly and accurately build the ground truth for your computer vision models.

Original URL: https://lionbridge.ai/articles/image-annotation-tools-for-computer-vision/

hackernoon.com

The best image annotation platforms for computer vision (+ an honest review of each) | Hacker Noon

Original URL: https://hackernoon.com/the-best-image-annotation-platforms-for-computer-vision-an-honest-review-of-each-dac7f565fea

🙌 Nathaniel Rindlaub

Howard L Frederick (simbamangu@gmail.com)

2020-12-17 11:56:03

… however there is a wide variety of features and I have yet to find one that does everything I want. Some don’t zoom, most don’t deal with teams, some have very heavy web/cloud requirements (vs. in-house) …

Koustubh Sharma (koustubh@snowleopard.org)

2020-12-15 01:15:15

Hi all, thrilled to be here!
Thank you @Dan Morris for the kind invite. We work with snow leopards, and recently partnered with Microsoft who developed a cool tool for us to scan through thousands of images of snow leopards from camera traps for analysis using spatial capture recapture. Here to learn and explore the rapidly growing overlap between mathematical statistics, computer science, ecology and biology, and technology!
My 30 seconds' worth of claim to fame has been the opportunity to feature on MS AI ad last year 😊: https://www.youtube.com/watch?v=68WheTADA-g&t=7s

YouTube

Infinetix South Africa (https://www.youtube.com/channel/UCBMWdRWHF5PNxiwfVHq8frQ)

Microsoft + Snow Leopard Trust Protecting A Threatened Species

Original URL: https://www.youtube.com/watch?v=68WheTADA-g&t=7s

🌍 Omiros Pantazis, Sara Beery, Lauren E. Gillespie, Manish Rai, Jason Holmberg (Wild Me)

🐆 Carly Batist, Megan Cromp, Ankita Shukla, Stefan Schneider, Ming Zhong, Jason Holmberg (Wild Me), Elizabeth Bondi

😍 Chris Yeh, Jason Holmberg (Wild Me)

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2020-12-16 21:17:55

Nice to meet you, and great video! We're working on individual ID for snow leopards with Pose Invariant Embeddings. Would love to talk further.

👍:skin_tone_4: Koustubh Sharma

Koustubh Sharma (koustubh@snowleopard.org)

2020-12-16 22:29:01

Hi @Jason Holmberg (Wild Me), heard about your team from Dan. Let me know please when a good time to connect!

👍 Jason Holmberg (Wild Me)

Sara Beery (sara.beery@gmail.com)

2020-12-15 16:40:45

New distributional robustness benchmark released today that includes camera trap data! https://twitter.com/sarameghanbeery/status/1338944867037147138

🎉 Jon Van Oast, Olivier Gimenez, gvanhorn, Caleb Robinson

👍 Benjamin Kellenberger, Caleb Robinson

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:09:38

@gvanhorn , @Christine Kaeser-Chen others, any advice for screening training data for label confusion in fine-grained classification? I'm building an autoencoder to try to find outliers (HSI data on trees), to look for probable cases where one tree has been labeled e.g. 'black oak', but the spectral signature clusters better with 'white oak' (80 species). Right now i'm doing 1) Train stacked autoencoder (https://github.com/weecology/DeepTreeAttention/blob/c7e39b0402943076caa0d151272b34cf36738430/DeepTreeAttention/generators/cleaning.py#L14), 2) Define outliers as >95th quartile of the reconstruction error. The whole idea works great for finding outliers compared the entire dataset (like this image where the labeled trees were since cut down...), but I feel like its missing the class information, not just, this sample looks different from all others, but this sample looks different from all other black oaks. Maybe have the autoencoder also predict the class to form sparse embeddings? I'm not sure how to define the loss to make sure the groups are compact.

gvanhorn (grv22@cornell.edu)

2020-12-16 15:16:34

Yeah, autoencoders are a cool way to go for finding outliers in the dataset as a whole. For identifying class label mistakes, there are lots of adhoc tactics out there (which you have probably thought of) all of which require some brute force examination of the data. Cross validation and looking at the entropy of the predictions coming from a classification network is one standard tactic. Essentially like an active learning setup.

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:17:14

Okay, so it would be a bad idea to concat the actual class to the autoencoder and try to get it to predict both the image and the label?

gvanhorn (grv22@cornell.edu)

2020-12-16 15:24:52

Maybe worth a shot. Depends on how much time you want to spend on the problem/coding/debugging. If you have prior knowledge about which classes might get confused, it might be quickest to just browse through that subset looking for mistakes.
With enough data, I typically embrace the noise in the training set and worry more about the quality of the evaluation set.

👍 Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:27:41

thanks. I appreciate it. Its an interesting problem because its not traditional mislabeling in the sense that a annotator thought it was a sharp-shinned hawk and it was a cooper's, but more subtle. Ground truth is captured by people marking the tree stems with points. So there is alot of guesswork about what tree crowns are visible for a canopy. In this example, we have our detection algorithm (blue box) predict one tree, but there are two species ground truth underneath. One correctly predicted, one incorrectly predicted. Obviously because there is only one crown. Looking for an algorithmic way to anticipate these sceneries. But yes, hand review!

gvanhorn (grv22@cornell.edu)

2020-12-16 15:30:06

Is there a box associated with each dot? (one-to-one?)

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:30:34

nope! ground truth gets a point. The box is predicted from my tree detection model. Fun, right?

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:30:52

so you have to choose with pixels go with each point for training.

gvanhorn (grv22@cornell.edu)

2020-12-16 15:31:29

hmm, interesting. Your ultimate goal is boxes or points?

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:32:03

that's an interesting question. boxes i suppose, since that entire thing is a tree.

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:32:08

no one has ever asked me that before.

gvanhorn (grv22@cornell.edu)

2020-12-16 15:34:57

No idea if this is helpful, but have you seen this work?
https://arxiv.org/abs/1602.08405
https://arxiv.org/abs/1708.02750

arXiv.org

We don't need no bounding-boxes: Training object class...

Training object class detectors typically requires a large set of images in which objects are annotated by bounding-boxes. However, manually drawing bounding-boxes is very time consuming. We...

Original URL: https://arxiv.org/abs/1602.08405

arXiv.org

Extreme clicking for efficient object annotation

Manually annotating object bounding boxes is central to building computer vision datasets, and it is very time consuming (annotating ILSVRC [53] took 35s for one high-quality box [62]). It...

Original URL: https://arxiv.org/abs/1708.02750

gvanhorn (grv22@cornell.edu)

2020-12-16 15:35:26

Not sure if it is related, but your example figure reminded me of those works

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:36:08

totally, thanks for the ideas.

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:36:32

@Sara Beery I see you lurking on my tree problems.

Sara Beery (sara.beery@gmail.com)

2020-12-16 15:39:27

@Ben Weinstein always lurking 🙂

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:39:42

🙂

Sara Beery (sara.beery@gmail.com)

2020-12-16 15:40:01

But also I'm running into the same issues so I'm hoping you just solve them for me ☺️

John Payne (drjohnpayne@gmail.com)

2020-12-16 15:42:33

Ben I think your question of whether “this black oak is different from all other black oaks” is an important one. This is kind of an inelegant, brute-force suggestion, but how about running the autoencoder on one class at a time (I mean in addition to running it on the whole dataset)? Would that be prohibitively expensive?

Sara Beery (sara.beery@gmail.com)

2020-12-16 15:43:16

I was actually kinda thinking the same @John Payne but it might be the case that you wouldn't have enough data per-class?

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:44:58

i mean, definitely not enough data. But no, easy to do, no computational challenge, we have enormous HPC resources. I'm not entirely sure its different than reconstructing all images, plus the label. So your autoencoder has two heads and you take the joint loss.

Sara Beery (sara.beery@gmail.com)

2020-12-16 15:48:02

I guess that is probably theoretically similar, and might help with the lack of data, but in the per-class sense you're letting the autoencoder learn only relevant features for reconstructing a specific class, which might let it separate out the clusters a bit more? As opposed to needing to learn to reconstruct all species?

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-16 15:49:32

i'll give it a try today. writing it now.

John Payne (drjohnpayne@gmail.com)

2020-12-16 21:22:08

I had another idea; not sure if it’s any good. You may have seen this recent paper on anomaly detection: https://arxiv.org/abs/1911.10676. They train an autoencoder to restore degraded images (rotated or color loss), which forces it to learn semantic features of the class (they run it on one class at a time). Then during inference, normal (non-anomolous) images can be restored properly but anomalous images that are missing the proper semantic features will not restore properly and so they cause large restoration losses. Their method performs extremely well on several anomaly detection tests.

So: I wonder if you could use that method, but also include the label as part of the input: basically pass a one-hot encoded vector of class labels to the autoencoder along with the image. During training, it would learn to tell the classes apart and how to restore them. Then during inference, if the image was a good example of the class, presumably it would restore properly, but if it was labeled wrong then the model might choose the wrong set of features to restore. I suppose you might need a fairly deep model if you had a lot of classes to learn, but in the paper they mention a Resnet 34, so perhaps it’s not out of reach?

arXiv.org

Attribute Restoration Framework for Anomaly Detection

With the recent advances in deep neural networks, anomaly detection in multimedia has received much attention in the computer vision community. While reconstruction-based methods have recently...

Original URL: https://arxiv.org/abs/1911.10676

👍 Sara Beery

Vienna Saccomanno (v.r.saccomanno@tnc.org)

2020-12-16 19:07:05

Greetings everyone - I just joined this slack channel (thanks for the invite @Sara Beery and @Nathaniel Rindlaub). I am a scientist at The Nature Conservancy and I recently began a project focused on using machine learning models to detect and count sea birds in drone imagery. I am somewhat new to the AI space and look forward to learning from the people in this community.

🎉 Sara Beery, Jason Holmberg (Wild Me), Carly Batist

🤗 Nathaniel Rindlaub, Jason Holmberg (Wild Me), Sara Beery

Sara Beery (sara.beery@gmail.com)

2020-12-16 19:07:45

Welcome!!!

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2020-12-17 03:36:28

Welcome indeed! 😄

Carly Batist (cbatist@gradcenter.cuny.edu)

2020-12-17 10:51:28

Great to have another person with a science background getting into the AI/ML stuff! We can commiserate together 😅 though I work with mammals not birds..

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-17 14:27:11

Thanks everyone who joined the wildlife drones meeting. Just placing a few links in this thread. @Benjamin Kellenberger can you link to those papers and AIDE for later reference.

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-17 14:28:34

Our pipelines uses Agisoft for orthorectification, Zooniverse for annotation, deepforest for transfer learning (https://deepforest.readthedocs.io/), R shiny for web dashboard (http://tree.westus.cloudapp.azure.com/everglades) hosted on azure VM instance.

Howard L Frederick (simbamangu@gmail.com)

2020-12-17 14:56:46

Very interesting discussion - needed another 4h or so …

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2020-12-18 04:58:52

Hi Ben,
Thanks a lot for making the meeting of yesterday happen! It was great to see and get to know you all.
Here’s some links to my works:

Detecting animals in drone images and using as much of the background for training as possible: https://arxiv.org/pdf/1806.11368.pdf

Weakly-supervised object detection (“image has/doesn’t have animals, localize them for me”): http://openaccess.thecvf.com/content_CVPRW_2019/papers/EarthVision/Kellenberger_When_a[…]ifference_Improving_Weakly-Supervised_CVPRW_2019_paper.pdf

AIDE paper: https://besjournals.onlinelibrary.wiley.com/doi/pdfdirect/10.1111/2041-210X.13489

AIDE GitHub link: https://github.com/microsoft/aerial_wildlife_detection

Conservation Technology Needs Assessment

Tools for detecting wildlife in aerial images using active learning

Stars

92

Language

Python

Casey Youngflesh (caseyyoungflesh@gmail.com)

2020-12-17 16:02:49

Hello, everyone! I’m new to the AI for Conservation Slack and thought I’d introduce myself. I’m a quantitative ecologist and postdoc at UCLA, broadly interested in how ecological systems are responding to global change (particularly with regard to phenology and population dynamics). Methodologically, most of what I do involves applying hierarchical Bayesian models to ‘large-scale’ ecological data. I’m fairly new to AI but currently working on a project to identify animals in satellite imagery. Looking forward to connecting with the community!

🌍 Sara Beery

👋 Jon Van Oast

🙌 Mari Reeves

Catherine Breen (cbreen@uw.edu)

2020-12-17 18:17:45

Hello! Thank you @Dan Morris for the invite! I am a PhD student at the University of Washington working on identifying snow and storms for wildlife application from a large-scale camera network based in Norway. I am new to AI and the community, and grateful to Dan and @Sara Beery for opportunities to learn, like Sara's awesome tutoring session on building your own model this past summer with WildLabs. I like running and cooking, and I look forward to connecting more with all of you!

🎉 Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

❄️ Sara Beery

👋 Jon Van Oast

🙌 Mari Reeves

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-17 18:32:54

Hey Katie, I assume you are working Laura Prugh at UW? If not that project is really similar!

👋 Catherine Breen

Catherine Breen (cbreen@uw.edu)

2020-12-18 15:55:36

Hey Ben, Yes that's the one! I am Laura Prugh's student. Currently the model is a (pretty simple) custom network, but I am looking for ideas to improve the accuracy. Would love to connect after the holidays if you have time!

Kim Goetz - NOAA Federal (kim.goetz@noaa.gov)

2020-12-18 03:09:30

Hi All, Thanks to @Dan Morris for the invite! I am a marine mammal spatial ecologist at NOAA and I am currently working on a project to detect whales in satellite imagery on an operational scale. We are starting with two endangered whale species but are getting a little hung up with some of the details. Our plan is to create an annotated image library that can feed into ML processes. We realize this is a bit of a 'choose your own adventure' in terms of who you use for the various aspects of this project (cloud storage, image tiling, annotating, ML) so are hoping to use this forum as a way to solicit advice from the bigger group. For example, to create an annotated dataset, you can use a myriad of programs - ie viame, picterra, VGG- but some of these have issues with dealing with satellite imagery. In addition, ML usually requires bounding boxes as opposed to dots so how big do you make a box around the object? Additionally we need to export the geographic coordinates of each bounding box, not pixel coordinates. If anyone has any thoughts or advise please share =).

🐋 Sara Beery

🙌 Mari Reeves

Nathan Hahn (nhahn@colostate.edu)

2020-12-18 13:29:32

Hello! I am an ecology PhD student at Colorado State University, and through my own research using a variety of camera trapping and GPS tracking tools have been pulled into the conservation tech world. Currently, I am working on a project to look at the interface between conservation practitioners and engineers, in terms of both feature importance and the collaborative process of developing and implementing technologies in practice. I have developed a short questionnaire around these topics, which should take ~6 minutes to fill out. We have received a lot of responses from practitioners, but are looking to get more thoughts from those on the tech side. Here is a link to the survey: http://colostate.az1.qualtrics.com/jfe/form/SV_7WBiscDDocCYIbb. If any of you have additional thoughts please share!

colostate.az1.qualtrics.com

Qualtrics sophisticated online survey software solutions make creating online surveys easy. Learn more about Research Suite and get a free account today.

Original URL: http://colostate.az1.qualtrics.com/jfe/form/SV_7WBiscDDocCYIbb

👋 Oisin Mac Aodha, Sara Beery

✅ Carly Batist

Ed Miller (ed@hypraptive.com)

2020-12-18 17:46:56

Hi @Nathan Hahn, I responded to the survey. As a technologist with no formal conservation background, I found many of the questions were not applicable for me. Many were more suited to the conservation scientist I collaborate with.

Howard L Frederick (simbamangu@gmail.com)

2020-12-19 06:06:36

Hi @Nathan Hahn great topic - the ecologist / engineer interface is a really important often often blocking link. Keen to hear more.

Armin Bazarjani (bazarjan@usc.edu)

2020-12-21 17:23:52

Hey all I’m new to the group so I apologize if this isn’t the correct channel for something like this. I’m currently finishing up my Masters in Electrical Engineering at USC and I’m very interested in computer vision. I would love to be able to apply that interest to my budding passion of sustainability and conservation.

Because I am thinking of applying to PhD programs next year I would love to get more research experience in this area, so I can then apply my own research to it in the future! So, if any of you are looking for volunteers or know of people who could use an extra hand, please let me know!

🎉 Sara Beery, Tony Chang

Heather Lynch (heather.lynch@stonybrook.edu)

2020-12-22 16:02:39

Dimitris Samaras and I collaborate on a number of problems in the area of computer vision and conservation, might be worth considering doing a PhD with him (https://www3.cs.stonybrook.edu/~samaras/). His lab does a lot of interesting things applying computer vision to both health and environmental applications.

👍 Armin Bazarjani

Armin Bazarjani (bazarjan@usc.edu)

2020-12-22 16:58:00

Thanks for the suggestion @Heather Lynch, I'll check his lab out!

Tony Chang (tony@csp-inc.org)

2020-12-25 16:12:52

@Armin Bazarjani we are working with on a suite of projects and would love to chat if you are looking for projects to gain some experience in sustainability and conservation.

Sarra Alqahtani (sarra-alqahtani@utulsa.edu)

2020-12-23 12:24:47

Hi all! I'm new to this group and I would like to introduce myself. I'm Sarra Alqahtani, a computer science assistant professor at Wake Forest University. I study the land change in Amazonian rainforests using computer vision, deep learning with imagery taken by UAVs and from different satellites. I currently got a huge fun from NASA and I'm looking for a postdoc to work with me and 3 other professors, 3 undergraduates. Please if you are interested, talk to me directly. Thanks!

🎉 Sara Beery, Tony Chang, Amrita Gupta

👍 Carly Batist

Armin Bazarjani (bazarjan@usc.edu)

2020-12-23 19:40:15

That’s awesome, excited to see what you do!

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-25 16:04:25

@Sarra Alqahtani in January, let's talk about your needs and how we can be helpful. My lab works on tree detection/species classification. Mostly in temperate zones, but have a paper in review in French Guiana. I was just talking to a colleague in the Ecuadorian amazon on potential applications there. See recent pubs for links, python packages, https://scholar.google.com/citations?hl=en&user=7POnELAAAAAJ&view_op=list_works&sortby=pubdate, https://github.com/weecology/DeepTreeAttention, http://visualize.idtrees.org/

Chimera: A Multi-Task Recurrent Convolutional Neural Network for Forest Classification and Structural Estimation

Implementation of Hang et al. 2020 "Hyperspectral Image Classification with Attention Aided CNNs" for tree species prediction

Stars

19

Language

Python

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-25 16:07:18

also @John Brandt and @Tony Chang for their experience on multi-temporal satellite fusion for land cover + deforestation analysis. https://www.mdpi.com/2072-4292/11/7/768

MDPI

Original URL: https://www.mdpi.com/2072-4292/11/7/768

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-25 16:09:46

We also are just starting (I can't get planet to return my emails!) work on UAV satellite fusion for wading bird colonies. Would love your input here, as the project is just forming.

Tony Chang (tony@csp-inc.org)

2020-12-25 16:11:02

Thanks for the intro @Ben Weinstein! Congrats Sarra on the NASA grant! Definitely interested in helping, I’ve been working lately on a US scale forest structure project, but could scale globally pending more data. We are focused on using Harmonized Sentinel Landsat data and fusing to other datasets. Haven’t looked into UAV data much, but would be interested to learn more.

Sarra Alqahtani (sarra-alqahtani@utulsa.edu)

2020-12-26 12:25:51

@Ben Weinstein @Tony Chang Thank you so much for the information! Would love to meet with you both to discuss the overlaps between our research.

Ben Weinstein (benweinstein2010@gmail.com)

2020-12-26 12:59:27

great, i'll organize something for January. Keeping @John Brandt and anyone else in mind as well. We did a recent open meetup on drones for wildlife monitoring and i think it was useful. Let's post agenda items of interest here. Immediate thoughts looking at that postdoc ad: 1) introductions, goals, and background, 2) integrating multi-sensor imagery for deep learning classification, 3) training multi-scale (resolution) classification workflows, 4) available datasets for benchmark and annotations. Feel free to add.

Sarra Alqahtani (sarra-alqahtani@utulsa.edu)

2020-12-26 14:06:02

I would include land change/deforestation detection using time series data as unsupervised way of classification.

👍 Ben Weinstein, Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-06 12:04:16

Okay everyone, let's try to set up a meeting in the next couple days. Anyone is welcome to come discuss the topics listed above, I think @John Brandt @Tony Chang @Sarra Alqahtani are interested. https://whenisgood.net/4k4kphn

whenisgood.net

WhenIsGood: Deforestation ML

Plan the time to have your meeting or event by co-ordinating availability with all the particpants using this fuss-free online tool.

Original URL: https://whenisgood.net/4k4kphn

Tony Chang (tony@csp-inc.org)

2021-01-06 12:06:50

Thanks for setting this up @Ben Weinstein!

John Brandt (John.Brandt@wri.org)

2021-01-06 13:00:06

Hey! Would love to join this. We are working on a massive tree planting project with Mastercard (https://www.mastercard.us/en-us/vision/corp-responsibility/priceless-planet.html) and are starting to do UAV classification work as well as change detection with medium res imagery

mastercard.us

Priceless Planet

Original URL: https://www.mastercard.us/en-us/vision/corp-responsibility/priceless-planet.html

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-07 10:10:55

bump here to fill out times @John Brandt and others

Sarra Alqahtani (sarra-alqahtani@utulsa.edu)

2021-01-07 12:20:56

Sorry for the late action. Just sent my available times

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-07 12:32:40

just confirming the time zone worked for you, tony and I are both west coast, it should have auto show you the correct timezone, I set it to PT. Did you mean 6pm ET?

👍 Ben Weinstein

Sarra Alqahtani (sarra-alqahtani@utulsa.edu)

2021-01-07 13:04:40

Yes ET

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-07 13:05:00

okay, just @John Brandt

Sara Beery (sara.beery@gmail.com)

2021-01-07 15:38:21

I added times in as well, I won't have much to add but I'd love to learn from you all :)

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-07 17:46:45

okay, @Sara Beery, @Tony Chang, @Sarra Alqahtani for the 4 of us Friday 1:30pm PT (tomorrow). I will email John, perhaps he isn't as on slack as we are. I think his feedback will be extremely useful, since he's basically devoted the last few years to this, plus i'm anxious to hear about the status of the WRI/Norway/planet tropical data program.

John Brandt (John.Brandt@wri.org)

2021-01-07 17:48:46

Hey @Ben Weinstein, sorry I'm on vacation right now so I haven't been on Slack much -- I'll be back in the office next week though and I'll catch up with you

🌴 Ben Weinstein, Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-07 17:49:26

no problem, when you get back drop me a note. I'll see if florence is around.

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-07 17:57:42

okay, looks like she left WRI.

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-08 16:27:56

Benjamin Weinstein is inviting you to a scheduled Zoom meeting.

Topic: Benjamin Weinstein's Personal Meeting Room

Join Zoom Meeting
https://ufl.zoom.us/j/4610898573

Meeting ID: 461 089 8573
One tap mobile
+12532158782,,4610898573# US (Tacoma)
+13462487799,,4610898573# US (Houston)

Dial by your location
+1 253 215 8782 US (Tacoma)
+1 346 248 7799 US (Houston)
+1 669 900 6833 US (San Jose)
+1 301 715 8592 US (Washington D.C)
+1 312 626 6799 US (Chicago)
+1 646 558 8656 US (New York)
Meeting ID: 461 089 8573
Find your local number: https://ufl.zoom.us/u/aewlP1nvmx

Join by SIP
4610898573@zoomcrc.com

Join by H.323
162.255.37.11 (US West)
162.255.36.11 (US East)
115.114.131.7 (India Mumbai)
115.114.115.7 (India Hyderabad)
213.19.144.110 (Amsterdam Netherlands)
213.244.140.110 (Germany)
103.122.166.55 (Australia)
149.137.40.110 (Singapore)
64.211.144.160 (Brazil)
69.174.57.160 (Canada)
207.226.132.110 (Japan)
Meeting ID: 461 089 8573

Join by Skype for Business
https://ufl.zoom.us/skype/4610898573

Tony Chang (tony@csp-inc.org)

2021-01-08 17:19:58

Thanks for this meeting! It was great meeting you all!

🙌 Sara Beery

Tony Chang (tony@csp-inc.org)

2021-01-08 17:22:41

@Sarra Alqahtani I’m having a conversation with Caleb Robinson who was working on land-use change questions in the US and he works with a combination of MODIS and NAIP data. Most of what he does is applying epitomic representations for low resolution labels to work in conjunction with high resolution images. Your problem is kind of the reverse case (high resolution labels with low resolution images), but I’ll see if he has any novel solutions. I agree, GANs may not be the best way…but maybe?

Sarra Alqahtani (sarra-alqahtani@utulsa.edu)

2021-01-08 17:33:52

Thanks everyone. It was a very informative meetign. @Tony Chang We are still adjusting our GAN model using downsampled drone images and hopefully we get better results.

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-14 12:08:10

@Sarra Alqahtani see the small blurb in there about gold mining https://maaproject.org/2021/norway-agreement/ this was the dataset I was mentioning, connected to @John Brandt’s work.

MAAP

MAAP #131: Power of Free High-resolution Satellite Imagery from Norway Agreement

This report demonstrates the powerful application of freely available, high-resolution satellite imagery recently made possible thanks to an agreement between the Government of Norway and several s…

Original URL: https://maaproject.org/2021/norway-agreement/

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-05 11:43:52

Cross-posting: Hi everyone,

We hope everyone had a restful holiday and we're looking forward to seeing you for our next discussion of machine learning for remote sensing applications! We'll kick back off this Friday January 8th at 11am PST / 2pm EST / 8pm CET. Jose Luis Holgado Alvarez (Technical University of Berlin) will present on Self-Supervised Adversarial Representation Learning for Binary Change Detection in Multispectral Images.

Bring your questions! As always, full details and updated schedule here.
Zoom link: https://umd.zoom.us/my/hkerner

🙌 Sara Beery, gvanhorn

👍 Vienna Saccomanno, Casey Youngflesh

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-05 11:44:20

@Sarra Alqahtani, we have a remote sensing machine learning zoom group for biology.

Petar Gyurov (pgyurov93@gmail.com)

2021-01-05 14:19:10

Hi all. For the past couple of months I have been working on a project involving MegaDetector that I am excited to share with you.

I have built a GUI around MegaDetector that speeds up the process of weeding out empty camera trap images.
The application detects animals in your photos, then lets you review the output of the model and make corrections. At the end your original images are sorted in folders.
It's designed to be relatively simple to use.

I have just published version v0.0.1-alpha which you can download from https://github.com/petargyurov/megadetector-gui/releases
If you don't have time to download it and try it, here is a short demo video: 📽️ https://streamable.com/h5pcdu

My next goal is to implement more models and the ability to train a custom model, as well as few other things.
Let me know what you think!

Streamable

kiwi-dem0-v0.0.1-alpha

Watch "kiwi-dem0-v0.0.1-alpha" on Streamable.

Original URL: https://streamable.com/h5pcdu

🙌 Omiros Pantazis, Sara Beery, Ben Weinstein, Olivier Gimenez, David Will

Petar Gyurov (pgyurov93@gmail.com)

2021-01-05 14:20:33

Also, for this project, I did some refactoring the base MegaDetector code, mostly just to improve usability.
If you want to use MegaDetector as a CLI or as part of your scripts, you may want to look at my other repo https://github.com/petargyurov/megadetector-api

petargyurov/megadetector-api

Abstracts Megadetector's model with a standalone, generalised API that can be built on top of. Can be used as a script, or wrapped as CLI tool or a GUI.

Language

Python

Last updated

8 days ago

Sara Beery (sara.beery@gmail.com)

2021-01-05 14:22:14

This is amazing!!!

❤️ Petar Gyurov

Sara Beery (sara.beery@gmail.com)

2021-01-05 14:30:12

Have you thought about adding the ability for the reviewer to box animals that weren't found initially? Might be useful for finetuning project-specific models down the line.

Petar Gyurov (pgyurov93@gmail.com)

2021-01-05 14:33:57

@Sara Beery Yes, that is something I want to add but it involves quite a bit of work. It's on the todo list 🙂

👍 Sara Beery

Caleb Robinson (calebrob6@gmail.com)

2021-01-05 17:19:56

@Siyu Yang

❤️ Sara Beery

Hannah Yin (hannah.yin@rice.edu)

2021-01-05 21:57:02

Hi everyone, and thank you @Sara Beery for the invite! I'm Hannah, a senior undergrad student double-majoring in Biology and Computer Science at Tufts University. Super grateful and excited to find an active community here bringing much needed technical guidance to ecology, conservation, and climate change response strategies. Looking forward to exploring how I can put my cs training to good use in these areas and learning from you all. I am fairly new to AI and just completed an introductory course in machine learning last semester. Much of my experience outside the classroom so far comes from working on research projects in wild vertebrate stress physiology and eco-evolutionary processes using empiricism (lab work and data) and theory (equations and biological concepts). That said, I do have some experience in the tech industry, albeit not directly related to bio or AI. Happy to chat and help when/where I can. Looking through the Wild Me documentation is already adding to my understanding of software design out in the real world. 😃

🎉 Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

:bearid: Ed Miller

Sarra Alqahtani (sarra-alqahtani@utulsa.edu)

2021-01-07 12:23:21

@Hannah Yin Happy to have you here. I would love to have you in my research group if you are willing to work remotely with my team. I have a group of wonderful undergraduates who taught themselves ML and RL and I think you will enjoy working with them. Please email me if you are interested: alqahtas@wfu.edu

Sara Beery (sara.beery@gmail.com)

2021-01-11 13:32:08

From the Computational Sustainability network:

CLAIRE Network is collaborating with AIhub.org on a focus series on “AI for Good: UN sustainable development goals”. This series highlights AI research relating to the UN sustainable development goals (SDGs). Each month we focus on a different UN SDG. The first SDG topic, released in January, was SDG: Good health and well-being. You can find the articles published so far collected here: https://aihub.org/tag/focus-on-good-health-and-well-being/

The next topic will be SDG: Climate action and they are looking for researchers who work in AI relating to climate action who are interested in writing or talking about their work. You could write (or describe on video) a short blog post about your research, write an opinion piece about an area of current interest, give a tutorial, or take part in an interview. If you are interested, please do get in touch at aihuborg@gmail.com. The deadline for contributions is 12 February 2021.”

aihub.org

Focus on good health and well-being | ΑΙhub

Original URL: https://aihub.org/tag/focus-on-good-health-and-well-being/

👍 Armin Bazarjani, Mikey Tabak

Sara Beery (sara.beery@gmail.com)

2021-01-12 19:36:15

New hub for climate datasets, with an AI focus: http://mldata.pangeo.io/

😎 Jon Van Oast, Jason Holmberg (Wild Me)

☀️ Ed Miller, Jason Holmberg (Wild Me)

🌎 Armin Bazarjani, Oisin Mac Aodha, Omiros Pantazis, Carly Batist

👍 Benjamin Kellenberger

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-14 21:34:36

Does anyone in channel use more than the standard DJI quadcopters for large scale surveys? @John Payne? Looking to spend alot more $ to be a more professional drone for large scale mapping for our everglades bird colony project. Wingtra drones?

Howard L Frederick (simbamangu@gmail.com)

2021-01-14 22:25:32

We have used microlights with wing mounted cameras! 3h+ endurance, triple the speed of quadcopter. Payload over 10kg. And no regulatory issues around use of UAV outside line of sight.

😎 Jon Van Oast, Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-14 22:33:33

thanks. I'd love to stop paying pilots, but yes, a good idea.

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-14 22:36:31

i didn't mention, we also fly in a CESNA fixed wing over the everglades.

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-14 22:36:36

its old, and a bit scary.

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-14 22:38:47

The thing going in our favor is that we are looking at wading bird colonies, which we tend to know their general location. We think (hope) that 90% of the nesting birds are in these known areas. Less spread out than many other wildlife monitoring programs.

Howard L Frederick (simbamangu@gmail.com)

2021-01-15 00:36:49

@Ben Weinstein what sort of area do you need to cover? What is the mission profile like?

John Payne (drjohnpayne@gmail.com)

2021-01-15 01:29:15

I don’t know Wingtra, but I think you’re on the right general track; fixed-wing drones are really the only option for covering distance. It’s been a while since I looked, but depending on how much you can spend you might look at a UAVFactory Penguin B or C, (https://uavfactory.com/en/penguin-c-uas) which blew away endurance records, or the UAS drones (http://www.ua-sp.com/c-astral) which have much less endurance but are aimed at people needing very precise positioning. If you can afford to spend as much as a modest house would cost, there are even better options :)

uavfactory.com

UAV Factory | Penguin C UAS

UAVFACTORY | High-performance composite airframe design and manufacture for the unmanned industry

Original URL: https://uavfactory.com/en/penguin-c-uas

ua-sp.com

C-Astral | Industrial Surveillance Drones | UAS

C-Astral unmanned systems are much smaller than manned aircraft, easier to maintain and transport and therefore much more cost-effective.

Original URL: http://www.ua-sp.com/c-astral

Howard L Frederick (simbamangu@gmail.com)

2021-01-15 01:30:37

@John Payne what sort of endurance / speed / payload do those have? #too-lazy-to-read-link

John Payne (drjohnpayne@gmail.com)

2021-01-15 01:42:30

For the penguin B: 10 kg, 25-50 hours(!), cruise speed of 43 knots, wide variety of payloads available including gyro-stabilized. Astral have only about a 3 hour range, but lots of available cameras including serious multispectral stuff and LIDAR.

Howard L Frederick (simbamangu@gmail.com)

2021-01-15 23:10:56

I’ve asked UAVFactory for the cost of a basic unit. Those sorts of speed/endurances mean we could cover 800km of flight lines (transects) in a day, which is meaningful. Would still be nearly impossible to get clearances for a large drone here in Tanzania but I’m keen to see what the possibilities are.

Howard L Frederick (simbamangu@gmail.com)

2021-01-18 08:37:39

… aaand UAVFactory is so far quite elusive about what their systems actually cost …

John Payne (drjohnpayne@gmail.com)

2021-01-18 15:03:44

From what I saw it’s around $15K to $20K for a basic system, but obviously depends on the extremely varied set of options chosen; esp. camera systems — I’ll be interested to hear what you find out.

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-18 15:26:01

one thing we are really worried about is wind sensitivity for the fixed wings.

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-18 15:26:18

My tech is convinced that wingtra drones will be too shaky in our typical 15mph wind.

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-18 15:26:26

Everglades is basically windy 100 days a year

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-18 15:26:59

so i contact https://www.atmosuav.com/ and they advertised a heavier drone for windier conditions.

atmosuav.com

ATMOS UAV | VTOL Drones for Mapping & Surveying

Atmos UAV empowers mapping and surveying professionals around the world to effortlessly gather geospatial data. To that end, the company has developed Marlyn, the VTOL fixed-wing drone (uav). 2020 cameras: SONY, Micasense (Altum and RedEdge MX) and more.

Original URL: https://www.atmosuav.com/

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-18 15:27:06

but its more like 25,000$

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-18 15:27:17

which is probably just outside our price range.

John Payne (drjohnpayne@gmail.com)

2021-01-18 15:46:27

Yes I think there are two issues: 1) tipping side-to-side makes the camera take photos that aren’t pointed straight down, which could violate your transect assumptions and/or add distortion at the edges of the images, and 2) shaking can cause blurring. Gyroscopes can help with blurring (and tipping to some degree), and gimbals can help with the tipping issue. But both also increase payload. I would think that the lighter weight the camera, the lighter the gyro/gimbal needed to control it, so it might be worth looking at integrated camera/gyro/gimbal systems that have been designed with light weight in mind. UAVFactory makes some (https://uavfactory.com/en/stabilized-payloads), but i haven’t checked them out and I don’t know what the prices are. (Howard, I notice they are ITAR-free). In general, I would think that a UAV with less lift would be less tippy, so a bullet-like UAV might be more stable than a glider-like UAV. I don’t know whether swept-wing designs like the C-Astral drone I pointed to or Marlyn are more stable than straight-wing designs in general — maybe Howard knows?

uavfactory.com

UAV Factory | Stabilized payloads

UAVFACTORY | High-performance composite airframe design and manufacture for the unmanned industry

Original URL: https://uavfactory.com/en/stabilized-payloads

Howard L Frederick (simbamangu@gmail.com)

2021-01-18 22:15:38

UAVFactory’s responses (I’ve had 3 sets of back-and-forth so far with no straight answers) were that my (fantasy) budget of $100K that I put down on the contact form was “far too little” for the base model.

Howard L Frederick (simbamangu@gmail.com)

2021-01-14 22:27:27

.... US regulatory environment much better I think, but for coverage of large areas definitely you should look into economics of microlight use.

Vienna Saccomanno (v.r.saccomanno@tnc.org)

2021-01-19 14:56:22

@Ben Weinstein my team with The Nature Conservancy uses a Wingtra to survey seabirds on Palmyra Atoll.

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-19 15:39:09

is it sensitive to the wind! can you send any sample images?

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-19 15:39:21

my techs are convinced it won't work in the wind.

Vienna Saccomanno (v.r.saccomanno@tnc.org)

2021-01-19 16:11:32

our tech says that, while Palmyra is a very windy place, the Wingtra handles steady winds up to 35MPH. Imagery is good, but suggest shifting to max shutter speed. Landing in high wind is risky. Very gusty winds will require a larger area for takeoff & landing. In general, his impression is that the Wingtra is much more capable in windy situations than a quadcopter.

I can look into a file share if that is still of interest.

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-19 16:11:48

please!

Ben Weinstein (benweinstein2010@gmail.com)

2021-01-19 16:35:57

If your tech is willing to talk by zoom/phone, I would love to hear.

Vienna Saccomanno (v.r.saccomanno@tnc.org)

2021-01-19 16:47:33

i'll move this to a DM

Sara Beery (sara.beery@gmail.com)

2021-01-21 17:59:56

New journal on "Environmental Data Science"

"OPEN FOR SUBMISSIONS: Environmental Data Science (cambridge.org/eds) is a new, peer-reviewed open access journal dedicated to the potential of artificial intelligence and data science to enhance our understanding of the environment, and to address climate change. Led by Claire Monteleoni (University of Boulder, Colorado) and a team of Editors, and published by Cambridge University Press, the journal is now open for submissions."

🎉 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Olivier Gimenez, Riccardo de Lutio, Ed Miller, Chris Yeh

Dan Morris (agentmorris@gmail.com)

2021-01-21 19:19:55

New dataset on lila.science, courtesy of the Wild Nature Institute (and lots of helpful volunteers at Zooniverse):

http://lila.science/datasets/wni-giraffes

This dataset contains keypoints for giraffe photogrammetry, so automating this is a potentially interesting ML problem that steps slightly outside the usual detection/classification paradigms, and WNI already makes extensive use of ML tools to accelerate their workflows, so if a student is looking for a nice self-contained ML project that could be used right away by a conservation organization, definitely take a look at this dataset!

Background on WNI's previous work integrating an AI pipeline for giraffe cropping and individual ID:

https://www.sciencedirect.com/science/article/abs/pii/S1574954118300426?via%3Dihub

LILA BC

WNI Giraffes - LILA BC

Overview Masai giraffes are endangered from illegal killing and habitat loss. As part of the world’s largest giraffe conservation science project, the Wild Nature Institute has identified thousands of individual Masai giraffes in Tanzania by their unique and unchanging coat patterns. Photogrammetry allows researchers to also track these giraffes’ heights and sizes over time from...

Written by

lilawp

Est. reading time

2 minutes

Original URL: http://lila.science/datasets/wni-giraffes

sciencedirect.com

An automated program to find animals and crop photographs for individual recognition

Detailed data on individual animals are critical to ecological and evolutionary studies, but attaching identifying marks can alter individual fates an…

Original URL: https://www.sciencedirect.com/science/article/abs/pii/S1574954118300426?via%3Dihub

:giraffe_face: Stefan Schneider, Sara Beery, Oisin Mac Aodha, Carly Batist, Ming Zhong, Mikey Tabak, Siyu Yang

👀 Alex Zhuang

Sara Beery (sara.beery@gmail.com)

2021-01-27 16:00:30

This year's FGVC Workshop just released its call for papers!! 4 page short submissions, deadline is April 2nd.

"The purpose of this workshop is to bring together researchers to explore visual recognition across the continuum between basic level categorization (object recognition) and identification of individuals within a category population. Topics of interest include:

Fine-grained categorization

Novel datasets and data collection strategies for fine-grained categorization

Appropriate error metrics for fine-grained categorization

Low/few shot learning

Self-supervised learning

Semi-supervised learning

Transfer-learning from known to novel subcategories

Attribute and part based approaches

Taxonomic predictions

Addressing long-tailed distributions

Human-in-the-loop

Fine-grained categorization with humans in the loop

Embedding human experts’ knowledge into computational models

Machine teaching

Interpretable fine-grained models

Multi-modal learning

Using audio and video data

Using geographical priors

Learning shape

Fine-grained applications

Product recognition

Animal biometrics and camera traps

Museum collections

Agricultural

Medical

Fashion"

https://sites.google.com/corp/view/fgvc8

🐦 Oisin Mac Aodha, Riccardo de Lutio

👍 Oisin Mac Aodha, Armin Bazarjani, gvanhorn, Mikey Tabak

👍:skin_tone_4: Ixchel Meza

Ana Usenko (usenkoa@wwu.edu)

2021-02-01 19:23:34

Hi everyone! I'm new to this group but very excited to be here and wanted to introduce myself! I'm a recent graduate from Western Washington University with BS & BA degrees in computer science and linguistics. I'm not particularly new to AI and have research experience through the university and the Pacific Northwest National Lab in deep learning and natural language processing. The work I've been involved in includes fewshot learning, audio classification, and authorship identification. Although I've not been directly involved in Conservation-focused AI pursuits, I am really inspired by the work @Gracie Ermi and others have done at Vulcan for wildlife conservation. Over the past year or so, I've become more interested in the topics of AI and ethics and although conservation isn't directly related to that, I'm always interested in seeing how others are using their skills in AI to improve the world around them and solve important, real-word, ethical problems. Looking forward to getting a peak into what you all are working on here!
Additionally, I'm currently searching for a full-time position as a recent grad and trying to break through into the field of AI. If anyone knows of any openings in ML/AI/NLP, Data Science, or even software development, please reach out to me! Would absolutely appreciate it, and hope to have the chance to contribute to this group more in the future.

👋 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Howard L Frederick, Ed Miller

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2021-02-02 13:18:51

Very excited that @Heather Lynch will be giving a talk at Harvard CRCS on Monday on penguins, satellite imagery, and AI!
How many penguins are there? (And other mysteries solved by satellites and AI
Monday, February 8, 2021 11 AM EST (UTC-5)

Sign up here: https://crcs.seas.harvard.edu/ai-social-impact

Satellite imagery and computer vision are two transformational technologies that have rapidly, and quite radically, expanded our capacity to study wildlife in the world’s most remote places. In this talk, I will describe my lab’s efforts to combine satellite imagery, drones, and other remote sensing technologies with good old fashioned field work to study the distribution and abundance of penguins and other wildlife in Antarctica. I’ll also discuss the threats facing Antarctic penguins and how scientists are bringing together new technology, artificial intelligence, and advanced predictive modelling to help guide policymakers in their work to protect one of the world’s last remaining wildernesses.

👍 Benjamin Kellenberger, Sara Beery, Talia Speaker, Armin Bazarjani, Casey Youngflesh, Ben Weinstein, Carly Batist, Michael Wethington, Frederic, Ana Usenko, Mikey Tabak

❤️ Sara Beery

Silvia Zuffi (silvia@mi.imati.cnr.it)

2021-02-04 02:12:14

https://www.cv4animals.com/

cv4animals.com

weecology/NeonTreeEvaluation

About

Original URL: https://www.cv4animals.com/

👍 gvanhorn, Sara Beery, Hemal Naik

Declan (declan.pizzino@consbio.org)

2021-02-05 16:37:14

Hi all, my name is Declan Pizzino and I'm a geospatial analyst with the Conservation Biology Institute. Stoked to share and see how other folks are applying AI to the field of conservation. Many thanks to @Björn Lütjens for the invite!

👋 Sara Beery, gvanhorn, Carly Batist, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Alex Borowicz, Petar Gyurov, Björn Lütjens, Oc, Mikey Tabak

👏 Sarra Alqahtani

Gyri Reiersen (gyri.reiersen@tum.de)

2021-02-06 10:10:38

Hey guys! 👋 Gyri here! Originally Norwegian, growing up on a farm between mountain and fjords, and currently a master student in AI in Munich, Germany. Really excited to be writing my thesis with @Björn Lütjens and @David on the topic of forest monitoring!

🌳🌲🌴We want to put together an open source “awesome-forest-datasets” github repo.

Do you know if such a collection already exists?

What forrest datasets/collections do you know of? (incl. satellite, drone, airborne, field data, etc)

🌍 Björn Lütjens, Ben Weinstein, Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, David, Petar Gyurov, Mikey Tabak, Lukas Liebel

Ben Weinstein (benweinstein2010@gmail.com)

2021-02-06 10:49:24

I can help here too. I know of most of the datasets, ping me if I don't update here during the week. https://github.com/weecology/NeonTreeEvaluation https://www.newfor.net/download-newfor-single-tree-detection-benchmark-dataset/

Benchmark dataset for tree detection for airborne RGB, Hyperspectral and LIDAR imagery

Stars

37

Language

Python

🙌 Gyri Reiersen, Sara Beery, David, Björn Lütjens

Ben Weinstein (benweinstein2010@gmail.com)

2021-02-06 11:00:46

https://www.planet.com/nicfi/

Planet

NICFI Program

Planet

Original URL: https://www.planet.com/nicfi/

🙌 Gyri Reiersen, Sara Beery, David, Björn Lütjens

🌴 Daniel Grzenda, David

Sara Beery (sara.beery@gmail.com)

2021-02-06 11:04:39

I'm currently curating a bunch of urban forest data, happy to add once it's ok'd for release!

🦅 Björn Lütjens, David

Daniel Grzenda (grzenda@uchicago.edu)

2021-02-06 14:30:06

I'm currently working on a project to use the NICFI dataset to monitor land use in the tropics. We're hoping to use transfer learning and looking for a model trained on 5 layer surface reflectance images (best resolution in the NICFI dataset). I'd be happy to share what we've found so far/bounce ideas off each other if you end up doing something similar.

Ben Weinstein (benweinstein2010@gmail.com)

2021-02-06 16:26:40

@Daniel Grzenda love to hear more, what measure are you trying to predict. I have access to the data, but don't see a massive use case yet.

Daniel Grzenda (grzenda@uchicago.edu)

2021-02-08 10:17:09

We're working on labeling a subset of the quads with segmentation masks of different industries (mines, mills, plantations, etc) and the plan is to use dice as our metric

Mikey Tabak (tabakma@gmail.com)

2021-02-10 09:16:31

@Sara Beery I have a question about MegaDetector. Why did you choose to use Faster-RCNN instead of a YOLO or SSD model? Did you find that it was more accurate? I couldn't find details on model selection in your paper, but sorry if you've already described this somewhere else. I ask because I'm working on similar problems (except that I'm trying to incorporate more classes) and finding that YOLO and SSD outperform Faster-RCNN on my data, so I'm wondering if there is something I'm missing (which is usually the case). Thank you!

Hemal Naik (hnaik@ab.mpg.de)

2021-02-10 10:57:03

Great question, I am also interested. I recently found yolo3 worked quite well in comparison with faster-rcnn type architecture.

👍 Mikey Tabak, Sara Beery

Sara Beery (sara.beery@gmail.com)

2021-02-10 12:28:50

The reason was that I wanted to work with the object centric features provided by two stage object proposal then object classification structure of Faster R-CNN. I think there's definitely a lot of interest in a lighter-weight version though! Adapting it in an object-centric way would mean thinking about how to appropriately extract object specific features to compare to, and where in the network you would want to place the attention block. I think there's even possibly a classification version of this where you add attention across entire frames before the final network layers.

👍 Mikey Tabak

Mikey Tabak (tabakma@gmail.com)

2021-02-10 13:47:39

Great. Thanks for explaining this Sara!

Sara Beery (sara.beery@gmail.com)

2021-02-12 14:19:35

Haha omg, I'm sorry, I just realized I answered a question you didn't ask! I just answered "Why did you start from Faster R-CNN for Context R-CNN"

Sara Beery (sara.beery@gmail.com)

2021-02-12 14:24:36

The answer to your actual question about the MegaDetector was that on the data we were evaluating on, at the time we trained the model, Faster R-CNN was the highest performing architecture (and did outperform YOLO). I think possibly because we're training on lots of data from different parts of the world the higher capacity model was able to capture more information. However, there have been a couple big advancements in object detection since then (2018!) and @Dan Morris and I were just talking a couple weeks ago about doing a new architecture comparison for MegaDetector V5, and including some of the newer, high-performing models such as RetinaNet, CenterNet, EfficientDet, etc. as well as re-evaluating on SSD/YOLO.

👍 Petar Gyurov

Olga Khroustaleva (okhroust@gmail.com)

2021-02-15 03:59:21

Hi everyone! New here, wanted to introduce myself. I am Olga, a long-time Googler with background in user experience. In addition to my Google job, I am also a Master's student in the Sustainable Resource Management program at TU Munich, where I plan to focus on wildlife preservation. Thank you @Sara Beery for inviting me to this community!

🌍 Omiros Pantazis, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Elijah Cole (Deactivated), Sara Beery, gvanhorn, David, Yumna, Megan Cromp, Gyri Reiersen, Björn Lütjens

Ben Weinstein (benweinstein2010@gmail.com)

2021-02-15 12:53:14

Hi Olga! We all definitely need more help in front end development. Welcome.

Sara Beery (sara.beery@gmail.com)

2021-02-16 18:19:25

Can anyone point me to work in NLP for conservation applications? Specifically anything on extracting scientific data from historic records/papers?

Ben Weinstein (benweinstein2010@gmail.com)

2021-02-16 19:26:37

They must be doing some NLP here, https://www.zooniverse.org/organizations/md68135/notes-from-nature I know there is transcription.

zooniverse.org

Zooniverse

The Zooniverse is the world’s largest and most popular platform for people-powered research.

Original URL: https://www.zooniverse.org/organizations/md68135/notes-from-nature

👍 Sara Beery

Ritwik (rittyun@yahoo.com)

2021-02-17 06:45:59

working on a similar thing now.. paper under review with minor revisions.. will post here soon 🙂

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2021-02-16 19:27:43

Our use cases are limited for our intelligent agent:

Machine translation from multiple languages to English then

Binary classifier: do the tags+title+description (standardized to English) describe a video of a whale shark in the wild or something else (non-data)

Date prediction with CoreNLP

👍 Sara Beery, Ankita Shukla

Sara Beery (sara.beery@gmail.com)

2021-02-18 19:58:11

Anyone know of any public datasets collected in the wild (not in a lab) that have behavior labels per-animal? Asking for a friend 🙂

Sara Beery (sara.beery@gmail.com)

2021-02-18 20:05:28

Snapshot serengeti has kinda messy crowdsourced sequence-level behavior labels, but they aren't matched to individuals

Sara Beery (sara.beery@gmail.com)

2021-02-18 20:06:59

I've seen a work recently on tracking and/or estimating pose of individual chimps (https://gdude.de/densepose-evolution/, https://advances.sciencemag.org/content/5/9/eaaw0736?utm_campaign=The%20Batch&utm_source[…]8wWvIUWHI963s-JBWbyXpTKSgk7bmicYo17It7L7cubxMB7g0nXbin1fw)

gdude.de

Transferring Dense Pose to Proximal Animal Classes

Transferring Dense Pose to Proximal Animal Classes.

Original URL: https://gdude.de/densepose-evolution/

Science Advances

Chimpanzee face recognition from videos in the wild using deep learning

Original URL: https://advances.sciencemag.org/content/5/9/eaaw0736?utm_campaign=The%20Batch&utm_source=hs_email&utm_medium=email&_hsenc=p2ANqtz-8CwLclbKbf40u8wWvIUWHI963s-JBWbyXpTKSgk7bmicYo17It7L7cubxMB7g0nXbin1fw

Sara Beery (sara.beery@gmail.com)

2021-02-18 20:07:45

But I'm not sure if those datasets are 1) public, 2) have behavior labels

Ben Weinstein (benweinstein2010@gmail.com)

2021-02-18 20:14:00

https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0158748

journals.plos.org

Quick, Accurate, Smart: 3D Computer Vision Technology Helps Assessing Confined Animals’ Behaviour

Mankind directly controls the environment and lifestyles of several domestic species for purposes ranging from production and research to conservation and companionship. These environments and lifestyles may not offer these animals the best quality of life. Behaviour is a direct reflection of how the animal is coping with its environment. Behavioural indicators are thus among the preferred parameters to assess welfare. However, behavioural recording (usually from video) can be very time consuming and the accuracy and reliability of the output rely on the experience and background of the observers. The outburst of new video technology and computer image processing gives the basis for promising solutions. In this pilot study, we present a new prototype software able to automatically infer the behaviour of dogs housed in kennels from 3D visual data and through structured machine learning frameworks. Depth information acquired through 3D features, body part detection and training are the key elements that allow the machine to recognise postures, trajectories inside the kennel and patterns of movement that can be later labelled at convenience. The main innovation of the software is its ability to automatically cluster frequently observed temporal patterns of movement without any pre-set ethogram. Conversely, when common patterns are defined through training, a deviation from normal behaviour in time or between individuals could be assessed. The software accuracy in correctly detecting the dogs’ behaviour was checked through a validation process. An automatic behaviour recognition system, independent from human subjectivity, could add scientific knowledge on animals’ quality of life in confinement as well as saving time and resources. This 3D framework was designed to be invariant to the dog’s shape and size and could be extended to farm, laboratory and zoo quadrupeds in artificial housing. The computer vision technique applied to this software is innovative in non-human animal behaviour science. Further improvements and validation are needed, and future applications and limitations are discussed.

Original URL: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0158748

🙌 Sara Beery

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2021-02-23 13:08:58

Hey all. We have struggled at Wild Me with finding small batch data annotation outsourcing options, either through volunteers, paid local contractors, or independent firms. Local contractors and volunteers have been problematic, rarely completing one project or lasting more than two. Given that our average batch size is 2000-2500 photos needing annotations, we're not at the volume or price point that a big vendor (iMerit, Samasource, etc.) is interested. We recently started working with a past collaborator (Dr. Richard Lamprey) and his Uganda-based aerial survey analytics team (https://www.wildspace-image-analytics.com/). We have completed three projects with them and are wrapping up a fourth. Their speed, quality and attention to detail has been impressive (the team is composed of experienced aerial survey reviewers), and the price point is good too. I highly recommend them if you are looking for an outsourcing option.

Full disclosure: I am a past collaborator with Dr. Richard Lamprey. I have no financial interest in Wildspace.

Wildspace-Image-Analytics

Image, Aerial Photograph, Wildlife, Machine Learning, Annotation

Image interpretation and analysis for wildlife populations and land use assessment, machine learning, annotation

Original URL: https://www.wildspace-image-analytics.com/

👍 Vienna Saccomanno, Sara Beery, Jon Van Oast, Carly Batist, Ben Weinstein, Ed Miller, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Armin Bazarjani, David

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-02-24 09:20:34

Hi all! Just wanted to give a heads-up that WILDLABS is running their annual week-long #tech4wildlife photo challenge starting today!! Check it out on Twitter: https://twitter.com/WILDLABSNET/status/1364521244620644352?s=20

❤️ Talia Speaker, Sara Beery, Rob Sinclair, Oc

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-02-24 09:21:51

More info here as well - https://www.wildlabs.net/tech4wildlife

WILDLABS.NET

Join the sixth annual #Tech4Wildlife Photo Challenge!

How are you using tech for wildlife conservation? A picture says a thousand words, so take a photo this World Wildlife Day and share how you're using #Tech4Wildlife!

Original URL: https://www.wildlabs.net/tech4wildlife

Petar Gyurov (pgyurov93@gmail.com)

2021-02-26 05:44:32

Has anyone here made production-ready Tensorflow apps that utilise the GPU? I want to simplify the process of enabling GPU support for non-tech users; I was thinking I could create a Docker image that has all the CUDA libraries, drivers, etc. installed but even that looks like it might not be straight-forward for the end user.

Frederic (frederic@apic.ai)

2021-02-26 07:54:46

I build production ready TF apps, but not for the deployment at "consumers/non tech people". I have control about the devices that TF GPU and CPU containers get deployed to or can define requirements about their environment.

To my knowledge the gpu driver, cuda, cudnn version hassle is still present in nvidia-docker. Specially when using new GPUs like the 3060, ... compatibility with every prebuild gpu TF containers is a real pain.

But one of the main benefits is, that you have way more control about the software stack with nvidia-docker, but the pain of supporting older and newer gpus is still present.

Further more I am not sure about the EULA, ... when distributing images publicly.

Side note:
If you want to use gpus within docker-compose ping be for workarounds, since its not officially supported yet ;)

Frederic (frederic@apic.ai)

2021-02-26 07:58:54

If you decide to go with docker images, i could share some of our internal guides privately, that might help you to set everything up and have some orientations for the README.md.

Petar Gyurov (pgyurov93@gmail.com)

2021-02-26 08:13:41

@Frederic Thanks -- I'm glad you've shared this knowledge with me before I started any Docker development. It looks like there may not be much to gain in my use case.

I've thought about just writing a script that downloads and installs all the requirements but NVIDIA don't make it easy (you need an account to download the cudnn libs 😠 ); and I doubt I can just bundle them into my installer without facing licensing issues. Will ping you if I decided to explore things further, thanks for the offer 👍

Ben Weinstein (benweinstein2010@gmail.com)

2021-02-26 10:00:29

@Petar Gyurov what do you have in mind? I took a meeting with an engineer not so long ago on, on device camera traps, may be relevant here.

Petar Gyurov (pgyurov93@gmail.com)

2021-02-26 10:12:46

@Ben WeinsteinEnabling GPU support in my project, MegaDetector-GUI, requires end users to download all the CUDA drivers, tookits and library patches. I've written a basic How-To guide but I was kind of hoping for a more elegant "automagical" solution that non-tech people can easily do.

Benjamin Hoffman (benjaminsshoffman@gmail.com)

2021-03-01 11:06:19

Hi everyone! I’ve been here for a while but I haven’t introduced myself. I’m a recent math PhD (in differential geometry). Now I’m pivoting to work in the conservation + ai realm. I’ve been collaborating with (and learning from) @gvanhorn at the Cornell lab of ornithology, working on some bird sound id problems. I’m also interested in remote sensing and its applications to conservation problems. Thanks to @Sara Beery for the invite! 🦉

👋 Sara Beery, Ben Weinstein, Oisin Mac Aodha, Declan, Oc, aruna, Omiros Pantazis, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Carly Batist, Daniel Grzenda, Vienna Saccomanno, Ed Miller, Björn Lütjens, Armin Bazarjani, Gyri Reiersen, Lukas Liebel

🐦 gvanhorn, Oc, Riccardo de Lutio, Hannah Yin

👋:skin_tone_4: Ixchel Meza

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-03 10:32:00

What's the latest in bird sound ID? What's new? Are we getting better? What's the main obstacle, annotations? geographic generalization? data quality?

Benjamin Hoffman (benjaminsshoffman@gmail.com)

2021-03-03 11:22:02

My perception (keeping in mind I’ve only been working on this for 6 months) is that there is a lot of room for improvement in getting more annotated data. There exists a ton of high quality recordings (eg at Macauley library), but it’s a bit of an expert task to box and correctly label each bird vocalization.

Benjamin Hoffman (benjaminsshoffman@gmail.com)

2021-03-03 11:22:28

Species recognition in soundscapes remains hard, because there tends to be many overlapping sounds, birds at varying distances from the microphone, and lots of background noise.

Benjamin Hoffman (benjaminsshoffman@gmail.com)

2021-03-03 11:23:02

I think most existing approaches to bird sound recognition use tweaks of off-the-shelf vision models (eg ResNets), and it would be exciting to construct some models which are more custom-built for the domain.

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-03 11:24:41

interesting. so its in the localization of the annotations within samples. I've heard this before. Thanks for your thoughts.

Benjamin Hoffman (benjaminsshoffman@gmail.com)

2021-03-03 11:25:50

no problem, i appreciate the interest!

Sara Beery (sara.beery@gmail.com)

2021-03-03 11:26:41

It's always interesting to see how many common threads there are across domains!!

Sara Beery (sara.beery@gmail.com)

2021-03-04 13:00:13

New python API for GBIF data, focused on pulling data for ML training!!! If you're interested in testing or contributing you can contact gbif-dl@inria.fr

"We released a first version of GBIF-DL (GitHub, pypi) a package that makes it simpler to obtain training images from the GBIF database to be used for training machine learning classification tasks. It wraps the GBIF API and supports directly querying the api to obtain and download a list of urls efficiently (based on asyncio) and with many options (balancing, bounding, filtering, etc.). We are looking for some potential users who would be willing to test it, provide feedbacks and/or maybe contribute to it."

plantnet/gbif-dl

GBIF classification dataloaders

Website

<a href="https://plantnet.github.io/gbif-dl/">https://plantnet.github.io/gbif-dl/</a>

Stars

1

PyPI

gbif-dl

Machine learning data loaders for GBIF

Original URL: https://pypi.org/project/gbif-dl/

gbif.org

GBIF REST API

Global Biodiversity Information Facility. Free and Open Access to Biodiversity Data.

Original URL: https://www.gbif.org/developer/summary

🐍 Stefan Schneider, Ben Weinstein, Declan, Megan Cromp, David

👀 Isaac Griswold-Steiner, Vienna Saccomanno, David

🌴 Riccardo de Lutio

😎 Jon Van Oast

Jon Van Oast (jon@wildme.org)

2021-03-04 15:51:10

this is so cool -- thanks for sharing! being a bit of a gbif fanboy myself, this has me pretty excited.

Ed Miller (ed@hypraptive.com)

2021-03-04 16:06:38

This may come in handy as we start to expand BearID to more species! Is there a way to filter for images that actually show the species in question rather than signs of the species (tracks, scat, etc.)?

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-04 22:16:03

Does anyone in our community have experience mounting cameras to the bottom of manned aircraft for wildlife surveys? My lab at the University of Florida performs manned aircraft surveys of wading bird populations to support everglades restoration by the US Army Corps, as well as endangered species management by Everglades National Park. The project has been going on for a couple decades. We are working to modernize the image processing and analysis tools. We've got a nice object detection model going from unmanned vehicles for small areas
and we are working to replace manned observation from our 1957 cessna 182 with an image capture solution. Ideally we would process the data during flight to throw away the vast majority of images that do not have birds present to keep storage more reasonable (?). We have an FAA approved housing to attach to the struts, so i'm looking for people with experience wiring the camera to a GPU in the plane. We have a Nikon D850 (~45 megapixels). Nadir or oblique angle? Wide angle lens? We fly at about 1000ft and 100 knots, but that was based on a human observation system. We survey a huge area (1000+ sq miles). I'm spoiled by all the innovation in an unmanned vehicles that abstracts away alot of these challenges. All thoughts welcome.

🐦 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Daniel Grzenda, Sara Beery, Björn Lütjens

Howard L Frederick (simbamangu@gmail.com)

2021-03-04 23:04:12

@Ben Weinstein yes - have been doing this for a few years with nadir and oblique images from Cessna 182/206.

Howard L Frederick (simbamangu@gmail.com)

2021-03-04 23:05:13

One of the systems is documented here: https://github.com/TZCRC/Lanner-CamPod

TZCRC/Lanner-CamPod

Build information and development for the Lanner camera pod.

Stars

1

Language

Python

😎 Jon Van Oast, Ben Weinstein, Sara Beery

Howard L Frederick (simbamangu@gmail.com)

2021-03-04 23:09:09

You mention “wiring the camera to a GPU” meaning you’d like to do in-flight processing to discard images? I’d suggest that the cost of storage is very cheap but the cost of missing data by poorly-filtered in-flight processing is high, so think about post-processing. We take images at 2 second intervals - Nikon raw @ 30+MB per image over 4 hours = 200 GB per flight; there are 256 and 512GB cards, but the biggest limitation is speed of the SD card (if it isn’t fast enough you get gaps in coverage while the camera catches up with writes).

Have Camera Will Travel - Reviews & Tips

Fastest MicroSD Card Speed Tests

If you're using a microSD card for 4K or 5K video or other demanding use, speed matters. These are the results from my independent, real-world speed tests.

Original URL: https://havecamerawilltravel.com/photographer/fastest-microsd-cards/

👍 Ben Weinstein, Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-04 23:47:51

Is someone physically changing out the SD cards as they fill during flight? My vision for the inflight processing is just have a pilot flying. That way we have a TB drive in the plane, but I guess we can just skip the detection mid-flight. Especially due to COVID, we can only have one person in the plane.

Howard L Frederick (simbamangu@gmail.com)

2021-03-05 00:30:20

No, it's a single SD card per flight. We were fine with 256GB until now but upgrading to 512 soon for longer flights. You planning on raw files?

Howard L Frederick (simbamangu@gmail.com)

2021-03-05 00:33:43

The reliability of a Nikon to get a photo on the sd card with every triggering is rock-solid. There are probably ways to record to an external drive or computer but don't lose the onboard reliability.

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-05 11:23:55

Great. I'll take this to the team. Thanks.

Pietro Perona (perona@caltech.edu)

2021-03-07 13:25:12

Hi Ben - are you sure that you need to store RAW files? If you stored high quality JPG you may be able to reduce the total storage just about to where youre fast SD cards will manage for the whole flight and then process once you are on the ground.

Pietro Perona (perona@caltech.edu)

2021-03-07 13:26:30

@Ben Weinstein Would you be willing to share your image dataset with the students in our class? As I understand it you need a good algorithm for detecting and counting the birds, where each bird is assigned a latitude and longitude. You probably want to also estimate the species. Correct?

Pietro Perona (perona@caltech.edu)

2021-03-07 13:27:14

@Ben Weinstein If you are inclined to share (part of) your dataset with us I will be in touch to work out the details.

Vincent Miele CNRS (vincent.miele@univ-lyon1.fr)

2021-03-05 08:29:03

Hi all, I am Vincent Miele from CNRS France. We are leading a nice collaboration in France to evaluate different strategies to deal with camera traps images from different places in the country. The idea is to transfer most of our future knownledge to the community and to help people in the field to use machine learning stuff on their own. Thanxs for all for being so inspiring. And thanks to @Sara Beery for the invitation.

👋 Daniel Grzenda, Sara Beery, Stefan Schneider, Ben Weinstein, Olivier Gimenez, Ed Miller, Declan

Ixchel Meza (ixchel.meza.ch@gmail.com)

2021-03-05 11:40:17

Hi, Vincent. We are developing a platform for capture, management and administration of biodiversity collections at our group in CONABIO and we are using/testing it internally. Our intention is that any research group could use it, but for now it is still in developing phase 😔 maybe later! 👋

🙌 Olivier Gimenez

Pietro Perona (perona@caltech.edu)

2021-03-07 13:34:32

Bonjour Vincent! Do you have camera trap datasets and image analysis challenges that you are willing to share with our class at Caltech?

Pietro Perona (perona@caltech.edu)

2021-03-07 13:36:53

@Vincent Miele CNRS We (@Sara Beery, @Elijah Cole (Deactivated) and Neehar Kondapaneni) are teaching a computer vision class at Caltech and would love to challenge our students with projects that help ecologists and field biologists.

Vincent Miele CNRS (vincent.miele@univ-lyon1.fr)

2021-03-08 03:52:17

Bonjour Pietro, we are discussing with different french partners to get image data, but these are their own private data... Maybe they will share their data at some point in the mid-term, but this is not planned at the moment. In any case, this would be a good idea!

Elizabeth Madin (emadin@hawaii.edu)

2021-03-05 20:51:35

Hello everyone, I recently heard about this space and have to thank @Heather Lynch for pointing me to it. I'm an Assistant Professor focused on marine conservation ecology at the Hawaii Institute of Marine Biology (part of the University of Hawaii). My lab's work focuses on understanding dynamics of ecosystems, primarily (but not exclusively) through landscapes of risk in marine systems. In particular, we're interested in how humans are changing landscapes of risk, and ecosystems, on local to global scales. Of relevance to this group is our current NSF project looking (in part) at developing an ML algorithm to classify and measure features of coral reefs called "reef halos" (aka "grazing halos"). We also do quite a bit of underwear 'camera trap' work looking at fish and invert behavior. My lab's website is https://www.oceansphere.org. Looking forward to learning more about what others are doing in this space!

👋 Sara Beery, Declan, gvanhorn, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Oc, aruna

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-05 21:20:56

what does underwater camera trapping look like? Can you post a photo? My old tool for video processing had some object detection that seem to get alot of downloads from the underwater ecology community. We are redesigning those tools, but don't have a well annotated underwater dataset.

Howard L Frederick (simbamangu@gmail.com)

2021-03-06 08:14:36

@Elizabeth Madin welcome … and do you have any writeups on underwater camera traps? Am trying to figure out how to make some for cichlid monitoring in freshwater and am struggling!

Oc (boulaisoceane@gmail.com)

2021-03-06 14:54:04

thirding @Ben Weinstein and @Howard L Frederick here - I would love to see you underwater camera set up! I’ve been building species ID models with the collected habcam data along the NE coast of the Gulf of Mexico that look like this and I’d love to know if folks have come up with resilient cages for their cameras that don’t have bars…

Oc (boulaisoceane@gmail.com)

2021-03-06 14:57:45

This is the preliminary results of species ID detection (trained with mmlab’s toolbox & resnet50 backbone), and I bet consistent accuracy would increase without these bars…

👍 Sara Beery, aruna

aruna (arunas@mit.edu)

2021-03-07 13:52:56

That looks so cool, @Oc! Thanks for sharing. ♥️

❤️ Oc

Howard L Frederick (simbamangu@gmail.com)

2021-03-08 01:11:50

@Oc that is a very solid looking camera, do you have any public build guides for how you did it?

Elizabeth Madin (emadin@hawaii.edu)

2021-03-08 14:24:24

@Ben Weinstein and @Howard L Frederick: this paper describes in some detail what we did recently with our underwater (coral reef) camera trap arrays. I call them camera traps - it's actually continuous video that we manually annotated (ugh!). We have a fairly large dataset (~3000 annotated individuals, mostly fish + some inverts) from Australia's Great Barrier Reef that could be put towards developing an algorithm if people are looking for a challenge like that! We're collecting and annotating more video now from Hawaii (some species overlap with Australia, but not complete) and would absolutely love to apply an algorithm like what you've posted, @Oc! I'll also post a photo from our earlier rounds (using now-old GoPros) - we don't bother with cages because we've never had a problem with cameras being dislodged, attacked, etc., We're working in fairly wave-limited coral reef lagoon environments.

Elizabeth Madin (emadin@hawaii.edu)

2021-03-08 14:26:49

Thread Reply:

🙌 Howard L Frederick, Oc

Elizabeth Madin (emadin@hawaii.edu)

2021-03-08 14:31:47

Thread Reply:

❤️ aruna, Howard L Frederick, Oc

Oc (boulaisoceane@gmail.com)

2021-03-08 18:25:09

That little camera is so cute! Is it mounted directly on the coral/rock?

Elizabeth Madin (emadin@hawaii.edu)

2021-03-08 19:29:16

@Oc: It's just temporarily cable-tied to dead coral. Easy on, easy off with no damage to anything. 🙂

❤️ Oc

Howard L Frederick (simbamangu@gmail.com)

2021-03-09 01:17:58

@Elizabeth Madin interesting paper, it’s amazing to be able to quantify feeding behaviour like that. I notice 7 cameras malfunctioned (out of 30?) - what were the problems there?

Elizabeth Madin (emadin@hawaii.edu)

2021-03-09 11:39:15

@Howard L Frederick: Ah, yes...we had signifiant camera failure because we had to modify the cameras slightly by using a larger waterproof housing case to accommodate the extended-life batteries we needed to get us through dusk and into the night. (We had to set out the cameras and get the boat back to the research station prior to dark.) A number of the modified housings failed overnight, despite having passed our (shorter-term) tests at depth prior to deployment. It was indeed pretty frustrating, but thankfully we still got enough data from the remaining cameras to get to the answer!

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-14 17:39:57

@Oc, @Elizabeth Madin, from ecolog if you didn't see it

Hello, my name is Austin Greene and I am a PhD student at the University of Hawaii at Manoa studying the ecology of coral reefs. I recently developed a low-cost camera system (KiloCam) that I hope will make habitat monitoring more accessible, and I am looking for other researchers to help me field test it. KiloCam is small, about 40x25x25mm, and fits inside of cheap GoPro housings to be made waterproof. The instrument is capable of taking 2 MP photos at user-specified intervals, can run off of many power sources, and is highly efficient for long deployment times. For example, on a set of two AA batteries KiloCam should operate for over a week at one photo per minute, or nearly a year with one photo per hour. While the camera isn't particularly high-resolution and is best suited for environments with ample ambient light, it is adequate for habitat monitoring. My hope is that the low cost of building KiloCam (~$30 USD) will make mass deployments possible, or allow improved habitat monitoring in underserved areas.

🙌 Oc

Pietro Perona (perona@caltech.edu)

2021-03-07 00:49:58

Dear all, @Sara Beery, @Elijah Cole (Deactivated) , @neehar Kondapaneni and I will be teaching a computer Vision class at caltech with focus on quantitative ecology. The students will be carrying out projects which, we hope, will push the envelope of what is possible. Please let us know about your datasets and image analysis needs!

❤️ Sara Beery, Jason Holmberg (Wild Me), Omiros Pantazis, aruna, Carly Batist, Elijah Cole (Deactivated), Subhransu Maji, Oc, David, Suzanne Stathatos, Gyri Reiersen, Armin Bazarjani

aruna (arunas@mit.edu)

2021-03-07 10:48:22

Hello! We have an earth observation dataset at MIT, of satellite images from Sentinel. It's about 100k pairs of images from the arctic in the winter and the summer. We are currently fixing how diverse the dataset is, so it could get smaller. But, please let us know if you want to use it for the class.

Elijah Cole (Deactivated) (ecole@caltech.edu)

2021-03-07 12:22:34

@aruna Sounds like it could be a good fit! What’s the task / label set like?

aruna (arunas@mit.edu)

2021-03-07 12:23:34

We have winter tiles, (pre-melt), summer tiles (post-melt), and segmentation masks that identify ice and non-ice regions.

Sara Beery (sara.beery@gmail.com)

2021-03-07 13:38:24

@aruna that sounds like a great dataset. Is it public? Or if not, would you be able to share it?

aruna (arunas@mit.edu)

2021-03-07 13:51:35

Yes, I can share it with you. We are currently working on paring it down a little to have more diversity in the tiles, but I can send you the details once it is published.

aruna (arunas@mit.edu)

2021-03-07 13:51:54

Can you give me a date I should aim for to get the dataset to you?

Sara Beery (sara.beery@gmail.com)

2021-03-07 13:52:20

The class will begin at the end of the month!

aruna (arunas@mit.edu)

2021-03-07 13:53:25

Sounds great. Will send it to you folks soon.

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-07 16:35:21

Just sharing a few things that might be fun starting with the same data. I can expand on any of these ideas.

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-07 16:35:25

Object detection from multiple sensors with uncertainty in spatial alignment. https://github.com/weecology/NeonTreeEvaluation

❤️ Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-07 16:36:15

Fusing 2D sensor and 3D point clouds in object detection. https://github.com/weecology/NeonTreeEvaluation

❤️ Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-07 16:38:08

Geographic generalization in fine-grained species classification https://github.com/Weecology/DeepTreeAttention

https://idtrees.org/competition/

Implementation of Hang et al. 2020 "Hyperspectral Image Classification with Attention Aided CNNs" for tree species prediction

Stars

26

Language

Python

❤️ Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-07 16:39:23

3D CNN versus stacked 2D CNN versus apriori feature engineering for hyperspectral deep learning (369 bands). https://github.com/Weecology/DeepTreeAttention

❤️ Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-07 16:40:33

Long tailed species classification and unknown class identification. https://github.com/Weecology/DeepTreeAttention

❤️ Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-07 16:42:54

Miniaturization of object detection networks for in-flight screening of remote sensing imagery during wildlife surveys http://tree.westus.cloudapp.azure.com/everglades/

❤️ Sara Beery

Sara Beery (sara.beery@gmail.com)

2021-03-07 16:45:21

Hey @Ben Weinstein if we get a student team or two interested in these would you be able to help mentor them?

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-07 16:46:02

sure, I can't promise a ton of time, but alittle

👍 Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-07 16:47:33

Then there is the hummingbird data. Key frame detection and tracking in low-frame rate videos. https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/2041-210X.13011

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-07 16:49:15

Some of these are more ready for students then the others. #6 and #2 are pretty plug and play.

❤️ Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-07 16:51:12

also we ran a competition last year that covers the general (easier) tree species classification and bounding box detection. The docs and data are all there.

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-07 16:51:14

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-07 16:51:45

https://zenodo.org/record/3700197#.YEVK7pNKiHE

Pietro Perona (perona@caltech.edu)

2021-03-07 20:33:47

@Ben Weinstein Ben - thank you very much for all the suggestions!!

Caleb Robinson (calebrob6@gmail.com)

2021-03-08 15:36:43

We are currently running a competition on high-resolution land cover change detection from weak low-resolution labels (https://www.grss-ieee.org/community/technical-committees/2021-ieee-grss-data-fusion-contest-track-msd/). The competition will be over next month (it is currently in the closed test phase), but the evaluation server/data will remain up.

👍 Sara Beery, Pietro Perona

Caleb Robinson (calebrob6@gmail.com)

2021-03-08 15:38:01

There is an interesting super-resolution component from the CV side (and of course identifying land cover change over time has strong ecological implications!)

Pietro Perona (perona@caltech.edu)

2021-03-09 12:35:49

Many thanks to @Caleb Robinson @Ben Weinstein @aruna

Armin Bazarjani (bazarjan@usc.edu)

2021-03-09 18:15:36

It would be great if you could also record and/or post the lectures somewhere! I would love to watch them, and I'm sure others would too.

➕ aruna, Sara Beery, Ștefan Istrate

Sara Beery (sara.beery@gmail.com)

2021-03-09 18:16:36

We're planning on hosting it all publicly 🙂

🙌 Armin Bazarjani, aruna

Subhransu Maji (smaji@cs.umass.edu)

2021-03-10 13:24:49

A bunch of us have been working on detecting tree swallow roosts from RADAR data. We are planning to release a standardized benchmark around this. Would be happy to share it. What’s the timeline for the course? @Dan Sheldon and I would could also do some lightweight mentoring if there’s interest on working on this. More info: https://people.cs.umass.edu/~zezhoucheng/roosts/

😍 Sara Beery

Sara Beery (sara.beery@gmail.com)

2021-03-10 13:34:46

The class will be starting end of this month, so the timeline is a bit tight, but I think that would be an amazing application to bring in and we would be incredibly happy to have you act as high-level mentors. Let us know if that timeline might be possible?

Sara Beery (sara.beery@gmail.com)

2021-03-10 13:37:09

Another option might be to bring in an UG team to help with testing and publishing the benchmark as part of their project? Different members could run different baselines and provide analysis of efficacy

Subhransu Maji (smaji@cs.umass.edu)

2021-03-10 13:39:43

Awesome! I think it’s doable --- we are already working on this and the class deadline is a good motivation 🙂

👍 Sara Beery

Pietro Perona (perona@caltech.edu)

2021-03-07 00:53:33

And if you wish to be involved in mentoring student teams please let us know.

Stefan Schneider (sschne01@uoguelph.ca)

2021-03-08 10:24:09

Hi Everyone! For those that don't know me, I'm Stefan Schneider! A post-doc at the University of Guelph, Canada focusing on AI for conservation and animal re-ID. I'm looking for a satellite imagery dataset to train a whale localization model to prevent shipping and fishing boats from coming within the detected whale proximity. Does anyone know of one of a public / upon request dataset? I've reached out to the author here. Cheers! 🐳

🐋 Sara Beery, Caleb Robinson

aruna (arunas@mit.edu)

2021-03-08 10:30:32

👋 @Stefan Schneider! Have you looked at requesting tiles from Pleiades, offered by the ESA?

aruna (arunas@mit.edu)

2021-03-08 10:31:22

It's free for research, and you can get all areas covered by sea. Please take a look at their area coverage, since I am not entirely sure if they photograph land and water bodies both. It has excellent resolution at 50 cm/px

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-08 11:21:02

@Stefan Schneider, have you met @Patrick Gray, who did this work while working for our old boss Ari Friedlaender. I can put you in touch with Ari as well. Also check with the NOAA authors who have this data. I also know of a talk by people at Vulcan Inc, interested in this as well. Being in canada, you probably talked to https://www.whaleseeker.com/about-us. It feels like someone needs to organize all these folks.

Whale Seeker

Whale Seeker — Whale Seeker

Meet our team

Original URL: https://www.whaleseeker.com/about-us

Alex Borowicz (alex.borowicz@stonybrook.edu)

2021-03-08 11:47:40

Hi @Stefan Schneider! I don't have any satellite imagery I can share - Maxar is typically protective of their licenses - but we trained our satellite whale model primarily using down-sampled aerial imagery, which is available here: https://github.com/lynch-lab/Borowicz_etal_Spacewhale

lynch-lab/Borowicz_etal_Spacewhale

Aerial-trained deep learning networks for surveying cetaceans from satellite imagery

Language

Python

Last updated

a year ago

Alex Borowicz (alex.borowicz@stonybrook.edu)

2021-03-08 11:49:07

If I remember right, Guirado et al. used Google Earth images - doable, but you lose a lot of control because of Google's pansharpening algorithm. They're also not full resolution and it's not super clear how much that matters if you want to feed in full-res imagery

Stefan Schneider (sschne01@uoguelph.ca)

2021-03-08 12:09:05

Hi Aruna. Thanks for this! I'll definitely take a look through the ESA tiles from Pleiadas

Stefan Schneider (sschne01@uoguelph.ca)

2021-03-08 12:13:36

@Ben Weinstein It would be amazing if you could put me in touch with Patrick Grey & Ari Friedlaender! Which NOAA authors are you referring to (sorry if I missed something)? Vulcan Inc sounds like a great contact. Same with this Montreal group that I'll definitely reach out to. Thanks so much!

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-08 12:14:20

@Patrick Gray should be wandering around somewhere. If doesn't jump in, i'll email him.

Stefan Schneider (sschne01@uoguelph.ca)

2021-03-08 12:18:13

@Alex Borowicz This is great! I'll definitely take a look at the repo/data and let you know if I have any questions. Cheers!

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-09 16:01:11

@Vienna Saccomanno.

Vienna Saccomanno (v.r.saccomanno@tnc.org)

2021-03-09 16:14:36

Hi all - The Nature Conservancy is scoping the potential for use of VHR satellite imagery as a tool to survey large whales to meet conservation monitoring objectives in California, so it's great to see mutual interest in this topic. @Stefan Schneider we'd love to hear more about you work in Canada and see if there are ways to collaborate.

Patrick Gray (patrick.c.gray@duke.edu)

2021-03-10 09:27:19

Hey @Stefan Schneider! I know some of your papers pretty well, good to hear from you. My advisor and some collaborators are still working on this though I've moved on mostly. They're taking a large archive we have of drone imagery of whales and downsampling to satellite imagery resolution and using that to train a model for right whales. Working with some folks from BAS and Canada's DFO. When I was doing this I was using all WorldView data and it was working alright but a huge effort to find suitable training data as you likely well know. I ended up largely just searching in common bays and breeding grounds to build up an initial dataset and then iteratively applied the model to find more. Happy to answer any specifics but again I'm a little out of date now on this work!

Stefan Schneider (sschne01@uoguelph.ca)

2021-03-10 09:36:21

@Vienna Saccomanno Great to make the connection! It's early days that are just ramping up, but I'm excited to connect with others and hopefully collaborate. Right now it's the pursuit of initial datasets to build a base line of what we can expect from an object detector.

Stefan Schneider (sschne01@uoguelph.ca)

2021-03-10 09:40:46

@Patrick Gray Thanks for reaching out! It's great to chat with someone who has tackled this problem. Down sampling from drone imagery sounds like a nice solution. Most definitely with the suitable training data. I'm more or less surveying right now trying to find any sort of public dataset. Would you happen to have a link the WorldView data you were using? Or anything that may act as a thread to pull? Cheers!

Patrick Gray (patrick.c.gray@duke.edu)

2021-03-10 09:43:33

Hey @Stefan Schneider unfortunately this was all from access through a NSF polar program grant (via the Polar Geospatial Center) and we can't share any of the data. I've also since lost access myself so can't pull any more. Maxar doesn't seem as interested in this problem as they did in their Digital Globe incarnation.

Stefan Schneider (sschne01@uoguelph.ca)

2021-03-10 11:21:49

@Patrick Gray no worries. This seems to be the way it goes for valuable datasets. I'm sure we all wish we could bring all this data together to train a universal performant model

aruna (arunas@mit.edu)

2021-03-08 16:18:35

Are there folks who have used https://github.com/knjcode/imgdupes?

knjcode/imgdupes

Finding and deleting near-duplicate images based on perceptual hash.

Stars

132

Language

Python

aruna (arunas@mit.edu)

2021-03-08 16:19:05

I am curious if there's a way to measure the similarity

aruna (arunas@mit.edu)

2021-03-08 16:19:19

For eg. delete if 60% similar but keep if 59.2% similar.

aruna (arunas@mit.edu)

2021-03-08 18:26:10

Ah, it's related to the hamming distance. 🙂 All good here.

Oc (boulaisoceane@gmail.com)

2021-03-08 18:46:41

Is there an easy way to access the FGVC7 extended abstract submissions from CVPR 2020?

Sara Beery (sara.beery@gmail.com)

2021-03-08 18:47:16

I believe most of them are on arxiv? Is there one you can't find?

Sara Beery (sara.beery@gmail.com)

2021-03-08 18:48:02

It looks like there are direct links here as well: https://sites.google.com/corp/view/fgvc7/program?authuser=0

sites.google.com

FGVC7 - Program

FGVC7 Schedule Overview

Original URL: https://sites.google.com/corp/view/fgvc7/program?authuser=0

🙌 Oc

🙏 Oisin Mac Aodha

Oc (boulaisoceane@gmail.com)

2021-03-08 18:48:50

Exactly what I was after. Thank you!

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-10 12:02:31

Are there remote sensing experts in channel? maybe @Sarra Alqahtani. For our everglades bird detection project. I'm trying to get a sense for what our image targets would look like at different image resolutions. We are considering buying new cameras for long-range flights or even invest in satellite data. I have 1cm drone imagery over known areas. My simple plan was to resample the train and test data to lower resolutions (5, 10, 20, 30, 50cm) and measure our ability to recover any signature of their presence visually and using our existing object detection models. Is there a way to make this more rigorous? Any other factors I should consider? What is the best resampling method to best approximate native collect at that resolution? Mean?

aruna (arunas@mit.edu)

2021-03-10 12:05:22

Hi Ben, I am not an expert in this area, but I am taking a computer vision class this semester, and one approach we spoke of here is to use a gaussian blur + sub-sampling. It's one of the recommended ways to downscale images, and between successive down-sampling, you lose mostly details. Plus up-sampling with Laplacian pyramids. Sorry if you already knew this, was excited about this class from today morning, and jumped in with a response. : )

👍 Ben Weinstein

Sara Beery (sara.beery@gmail.com)

2021-03-10 12:20:31

I remember @Sarra Alqahtani discussing challenges surrounding switching between actual data collected with different resolutions vs subsampling high resolution data in her mine detection work?

Sarra Alqahtani (sarra-alqahtani@utulsa.edu)

2021-03-11 17:50:38

Hi @Ben Weinstein. Sorry for the late response. What we are currently doing is to train our model on a mixed of high resolution images (from drone) and downsampled images (low resolution from satellite) then during testing we only use low resolution images. We use GANs and it is a risky technique since the model may populate unreal details but our point is to improve the accuracy of the object detection so we are not that worried about the accuracy of the resolved images (landscapes with gold mining objects). I would think your goal is the same unless you want to differentiate between birds. I hope this helps.

👍 Ben Weinstein

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-11 17:51:34

have you seen this strategy elsewhere? When you say a mix, do you mean at the same time, or sequentially?

Sarra Alqahtani (sarra-alqahtani@utulsa.edu)

2021-03-11 18:35:55

it is like 80% of the dataset is hr and the rest for the lr. I will find the papers we got this idea from. You could try the sequential and see if it will improve the results just as a form of TL

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-11 18:40:40

awesome. We have alot of use cases for cross resolution work, and never really explored the literature. Thanks.

Grace Hansen (ghansen33@gatech.edu)

2021-03-11 12:08:32

Hi everyone, my name is Grace Hansen and I am working with @Isha Palakurthy at the Georgia Institute of Technology to develop a vaccine dispensing device to prevent the spread of rabies among foxes on our campus. Our goal is to have the device be triggered by the appearance of a fox in a wooded area, so this requires image recognition without a reliable internet connection.

Would anyone have any suggestions/resources regarding an edge computing project such as this? We are also grateful for any advice just in approaching the problem. At the moment, we have a dataset of images including images of foxes, but we are also gathering camera trap images here on campus, though this is unlabeled and still limited. Also, most of the tools we are looking at involving using a CNN, but we are also trying to figure out what preprocessing may be necessary given that we want to deploy this on a remote device. We’re aiming to start off deploying a prototype at a smaller scale for now. Any advice (or further questions) is greatly appreciated!

Thank you @Ben Seleb for the invite to this excellent community!

Elijah Cole (Deactivated) (ecole@caltech.edu)

2021-03-11 12:09:50

@Sam Kelly might have a good sense for the right edge device?

❤️ Sara Beery, Grace Hansen

Sam Kelly (sam@conservationxlabs.org)

2021-03-11 12:14:20

I am personally a fan of Google Coral ecosystem - very easy to use. As with any edge device, figuring out a way to keep the power down is the key! (This is something we have been working on at CXL) I would be more than happy to jump on a call with you to help out. just DM me

Grace Hansen (ghansen33@gatech.edu)

2021-03-11 12:16:48

Thank you! That would be great, we will message

Rosho (rbam.vc@gmail.com)

2021-03-15 08:42:12

TrailGuard AI project has used an Intel’s movidius myriad 2 VPU .
Edge devices, you can also test: nvidia jetson nano.
I made a camera trap with the Google Coral devboard.

Sara Beery (sara.beery@gmail.com)

2021-03-11 12:44:08

iWildCam 2021 is LIVE!!! https://www.kaggle.com/c/iwildcam2021-fgvc8

"Camera traps enable the automatic collection of large quantities of image data. Ecologists all over the world use camera traps to monitor biodiversity and population density of animal species. In order to estimate the abundance and density of species in camera trap data, ecologists need to know not just which species were seen, but also how many of each species were seen. However, because images are taken in motion-triggered bursts to increase the likelihood of capturing the animal(s) of interest, object detection alone is not sufficient as it could lead to over or undercounting. For example, if you get 3 images taken at one frame per second and in the first image you see 3 gazelles, in the second you see 5 gazelles, and in the last you see 4 gazelles, how many total gazelles have you seen? This is more challenging than strictly detecting and categorizing species as it requires reasoning and tracking of individuals across sparse temporal samples."

Please share to your communities, and if you're on twitter help me out by sharing our twitter thread! https://twitter.com/sarameghanbeery/status/1370065404148678657

kaggle.com

iWildcam 2021 - FGVC8

Count the number of animals of each species present in a sequence of images

Original URL: https://www.kaggle.com/c/iwildcam2021-fgvc8

📸 Stefan Schneider, Declan, aruna, Oisin Mac Aodha, Henrik Cox (Sentinel), Daniel Grzenda, Omiros Pantazis, Armin Bazarjani, Gyri Reiersen, Olivier Gimenez, Rosho, Dan Morris

📷 Ixchel Meza, Armin Bazarjani, Gyri Reiersen

gvanhorn (grv22@cornell.edu)

2021-03-11 13:03:06

Similarly, the iNat Competition is back for 2021!

https://www.kaggle.com/c/inaturalist-2021/overview
https://github.com/visipedia/inat_comp/tree/master/2021

After taking a year out, we are excited to announce the latest iNaturalist Species Classification Challenge. This challenge is part of the Eighth Workshop on Fine-Grained Visual Categorization (FGVC8) at CVPR 2021.

The iNat Challenge 2021 dataset contains 10,000 species, with a training dataset of 2.7M images that have been collected and verified by multiple users from iNaturalist. There is also a more manageable "mini" dataset with 50 images per species, for a total of 500K training images. The dataset features many visually similar species, captured in a wide variety of situations, from all over the world.

We have made several modifications to the competition this year. Similar to the 2017 competition, we are releasing the species names immediately, instead of obfuscating them. Our reason for obfuscating them in 2018 and 2019 was to make it difficult for competitors to scrape the web (or iNaturalist itself) for additional images. Because we are releasing 2.7M training images and the dataset doesn't necessarily focus on the long tail problem we feel that we can release the species names without worry. This does not mean that scraping is allowed. Please do not scrape for additional data, especially from iNaturalist or GBIF. Having the species names also makes interpreting validation results easier when examining confusion matrices and accuracy statistics.

We are also releasing location and date information for each image in the form of latitude, longitude, location_uncertainty, and date values. We have retroactively added this information to the 2017 and 2018 datasets, but this year competitors are able to utilize this information when building models. We hope this motivates competitors to devise interesting solutions to this large scale problem. You will find more information on the challenge Github page.

We are looking forward to seeing the creative solutions you come up with!

Please share to your communities!
https://twitter.com/oisinmacaodha/status/1370050350779150340

kaggle.com

iNat Challenge 2021 - FGVC8

10,000 Species Recognition Challenge with iNaturalist Data - FGVC8

Original URL: https://www.kaggle.com/c/inaturalist-2021/overview

🌿 Oisin Mac Aodha, Sara Beery, Omiros Pantazis, Armin Bazarjani, Gyri Reiersen

🦌 Oisin Mac Aodha, Sara Beery, Armin Bazarjani, Gyri Reiersen

🐦 Oisin Mac Aodha, Sara Beery, Elijah Cole (Deactivated), Armin Bazarjani, Gyri Reiersen

:giraffe_face: Daniel Grzenda, Sara Beery, Armin Bazarjani, Gyri Reiersen

:zebra_face: Daniel Grzenda, Sara Beery, Armin Bazarjani, Gyri Reiersen

🎉 Sara Beery, Armin Bazarjani, Gyri Reiersen

🌎 Ixchel Meza, Armin Bazarjani, Gyri Reiersen

👍 Holger Klinck, Armin Bazarjani, Gyri Reiersen, Olivier Gimenez, Rosho

Oc (boulaisoceane@gmail.com)

2021-03-15 19:32:23

Howdy folks! Is anyone here aware of a fish (underwater) dataset for fine-grained categorization/verification?

Sara Beery (sara.beery@gmail.com)

2021-03-15 19:33:23

I just saw https://www.frontiersin.org/articles/10.3389/fmars.2021.629485/full

Frontiers

Annotated Video Footage for Automated Identification and Counting of Fish in Unconstrained Seagrass Habitats

Abstract Computer vision techniques such as deep learning have quickly become of interest to ecologists for automatically processing large volumes of video and image-based data. However, training deep learning models often require large volumes of manually annotated footage to create a robust and accurate model. The collection and annotation of these training datasets can incur high initial labour cost and may not be feasible for some research projects. The accessibility of publicly available datasets that are pre-annotated for easy implementation is imperative for continued research and understanding of computer vision technology as a viable method to automate the processing of visual data. In this report, we provide a dataset containing ~ 9,000 annotated fish in unconstrained conditions in a key coastal habitat - seagrass meadows - collected via remote underwater video. These images include object instance annotations which consist of a corresponding image, label, bounding box and segmentation mask. These data can be used for training several different computer vision models and for investigating the effects of pre- or post-processing steps to improve model performance when predicting data in aquatic habitats. The purpose of this manuscript, in conjunction with the annotated dataset, is to advance the use of computer vision techniques and further the growth in labelled fish datasets publicly available. Keywords: annotated dataset, automation, machine learning, seagrass I...

Original URL: https://www.frontiersin.org/articles/10.3389/fmars.2021.629485/full

🐠 Oc

Sara Beery (sara.beery@gmail.com)

2021-03-15 19:33:49

Not sure how many species, so it may not be very fine-grained

Sara Beery (sara.beery@gmail.com)

2021-03-15 19:35:24

And there was a fish task in LifeClef a while ago https://www.imageclef.org/lifeclef/2015/fish

Elizabeth Madin (emadin@hawaii.edu)

2021-03-15 20:52:10

We may have a relevant dataset...will send details when I reply to your email, but feel free to remind me if I forget.

🙌 Oc

Dan Morris (agentmorris@gmail.com)

2021-04-01 12:29:17

Depending on exactly what you're looking for, here are a few more:

https://github.com/Microsoft/Project_Natick_Analysis/releases/tag/annotated_data

https://www.fishnet.ai/

http://www.inf-cv.uni-jena.de/fine_grained_recognition.html#datasets
(link is called "Croatian Fish Dataset")

...and, yes, of course there's a dataset called "DeepFish":

https://alzayats.github.io/DeepFish/

...in fact it's a pretty amazing dataset!

fishnet.ai

Forest Damages - Larch Casebearer - LILA BC

AI Training Data for Fisheries More than a thousand commercial fishing boats carry electronic monitoring systems, which use on-board cameras to track catch and provide accountability in the global seafood market. These systems produce large volumes of video data that are screened by trained human

Original URL: https://www.fishnet.ai/

inf-cv.uni-jena.de

Fine-grained Recognition

The activities of the Computer Vision Group concern with teaching and research.

Original URL: http://www.inf-cv.uni-jena.de/fine_grained_recognition.html#datasets

alzayats.github.io

DeepFish

DeepFish Dataset

Original URL: https://alzayats.github.io/DeepFish/

Dan Morris (agentmorris@gmail.com)

2021-03-22 14:44:08

New data set on LILA containing ~100k bounding boxes on trees in drone images, including classification of invasive moth damage on a subset of those bounding boxes.

http://lila.science/datasets/forest-damages-larch-casebearer/

Courtesy of the Swedish Forest Agency.

LILA BC

Overview The larch casebearer, Coleophora laricella, is a moth that mainly attacks larch trees and has caused significant damage in larch stands in Västergötland, Sweden. The Swedish Forest Agency, supported by Microsoft’s AI for Earth program, started a project that utilizes artificial intelligence for identifying, inventorying, mapping, and monitoring forest areas affected by the larch...

Written by

lilawp

Est. reading time

3 minutes

Original URL: http://lila.science/datasets/forest-damages-larch-casebearer/

👍 gvanhorn, Oisin Mac Aodha, Srishti, Vienna Saccomanno, Benjamin Kellenberger, Riccardo de Lutio, Mitch Fennell, David, Halil Radogoshi, Chris Yeh, Sara Beery, Gyri Reiersen

👀 David

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-22 15:24:55

Ooo! Me! I will download and train on this soon

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-23 11:49:08

This is what a sample image looks like. If I get a moment I will generate a single csv file so others don't need to parse the xml.

❤️ Halil Radogoshi, Gyri Reiersen

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2021-03-22 18:08:27

Consider adding EAAMO to your annual rotation of conference deadlines! Your work on AI for conservation (both technical and applied) will be very welcome.

Stemming from the Mechanism Design for Social Good (MD4SG) initiative, the inaugural ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization (EAAMO’21) aims to highlight work where techniques from algorithms, optimization, and mechanism design, along with insights from the social sciences and humanistic studies, can help improve equity and access to opportunity for historically disadvantaged and underserved communities.

Website: https://eaamo.org/
Submission date: June 3
Conference: October 5–9

eaamo.org

EAAMO'21

ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization

Original URL: http://eaamo.org/

eaamo.org

EAAMO'21

ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization

Original URL: https://eaamo.org/

👍 Caleb Robinson, Oc, David, Chris Yeh, Sara Beery

👀 Caleb Robinson

Halil Radogoshi (halil.radogoshi@skogsstyrelsen.se)

2021-03-23 02:44:49

Great to meet all of you on this channel. My name is Halil Radogoshi and I am a Senior Advisor on knowledge management and AI at the Swedish Forest Agency. We have started several AI projects at the agency and the projects mostly deal with forest damages. Here is a link to a video https://www.youtube.com/watch?v=WcVzys6ECys from one of our projects.
Looking forward to exchange experience on this channel. Thank you @Dan Morris for inviting me here.

YouTube

Skogsstyrelsen (https://www.youtube.com/user/Skogsstyrelsen)

Lärksäckmalar och AI - english subtitles

Original URL: https://www.youtube.com/watch?v=WcVzys6ECys

🌳 Omiros Pantazis, Rosho, Sara Beery, Gyri Reiersen

👏 Srishti, Gyri Reiersen

👋 Petar Gyurov, Gyri Reiersen

Oisin Mac Aodha (macaodha@caltech.edu)

2021-03-29 09:05:08

Hi everyone. Just a reminder that the deadline for papers for the workshop on Fine-Grained Visual Categorization is this Friday. You can find more info on our website: https://sites.google.com/view/fgvc8

sites.google.com

FGVC8

Announcements The CMT site for paper submissions is now live. We have launched a set of competitions as part of the workshop. Our paper submission deadline is April 2nd, additional information can be found here. A list of our workshop speakers is now online. FGVC8 will be held in conjunction

Original URL: https://sites.google.com/view/fgvc8

:bearid: Omiros Pantazis, Subhransu Maji, Sara Beery, Oc

Sara Beery (sara.beery@gmail.com)

2021-03-31 14:42:33

Our CVPR 2021 paper just went up on arxiv! We investigate supervised and self-supervised representation learning for efficient adaptation to a diverse set of real-world Conservation/Ecology tasks, such as detecting specific animal behaviors, age, or health issues. The paper was led by @gvanhorn, along with @Elijah Cole (Deactivated), myself, Kimberly Wilber, Serge Belongie, and @Oisin Mac Aodha. Check it out!

https://twitter.com/oisinmacaodha/status/1377286648489201668

👍 Oisin Mac Aodha, Holger Klinck, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Rosho, gvanhorn, Stefan Schneider, Hemal Naik, Riccardo de Lutio, Dan Morris

🎉 Vienna Saccomanno, Omiros Pantazis, Stefan Schneider, Mitch Fennell, Talia Speaker, Catherine, Oc, Hannah Yin

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-31 16:28:30

so good.

😍 Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-31 16:30:20

I can't quite find a list of tasks, was 'is this hummingbird visiting a flower' an example, or an actual task?

Ben Weinstein (benweinstein2010@gmail.com)

2021-03-31 16:30:32

I have a few dozen TB of video that throw at that question.

Sara Beery (sara.beery@gmail.com)

2021-03-31 16:30:44

An actual task.

👍 Ben Weinstein

Hemal Naik (hnaik@ab.mpg.de)

2021-04-01 10:38:27

Awesome work congratulations

Holger Klinck (hk829@cornell.edu)

2021-04-01 11:18:56

BirdCLEF 2021 competition is live: https://www.kaggle.com/c/birdclef-2021/overview

🐦 Oisin Mac Aodha, Riccardo de Lutio, Declan, Daniel Grzenda

Dan Morris (agentmorris@gmail.com)

2021-04-01 12:20:42

New data set on LILA; ~50k bounding boxes on bees and pollen:

http://lila.science/datasets/boxes-on-bees-and-pollen

...courtesy of the BeeLivingSensor project.

LILA BC

Boxes on Bees and Pollen - LILA BC

Overview The goal of the BeeLivingSensor project is to non-invasively track honey bees at hive entrances, and to track the type and volume of pollen they bring into the hive. By analyzing the color of the pollen and aggregating it with other data, the project aims to to determine the plant biodiversity around the beehive....

Written by

lilawp

Est. reading time

1 minute

Original URL: http://lila.science/datasets/boxes-on-bees-and-pollen

😍 Sara Beery, aruna, Rosho, Oc

🐝 Declan, Riccardo de Lutio, Omiros Pantazis, Carly Batist, aruna, Oc

Riccardo de Lutio (riccardo.delutio@geod.baug.ethz.ch)

2021-04-01 12:32:52

Hi everyone, the Herbarium Challenge 2021 is currently running! 🌸🌿
Still two months to go before the final submission deadline.
https://www.kaggle.com/c/herbarium-2021-fgvc8/overview

kaggle.com

Herbarium 2021 - Half-Earth Challenge - FGVC8

Identify plant species of the Americas, Oceania and the Pacific from herbarium specimens

Original URL: https://www.kaggle.com/c/herbarium-2021-fgvc8/overview

🌿 Sara Beery, Omiros Pantazis, Declan, Yumna

👍 Rosho

Sara Beery (sara.beery@gmail.com)

2021-04-06 13:18:12

Just came across this cool list of "fish tech" projects: https://docs.google.com/spreadsheets/d/1G4XX7WB5dt4D5SFQmecEKVk2xSDRxWVpPOgTDHZf9-M/edit?usp=sharing

😎 Jon Van Oast

🐟 Carly Batist, Talia Speaker, Oc

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-04-06 15:31:39

thanks for sharing here! I posted it on Twitter but forget to share here as well

Ben Best (ben@ecoquants.com)

2021-04-08 14:44:47

Hi Carly, do you manage this? Very cool to see all these! Here’s another one: https://www.sharkeye.org/

Ben Best (ben@ecoquants.com)

2021-04-08 14:46:12

And https://www.fishnet.ai/

fishnet.ai

https://www.katewing.net/

Original URL: https://www.fishnet.ai/

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-04-09 09:42:38

I don’t! I just came across it and wanted to share. I believe it is managed by Kate Wing? kate@katewing.net, https://www.katewing.net/projects

Kate Wing

Projects — Kate Wing

Essays, tools, templates, reports

Original URL: https://www.katewing.net/projects

Ted Schmitt (teds@allenai.org)

2021-04-06 14:51:07

Wow, we are clearly doing a bad job of making Skylight visible. Do you know the owner of the list? This is a good list but I see several things missing. Is this something they want to share and have the “community” add to?

vulcan.com

Skylight

Skylight helps improve maritime transparency and protect our oceans.

Original URL: https://vulcan.com/skylight

Sara Beery (sara.beery@gmail.com)

2021-04-06 15:01:38

I'm sure they would love additions (or at least I would if I were them!). but I don't know them personally.

Looks like the owner is Kate Wing, maybe you could reach out to her directly? https://www.linkedin.com/in/kate-wing-11b7029

Sara Beery (sara.beery@gmail.com)

2021-04-06 15:02:03

Sara Beery (sara.beery@gmail.com)

2021-04-06 15:02:28

kate@katewing.net

Ted Schmitt (teds@allenai.org)

2021-04-06 18:05:14

Cool, thanks

Petar Gyurov (pgyurov93@gmail.com)

2021-04-07 09:30:36

Wondering if anyone has any conservation (or related) contacts/projects in Costa Rica :flag_cr: ? Will be travelling there soon and I'm interested in opportunities and networking. Cheers!

🤩 Sara Beery

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-04-07 09:43:04

definitely reach out to the folks at Osa Conservation! https://osaconservation.org/
They do a ton of work in the Osa peninsula region, working with communities to conserve forests, wetlands, coasts, and they have a sustainable ag program

Osa Conservation

Conserving Costa Rica's Natural Treasure - Osa Conservation

Osa Conservation is a nonprofit organization dedicated to protecting the globally significant biodiversity of the Osa Peninsula, Costa Rica.

Est. reading time

3 minutes

Original URL: https://osaconservation.org/

👍 Petar Gyurov, Mitch Fennell

🌿 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Gyri Reiersen (gyri.reiersen@tum.de)

2021-04-08 07:02:02

@David?

David (dwddao@gmail.com)

2021-04-08 07:50:32

I’m currently at La Cotinga in Osa 🙂 but heading towards a large restoration project at the border of Panama, send me a DM!

😃 Carly Batist

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-08 15:26:20

I think I would have known, but is anyone else creating drone images over bird colonies in the group? I'm testing the waters for a generalization study. I've got 4 datasets (one since below). Annotated or unannotated. Let me know if there are people to contacts.

Sara Beery (sara.beery@gmail.com)

2021-04-08 15:27:25

Did you talk to @Benjamin Kellenberger?

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-08 15:27:35

ya, i wrote him.

Sara Beery (sara.beery@gmail.com)

2021-04-08 15:29:48

Also I know you chatted with Mark Koneff, but just in case I think there might be several others with datasets in the Community of Practice on migratory bird surveys he's running. Maybe we could email that group?

👍 Ben Weinstein

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-08 15:30:11

I talked to FWS this morning, but didn't know there was a group email.

Sara Beery (sara.beery@gmail.com)

2021-04-08 15:30:40

He has a mailing list for everyone who has been involved

Sara Beery (sara.beery@gmail.com)

2021-04-08 15:31:02

Could probably pass on a request if you write up a blurb

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-08 15:32:01

k, i'll DM you.

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2021-04-09 02:27:14

Hi! I can answer @Ben Weinstein’s question here: our birds dataset over West Africa is publicly available on lila.science: http://lila.science/datasets/aerial-seabirds-west-africa/
The accompanying paper can be found under this link: 21 000 birds in 4.5 h: efficient large‐scale seabird detection with machine learning - Kellenberger - - Remote Sensing in Ecology and Conservation - Wiley Online Library

LILA BC

Aerial Seabirds West Africa - LILA BC

Written by

lilawp

Est. reading time

2 minutes

Original URL: http://lila.science/datasets/aerial-seabirds-west-africa/

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-09 11:42:28

my bad! its not listed here http://lila.science/datasets

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-09 11:43:50

@Dan Morris or @Siyu Yang, bump to add to front page. Is there another way to look at this list besides the dataset tab. In case I missed anything else.

Dan Morris (agentmorris@gmail.com)

2021-04-09 12:10:25

Oops! Fixed.

Dan Morris (agentmorris@gmail.com)

2021-04-09 12:11:15

Those of you who have contributed data to LILA know that we send out a password-protected preview page first; when we make the page live, we have to un-check two boxes (for password and listing), and I forgot to un-check one of the two boxes.

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-09 12:14:48

thanks for all this organization work, it is crazy important. I'm rounding up UAV datasets for bird detections. No idea yet is atleast a baseline model is tangible in this area, the problem is pretty unconstrained across taxa, background, spatial resolution, etc.

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-14 11:43:36

Thanks @David Will, I have data from @Vienna Saccomanno already. I definitely still look for more data, either annotated (preferably) or unannotated that fits the general specs of visual birds in UAV imagery <3cm resolution.

🙌 Vienna Saccomanno

Sara Beery (sara.beery@gmail.com)

2021-04-29 16:59:42

Anyone looked at an approach like this for counting birds? https://arxiv.org/abs/2104.08391

arXiv.org

Learning To Count Everything

Existing works on visual counting primarily focus on one specific category at a time, such as people, animals, and cells. In this paper, we are interested in counting everything, that is to count...

Original URL: https://arxiv.org/abs/2104.08391

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-29 17:07:31

Thread Reply:

👍 Sara Beery

Mari Reeves (mari_reeves@fws.gov)

2021-05-21 16:36:36

I'm interested in this for boobies in Lehua. Can we talk?

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2021-04-09 13:36:05

NeurIPS 2021 will include a Datasets and Benchmarks Track. Might be a great opportunity to get some wildlife and ecological datasets out to the broader ML community!

https://neuripsconf.medium.com/announcing-the-neurips-2021-datasets-and-benchmarks-track-644e27c1e66c

Medium

Announcing the NeurIPS 2021 Datasets and Benchmarks Track

Serena Yeung and Joaquin Vanschoren

Reading time

4 min read

Original URL: https://neuripsconf.medium.com/announcing-the-neurips-2021-datasets-and-benchmarks-track-644e27c1e66c

👍 Sara Beery, Elijah Cole (Deactivated), Oc

Silvia Zuffi (silvia@mi.imati.cnr.it)

2021-04-11 08:08:51

April 13, 2021 9:00 AM - 9:50 AM CEST
Where Conservation Meets AI Enabled Vision [E32301]

Carl Chalmers
Senior Lecturer in Machine Learning and Applied Artificial Intelligence, Liverpool John Moores University
Paul Fergus
Reader in Machine Learning, Liverpool John Moores University
In response to an escalating crisis of wildlife poaching, the Conservation AI project aims to harness machine learning with a focus on video and image analytics to gain insights into animal/people interactions in the wild. We’ll present our work and the technologies implemented in our conservationai.co.uk system including projects with Knowsley Safari in the UK, the Endangered Wildlife Trust in South Africa, and the Greater Mahale Ecosystem Research and Conservation team. The session will cover both accelerated training and inference pipelines using Docker, TensorFlow-serving using NVIDIA Quadro RTX 8000, Tesla T4-GPU and NVIDIA cuDNN.

👍 Oisin Mac Aodha, Sara Beery, Ted Schmitt

👀 Srishti, Rosho

Silvia Zuffi (silvia@mi.imati.cnr.it)

2021-04-11 08:09:42

The above is an event at GTC

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-15 13:35:31

@Oisin Mac Aodha or @Elijah Cole (Deactivated), I was just talking with @Vienna Saccomanno about species classification with human guided inference. i'm rereading your paper (https://arxiv.org/abs/1906.05272) on geography priors and thinking about how to do a human-constructed prior for predicting into novel areas. Given that we may incomplete geographic information in train, or that test data might include areas outside of train, it would be useful to ask a researcher to say, what class do you expect in location X and then either use the embeddings you have done, or a more formal bayesian model (which was my plan) in jags/bugs/stan that is a function of a feature layer (e.g right before the softmax in a vanilla CNN). @gvanhorn may have thoughts here.

arXiv.org

Presence-Only Geographical Priors for Fine-Grained Image Classification

Appearance information alone is often not sufficient to accurately differentiate between fine-grained visual categories. Human experts make use of additional cues such as where, and when, a given...

Original URL: https://arxiv.org/abs/1906.05272

Sara Beery (sara.beery@gmail.com)

2021-04-15 13:37:08

One way we've thought about it is incorporating species distribution models from something like Map of Life. @Elijah Cole (Deactivated) or @Kevin Winner would be able to say more 🙂

👍 Vienna Saccomanno, Riccardo de Lutio

Oisin Mac Aodha (macaodha@caltech.edu)

2021-04-15 13:49:35

Hey @Ben Weinstein! We have had similar thoughts, but the closest we have currently gotten is to use of externally generated range maps for eval (a bit like @Sara Beery's comment). These are still not perfect e.g. if you click on the Map tab halfway down this page, you see that there have been observations in regions that are not inside the range (pink area).
https://www.inaturalist.org/taxa/42412-Ovibos-moschatus

Oisin Mac Aodha (macaodha@caltech.edu)

2021-04-15 13:50:34

Obviously, this is a very anecdotal example, but issues of this type as well as "range creep" do occur:

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-15 13:51:06

yeah, its interesting, your problem is incredibly hard because you couldn't have a person write a prior for every grid cell (though a hand drawn map would be cool).

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-15 13:51:30

for @Vienna Saccomanno she has islands that are a fixed unit and the species classes occur in somewhat known combinations at each island.

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-15 13:51:47

not every island is in train, or else we would just do a site level embedding.

Oisin Mac Aodha (macaodha@caltech.edu)

2021-04-15 13:52:05

The other question would be, is there a way to do the human in the loop version efficiently for tens of thousands of species. Im also assuming that the common (i.e. easy to guess the range of species) are well represented in your dataset anyway.

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-15 13:52:16

so I was thinking about having her parameterize a multinomial distribution by hand.

Oisin Mac Aodha (macaodha@caltech.edu)

2021-04-15 13:52:18

The island example is a great one.

Oisin Mac Aodha (macaodha@caltech.edu)

2021-04-15 13:53:06

The current hack would be to just manually generate some "pseudo positive" training examples and add them to the training set.

👍 Ben Weinstein, Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-04-15 13:53:33

rather than going to https://botorch.org/?

botorch.org

BoTorch · Bayesian Optimization in PyTorch

Bayesian Optimization in PyTorch

Original URL: https://botorch.org/

Oisin Mac Aodha (macaodha@caltech.edu)

2021-04-15 13:55:28

The challenge with any of this within the current framework (bayesian or not) is that I suspect it will be very difficult to influence the output of the network with a very small number of examples in very localized regions. The network will basically have to devote capacity to remembering just those observations.

👍 Ben Weinstein, Vienna Saccomanno

Heather Lynch (heather.lynch@stonybrook.edu)

2021-04-15 16:40:52

I think integrating HSM (vis a vis @Sara Beery’s suggestion) is really the way to go here. If ecologists believe those HSMs actually mean something, they should be useful for narrowing down a species identification.

👍 Oisin Mac Aodha

Sara Beery (sara.beery@gmail.com)

2021-04-19 15:13:48

Call for papers for ICML Climate Change AI Workshop:

"Climate Change AI will be hosting a virtual workshop on climate change and machine learning alongside the ICML 2021 machine learning conference in July. The goal of this workshop is to facilitate networking and exchange between those working in machine learning and those working in energy/climate-related areas.

There are several ways to participate:

Be a mentor or mentee in our submission mentorship program (applications due April 28th)

Submit a paper or proposal on your work at the intersection of climate change and machine learning (submissions due May 31st)

Attend! The workshop will be virtual on July 23rd or 24th, 2021.

Thinking about submitting? We are holding webinars to explain what we are looking for on Friday, April 23rd (at 1.30PM Eastern Time / 6:30 PM London time) and Tuesday, April 27th (at 9:00 AM Eastern Time / 2:00 PM London time / 9:00 PM Beijing Time). We will give advice on how to prepare a successful submission, and an opportunity to ask questions regarding the mentorship program.

For more details, see: https://www.climatechange.ai/events/icml2021.html
or reach out to climatechangeai.icml2021@gmail.com with any questions"

Climate Change AI

Tackling Climate Change with Machine Learning

NeurIPS 2020 Workshop: Tackling Climate Change with Machine Learning

Original URL: https://www.climatechange.ai/events/neurips2020#informational-webinar

Climate Change AI

Tackling Climate Change with Machine Learning

ICML 2021 Workshop: Tackling Climate Change with Machine Learning

Original URL: https://www.climatechange.ai/events/icml2021.html

👍 Oisin Mac Aodha, Elijah Cole (Deactivated), Riccardo de Lutio, Scott Hosking, Hemal Naik, Gyri Reiersen

🎉 Jon Van Oast

Jon Van Oast (jon@wildme.org)

2021-04-24 22:32:54

thought this might be of interest to folks here who might be curious about combining their research with artistic expression. proposals due by may 2.
https://zkm.de/en/open-call-viable-data-remote-residency

ZKM

OPEN CALL: Viable Data Remote Residency | ZKM

Criteria for submission of the OPEN CALL »Viable Data Remote Residency« of the ZKM | Hertz-Lab.

Original URL: https://zkm.de/en/open-call-viable-data-remote-residency

👍 Sara Beery, Frederic, Riccardo de Lutio

Anne Dangerfield (anne@arribada.org)

2021-04-27 12:33:28

Hello all! I work with the Arribada Initiative on developing projects that combine affordable tech and AI for conservation. Great to be on the channel!

👋 Declan, Jason Holmberg (Wild Me), Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Sara Beery, Carly Batist, Thijs, Alasdair Davies

👀 Alasdair Davies

💚 Alasdair Davies

John Payne (drjohnpayne@gmail.com)

2021-04-29 16:17:55

Is anyone else working with Detectron2? I’m trying to solve a problem with multi-GPU inference, which it wasn’t designed for. I’ve been using it for a year so I can reciprocate with my own hard-won knowledge of it.

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:22:24

Hi everyone --- I'm Barry Brook from the University of Tasmania. I run an extensive network of camera traps, and use the MegaDetector as part of my image-processing pipeline. My group is also developing a species classifier in EfficientNet.

👋 Sara Beery, Riccardo de Lutio, Ștefan Istrate, Tomer Nahshon

👍 Stuart Neilon

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:26:48

Awesome! I love hearing that MegaDetector is useful 🙂

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:27:38

Extremely! And it's accurate too, even for finding creatures in deep, dark Tasmanian rain forests ;)

😍 Sara Beery

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:28:10

How is the species ID going?

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:30:39

Our latest version is very good, and we're working on various ways to improve it further. We recently did a test on 22K out-of-sample images. This was the result (top-1):

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:31:24

Wow, that's fantastic! Those numbers look great 🙂

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:31:47

Thread Reply:

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:31:55

Just a better format, sorry for the replacement.

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:32:03

Yes, we're very happy with it's performance!

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:32:33

Super impressive for the Bassian thrush with 27 examples

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:32:46

It's currently trained on ~300K expert-labelled images, but next round will be on >500K, with some new categories added, and higher version of EfficientNet (the above is B3).

👍 Sara Beery

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:33:10

Oh, remember, that was just for the test set. It was trained on about 2K Bassian images.

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:33:23

ahhhhh gotcha, I thought that was training n

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:33:32

Still awesome 🙂

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:34:00

It is, given that Bassian Thrushes are actually super camouflage experts!

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:34:59

How do you use MegaDetector in combo with this model? Are you training on crops? Or is it more of an ensemble approach?

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:36:08

Exactly. We take the MegaDetector Crops, and resize them to 256px for EfficientNet. Then we write the species-label to an unused metadata field on the image, and upload to Camelot for final human verification.

👍 Olivier Gimenez

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:36:36

Oh, very cool that your workflow integrates with Camelot too!

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:37:36

Yep, it's an essential part of the pipeline. We then analyse the data by importing the Camelot database (full export) in to R, for ecological metrics etc.

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:38:16

With the images 'pre-classified' by our EfficientNet model, the human labelling of images in Camelot is about 10-50 times faster than if we were presented with the usual mixed bag one would get from a direct import of images.

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:39:01

I love that you have human verification built in to your pipeline, and that the species labels help with speedup. Have you thought about how to make humans more efficient? So far I still feel like we need to have human eyes on the IDs before using them to help mitigate ML biases, but it's hard to scale.

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:40:27

I've been thinking about ways to prioritize corrections/verification or potentially use a subset as quality control to estimate error, but everything is so imbalanced that it's hard to do robustly

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:40:38

Well, with the Camelot interface, a human is presented with 150 thumbnails on the left sidebar, which on a large screen still show quite a bit of detail. These can then be bulk-selected, and then the human simply has to look for anomalies within the 150, and deselect those. This is what makes it so fast for us.

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:41:34

That's awesome. Have you done any analysis of whether this causes any bias against small animals?

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:41:39

Thread Reply:

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:41:43

This is an example search for class 3 (Tasmanian devils).

👍 Sara Beery

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:42:01

Very Tasmania 🙂

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:42:23

Re: my above comment, just thinking about thumbnails

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:42:26

Indeed, haha.

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:42:37

and also, do humans verify the empties too?

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:42:55

It might be that some small mammals are overlooked, yes, although the detector does a good job at finding/boxing them in general, and the boxes are all quite visible on the thumbnails.

👍 Sara Beery

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:43:08

Yes, all the empties are checked.

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:43:54

that's rad. Do you have any stats on corrections? How many missed animals you see?

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:44:35

After we run the detection over our images, we partition each camera into “animal”, “human” and “blank” folders. I then manually scan the blank folder, and pick up any of the rejections and move them over to the animals folder. It doesn’t take too long to do this (typically an hour or two on a 25K image service), and it avoids having lots of false detections in the final upload to Camelot (where they would otherwise have to be rejected by a person anyway). So it’s a trade-off as to where we want the effort focused --- I like to have a clean dataset going into the database!

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:44:52

in the last set of images we processed, we set its confidence threshold to 95%. For 23,894 new images, it failed to detect 1,054 real images (4.7%) and falsely detected 331 (1.4%). Pretty good!

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:45:14

that's really cool 🙂

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:45:25

Thanks for answering so many questions!!

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:45:37

My pleasure -- thanks for all your work on this Sara!

❤️ Sara Beery

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:46:24

We were talking in our group about ways the detector could be further improved, with a view to helping classifiers. The ideal would be if a future ‘bounding box’ was not a rectangle, but a complex polygonal outline (i.e., tracing, even roughly, the animal profile). This would mean the image information about the animal would be extracted with as little ‘noise-injecting’ background as possible (even with tight boxes, often up to a half of the pixels is still non-animal). But I acknowledge that such an approach might be quite difficult/tedious to train!

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:47:54

Sooooo, if you want to try something fun, test your data on this demo: https://github.com/tensorflow/models/blob/master/research/object_detection/colab_tutorials/deepmac_colab.ipynb

research/object_detection/colab_tutorials/deepmac_colab.ipynb

``` { "nbformat": 4, "nbformat_minor": 0, "metadata": { "colab": { "name": "deepmac_demo.ipynb", "provenance": [], "collapsed_sections": [] }, "kernelspec": { "name": "python3", "display_name": "Python 3" }, "language_info": { "name": "python" }, "accelerator": "GPU" }, "cells": [ { "cell_type": "markdown", "metadata": { "id": "P-esW81yhfCN" }, "source": [ "# Novel class segmentation demo with Deep-MAC\n", "\n", "Welcome to the Novel class segmentation (with Deep-MAC) demo --- this colab loads a Deep-MAC model and tests it interactively with user-specified boxes. Deep-MAC was only trained to detect and segment COCO classes, but generalizes well when segmenting within user-specified boxes of unseen classes.\n", "\n", "Estimated time to run through this colab (with GPU): 10-15 minutes.\n", "Note that the bulk of this time is in installing Tensorflow and downloading\n", "the checkpoint then running inference for the first time. Once you've done\n", "all that, running on new images is very fast." ] }, { "cell_type": "markdown", "metadata": { "id": "Kq1eGNssiW31" }, "source": [ "# Prerequisites\n", "\n", "Please change runtime to GPU." ] }, { "cell_type": "markdown", "metadata": { "id": "UT7N0HJhiRKr" }, "source": [ "# Installation and Imports\n", "\n", "This takes 3-4 minutes." ] }, { "cell_type": "code", "metadata": { "id": "nNdls0Pe0UPK" }, "source": [ "!pip install -U --pre tensorflow==\"2.2.0\"\n", "\n", "import os\n", "import pathlib\n", "\n", "# Clone the tensorflow models repository if it doesn't already exist\n", "if \"models\" in pathlib.Path.cwd().parts:\n", " while \"models\" in pathlib.Path.cwd().parts:\n", " os.chdir('..')\n", "elif not pathlib.Path('models').exists():\n", " !git clone --depth 1 <a href="https://github.com/tensorflow/models\n">https://github.com/tensorflow/models\n</a>" ], "execution_count": null, "outputs": [] }, { "cell_type": "code", "metadata": { "id": "WwjV9clX0n7S" }, "source": [ "# Install the Object Detection API\n", "%%bash\n", "cd models/research/\n", "protoc objectdetection/protos/**.proto --pythonout=.\n", "cp object_detection/packages/tf2/setup.py .\n", "python -m pip install ." ], "execution_count": null, "outputs": [] }, { "cell_type": "code", "metadata": { "id": "sfrrno2L0sRR" }, "source": [ "import glob\n", "import io\n", "import logging\n", "import os\n", "import random\n", "import warnings\n", "\n", "import imageio\n", "from IPython.display import display, Javascript\n", "from IPython.display import Image as IPyImage\n", "import matplotlib\n", "from matplotlib import patches\n", "import matplotlib.pyplot as plt\n", "import numpy as np\n", "from objectdetection.utils import colabutils\n", "from object_detection.utils import ops\n", "from objectdetection.utils import visualizationutils as viz_utils\n", "from PIL import Image, ImageDraw, ImageFont\n", "import scipy.misc\n", "from six import BytesIO\n", "from skimage import color\n", "from skimage import transform\n", "from skimage import util\n", "from skimage.color import rgb_colors\n", "import tensorflow as tf\n", "\n", "%matplotlib inline\n", "\n", "COLORS = ([rgbcolors.cyan, rgbcolors.orange, rgb_colors.pink,\n", " rgbcolors.purple, rgbcolors.limegreen , rgb_colors.crimson] +\n", " [(color) for (name, color) in color.color_dict.items()])\n", "random.shuffle(COLORS)\n", "\n", "logging.disable(logging.WARNING)\n", "\n", "\n", "def read_image(path):\n", " \"\"\"Read an image and optionally resize it for better plotting.\"\"\"\n", " with tf.io.gfile.GFile(path, 'rb') as f:\n", " img = Image.open(f)\n", " return np.array(img, dtype=np.uint8)\n", "\n", "\n", "def resizefordisplay(image, max_height=600):\n", " height, width, _ = image.shape\n", " width = int(width ** max_height / height)\n", " with warnings.catch_warnings():\n", " warnings.simplefilter(\"ignore\", UserWarning)\n", " return util.imgasubyte(transform.resize(image, (height, width)))\n", "\n", "\n", "def getmaskprediction_function(model):\n", " \"\"\"Get single image mask prediction function using a model.\"\"\"\n", "\n", " @tf.function\n", " def predict_masks(image, boxes):\n", " height, width, _ = image.shape.as_list()\n", " batch = image[tf.newaxis]\n", " boxes = boxes[tf.newaxis]\n", "\n", " detections = model(batch, boxes)\n", " masks = detections['detection_masks']\n", "\n", " return ops.reframeboxmaskstoimage_masks(masks[0], boxes[0],\n", " height, width)\n", "\n", " return predict_masks\n", "\n", "\n", "def plotimageannotations(image, boxes, masks, darken_image=0.5):\n", " fig, ax = plt.subplots(figsize=(16, 12))\n", " ax.setaxisoff()\n", " image = (image ** darken_image).astype(np.uint8)\n", " ax.imshow(image)\n", "\n", " height, width, _ = image.shape\n", "\n", " num_colors = len(COLORS)\n", " color_index = 0\n", "\n", " for box, mask in zip(boxes, masks):\n", " ymin, xmin, ymax, xmax = box\n", " ymin **= height\n", " ymax **= height\n", " xmin **= width\n", " xmax **= width\n", "\n", " color = COLORS[color_index]\n", " color = np.array(color)\n", " rect = patches.Rectangle((xmin, ymin), xmax - xmin, ymax - ymin,\n", " linewidth=2.5, edgecolor=color, facecolor='none')\n", " ax.add_patch(rect)\n", " mask = (mask > 0.5).astype(np.float32)\n", " colorimage = np.oneslike(image) ** color[np.newaxis, np.newaxis, :]\n", " colorandmask = np.concatenate(\n", " [color_image, mask[:, :, np.newaxis]], axis=2)\n", "\n", " ax.imshow(colorandmask, alpha=0.5)\n", "\n", " colorindex = (colorindex + 1) % num_colors\n", "\n", " return ax" ], "execution_count": null, "outputs": [] }, { "cell_type": "markdown", "metadata": { "id": "ry9yq8zsi0Gg" }, "source": [ "# Load Deep-MAC Model\n", "\n", "This can take up to 5 minutes." ] }, { "cell_type": "code", "metadata": { "id": "PZ-wnbYu05K8" }, "source": [ "print('Downloading and untarring model')\n", "!wget <a href="http://download.tensorflow.org/models/object_detection/tf2/20210329/deepmac_1024x1024_coco17.tar.gz\n">http://download.tensorflow.org/models/object_detection/tf2/20210329/deepmac_1024x1024_coco17.tar.gz\n</a>", "!cp deepmac1024x1024coco17.tar.gz models/research/objectdetection/testdata/\n", "!tar -xzf models/research/objectdetection/testdata/deepmac1024x1024coco17.tar.gz\n", "!mv deepmac1024x1024coco17 models/research/objectdetection/testdata/\n", "modelpath = 'models/research/objectdetection/test_data/dee…

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:48:23

It does class-agnostic segmentation if you provide boxes, and I've found it works astonishingly well on my CT data

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:49:11

OK, we'll give it a go!

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:49:25

Could be fun to use it as weak attention for a classifier, or use this as a rough way to build the polygon labels you mentioned without needing to explicitly collect a bunch of them

Sara Beery (sara.beery@gmail.com)

2021-04-29 21:50:05

fun little research project 🙂

Barry Brook (barry.brook@utas.edu.au)

2021-04-29 21:50:15

Indeed!

Ritwik (rittyun@yahoo.com)

2021-04-30 04:50:13

great stuff👍

Dan Morris (agentmorris@gmail.com)

2021-05-01 21:31:00

New data set on LILA, courtesy of NOAA Fisheries:

http://lila.science/datasets/noaa-arctic-seals-2019/

~28k bounding boxes on seals in ~80k aerial images (color + IR).

Not to be confused with the previous big data set of seals and bounding boxes on LILA. This newer data set represents totally new images from a more recent survey, and a substantial improvement in both the precision of the annotations and the precision of the color/IR alignment.

LILA BC

NOAA Arctic Seals 2019 - LILA BC

Overview This dataset consists of around 80,000 color and IR (thermal) images, associated with flights conducted in Alaska by the NOAA Alaska Fisheries Science Center in 2019. Images have been annotated with around 28,000 bounding boxes (14,000 on color images, 14,000 on thermal images) on ice seals. Data format Metadata are provided as a .csv...

Written by

lilawp

Est. reading time

2 minutes

Original URL: http://lila.science/datasets/noaa-arctic-seals-2019/

🦭 Sara Beery, Barry Brook

👏 Alex Borowicz

Jes Lefcourt (jesl@vulcan.com)

2021-05-03 14:43:25

Shameless plug, but there are so few jobs in this field that I figure it's also a public service announcement: The EarthRanger team has three open job positions in Seattle for people who are passionate about conservation and conservation technology:

UI/UX Designer

Mobile Software Development Lead

Software QA Lead

If you know of anyone interested, please let me know and/or direct them to https://vulcan.com/Careers.aspx . Thanks!

vulcan.com

Vulcan Inc.

Working with partners to solve some of the world's biggest challenges facing oceans, climate, conservation, and communities.

Original URL: https://vulcan.com/Careers.aspx

😍 Sara Beery, Gyri Reiersen

👍 Oisin Mac Aodha, Carly Batist, Gracie Ermi

🙌 Tanya Birch

Anne Dangerfield (anne@arribada.org)

2021-05-04 11:32:42

OPPURTUNITY
Arribada Initiative and our partners are looking for ML experts to help ID individual gorillas. If anyone is interested or has a connection to someone who could help, please send a DM to @Anne Dangerfield or @Alasdair Davies

❤️ Sara Beery, Bistra Dilkina, Mike C

🦍 Alasdair Davies, Talia Speaker

Sara Beery (sara.beery@gmail.com)

2021-05-04 11:42:32

@Maxime Vidal has been looking at Re-ID recently

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-05-04 12:02:09

Perhaps also worth posting on Wildlabs in the AI for conservation group?

👍 Sara Beery

Anne Dangerfield (anne@arribada.org)

2021-05-05 09:38:59

@Carly Batist that's a good idea

😃 Carly Batist

Mike C (mike@mikecee.solutions)

2021-05-11 04:03:49

@Utkarsh Goel from the OOCAM team may be able to help

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-05-12 11:33:07

Potential event of interest - https://www.meetup.com/DataForGood-CorrelAid-X-Netherlands/events/277277935/

Meetup

Using data to conserve nature and stop wildlife crime

Wed, May 12, 2021, 7:30 PM: [Want to join without using MeetUp? Go here: <https://forms.gle/tVxUT7XC1uZSr3DT8> ]How can data and data technologies and methods be used to fight wildlife crime and support

Where

Online event

When

Wed, May 12, 2021, 7:30 PM

Original URL: https://www.meetup.com/DataForGood-CorrelAid-X-Netherlands/events/277277935/

👍 Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Jon Van Oast, Thijs

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-05-12 11:33:38

**note the website will open up in Dutch, but the event description is English and the event will be in English
Sorry for the late notice, I just came across it myself!

Ben Weinstein (benweinstein2010@gmail.com)

2021-05-13 12:53:37

I know a couple projects here are using the tree data from our group. I cleaned up training data. @Rebekah Loving https://zenodo.org/record/4746605

😍 Sara Beery, Declan, Elijah Cole (Deactivated), Rebekah Loving, Gyri Reiersen

❤️ Halil Radogoshi

Rebekah Loving (rloving@caltech.edu)

2021-05-13 14:19:53

Thank you, @Ben Weinstein!

Sara Beery (sara.beery@gmail.com)

2021-05-14 11:38:57

Pretty awesome paper:

Priority list of biodiversity metrics to observe from space

Monitoring global biodiversity from space through remotely sensing geospatial patterns has high potential to add to our knowledge acquired by field observation. Although a framework of essential biodiversity variables (EBVs) is emerging for monitoring biodiversity, its poor alignment with remote sensing products hinders interpolation between field observations. This study compiles a comprehensive, prioritized list of remote sensing biodiversity products that can further improve the monitoring of geospatial biodiversity patterns, enhancing the EBV framework and its applicability. The ecosystem structure and ecosystem function EBV classes, which capture the biological effects of disturbance as well as habitat structure, are shown by an expert review process to be the most relevant, feasible, accurate and mature for direct monitoring of biodiversity from satellites. Biodiversity products that require satellite remote sensing of a finer resolution that is still under development are given lower priority (for example, for the EBV class species traits). Some EBVs are not directly measurable by remote sensing from space, specifically the EBV class genetic composition. Linking remote sensing products to EBVs will accelerate product generation, improving reporting on the state of biodiversity from local to global scales.

https://www.nature.com/articles/s41559-021-01451-x

Nature Ecology & Evolution

Priority list of biodiversity metrics to observe from space

Remote sensing of geospatial biodiversity patterns is an important complement to field observations. This priority list suggests how remote sensing observations can be better integrated into the essential biodiversity variables.

Original URL: https://www.nature.com/articles/s41559-021-01451-x

😎 Jon Van Oast, Omiros Pantazis, Riccardo de Lutio, Ben Weinstein, Declan, Carly Batist, Tony Chang, Gyri Reiersen

📡 Stefan Schneider, Oc, aruna

Oc (boulaisoceane@gmail.com)

2021-05-18 12:17:58

@aruna and @Björn Lütjens 👀

👀 aruna, Björn Lütjens

🙌 aruna

Justin Kay (justinkay92@gmail.com)

2021-05-19 17:02:17

This sounds rad - anyone have a pdf version they can share?

Björn Lütjens (bjoern.luetjens@gmail.com)

2021-05-20 12:49:59

Thread Reply:

🙏 Oc, Justin Kay

Oc (boulaisoceane@gmail.com)

2021-05-19 15:36:46

Does anyone know how accepted abstracts will virtually present their posters for FGVC8 during CVPR21? Will they be assigned a specific time/zoom room where they’ll need to be available?

Oisin Mac Aodha (macaodha@caltech.edu)

2021-05-19 15:39:04

Hi there. We will be sending out instructions very soon.

Oc (boulaisoceane@gmail.com)

2021-05-19 15:39:45

Thank you @Oisin Mac Aodha! Response so speedy. much fast.

Oisin Mac Aodha (macaodha@caltech.edu)

2021-05-19 15:40:09

PS the workshop is on June 25th.

🙌 Oc, Sara Beery

Oc (boulaisoceane@gmail.com)

2021-05-19 15:40:29

Thanks! That gives me a window to plan around in.

Justin Kay (justinkay92@gmail.com)

2021-05-19 20:50:56

Hi everyone! I’m new here, figure I’ll introduce myself - I’m Justin Kay, I work on computer vision applications for fisheries and marine conservation/ecology through my company ai.fish. I’m interested in getting more involved in research in this area and I’m excited to learn more about all the other work going on - it’s exciting to see such a large community here (thanks for the invite @Sara Beery!), and some familiar names from my Twitter lurking…look forward to getting to know you all 🙂

❤️ Sara Beery, Caleb Robinson, aruna, Jason Holmberg (Wild Me), Bistra Dilkina, Oisin Mac Aodha, Aarnav Sawant, Erik Young

🐟 Carly Batist, Thijs, Omiros Pantazis, Mitch Fennell, Scott Hosking, Erik Young

👋 Benjamin Kellenberger, Ben Weinstein

Sara Beery (sara.beery@gmail.com)

2021-05-20 10:58:34

Nice article on the role of nature/biodiversity for human health (thanks @Oisin Mac Aodha for sharing!)

https://www.euro.who.int/en/health-topics/environment-and-health/pages/news/news/2021[…]nd-biodiversity-play-a-vital-role-in-protecting-human-health

👍 Oisin Mac Aodha, Omiros Pantazis, Mitch Fennell, Scott Hosking, Riccardo de Lutio, John Beuving

Vienna Saccomanno (v.r.saccomanno@tnc.org)

2021-05-20 17:25:47

Is anyone aware of an existing AI vegetation detection and species classification model for tree seedlings? The Nature Conservancy Palmyra has a restoration project that would benefit greatly from a veg cover map. We fly the island somewhat regularly with a Wingtra with 0.7cm GSD - example image attached. Thank you 🌴

Ben Weinstein (benweinstein2010@gmail.com)

2021-05-20 17:55:36

No, I don't believe that exists. My tree model does an adequate, but not great job, you could retrain it.

from deepforest import main
from deepforest import visualize
from matplotlib import pyplot as plt
from skimage import io

m = main.deepforest()
m.use_release()
img = io.imread("/Users/benweinstein/Downloads/image.png")

#Remove sneaky 4th transparent alpha channel
m.config["score_thresh"] = 0.01
boxes = m.predict_image(img[:,:,:3].astype("float32"), return_plot=False)
boxes.label = 0
image = visualize.plot_predictions(img, boxes, color = (255,0,255))
plt.imshow(image)

Vienna Saccomanno (v.r.saccomanno@tnc.org)

2021-05-21 13:11:12

@Ben Weinstein this is great. Will follow up, thank you.

Ben Weinstein (benweinstein2010@gmail.com)

2021-05-24 13:48:31

I know i've had meetings with a number of people here on annotation platforms and workflows. Just FYI i just sat through the 45 minutes on azure data labeling, its feeling like something promising https://docs.microsoft.com/en-us/azure/machine-learning/how-to-create-labeling-projects. I'll probably be giving it a try in a few months if anyone wants to come back to this thread. We currently use a combination of Zooniverse and QGIS.

docs.microsoft.com

Image labeling and text labeling - Azure Machine Learning

Learn how to create and run projects to label images or label text. Use ML assisted labeling, or human in the loop labeling, to aid with the task.

Original URL: https://docs.microsoft.com/en-us/azure/machine-learning/how-to-create-labeling-projects

👍 Gyri Reiersen

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2021-05-25 03:32:59

Related to @Ben Weinstein’s post I will shamelessly use this opportunity to ofﬁcially introduce AIDE v2.0.

This version brings many new exciting functionalities, including:

New models: AIDE v2.0 integrates Detectron2 and comes with 14 new deep learning models built-in! Whether it’s ResNe(X)t for image classiﬁcation or Faster R-CNN or RetinaNet for object detection, you’re now covered.

Model Marketplace: share your models across projects for maximum performance and reuse. Browse an ever-growing catalog for suitable pre-trained model states, upload your own from the disk, import public models from the Web, etc.

AIDE Model Zoo (teaser): AIDE will come with many pre-trained models for ecologists as time progresses, along with a new code base to train your own offline and publish to AIDE (coming soon). The ﬁrst models on camera trap imagery are already in the making and will be made available soon, straight through the Model Marketplace!

Workﬂow Designer: create complex model training graphs in the Web browser with zero lines of code.

Much more: accuracy evaluation, progress monitoring, versatile data management, a million bug ﬁxes, etc.

You can get it for free here: https://github.com/microsoft/aerial_wildlife_detection

🎉 Siyu Yang, Riccardo de Lutio, Vienna Saccomanno, Bistra Dilkina, Sara Beery, Mitch Fennell

🙌 Diego Marcos, Bistra Dilkina, Sara Beery

👍 Dan Morris

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2021-05-25 05:23:46

I have shown a demo of the beta of AIDE v2.0 to some of you a couple of months ago.
To the others (and to you, if you don’t mind a refresher): would you be interested in a demo session? In that case I’d gladly prepare one and communicate a day and time. Please quickly respond below or mark this comment with a ✅ if you’d like to see a demo.
Thanks!

✅ Diego Marcos

Howard L Frederick (simbamangu@gmail.com)

2021-05-30 02:21:24

@Benjamin Kellenberger fantastic news! Keen to set up with our models asap ..

Matt McCann (m_k_mccann@proton.me)

2021-05-25 08:56:22

Hey everyone, I'm new here (and to the field in general). I'm Matt McCann, currently a neurobiology PhD student and have a background in bioengineering. Right now I'm using computer vision and ML for pose segmentation/behavioral classification during predation (in the lab), as well as for more sophisticated ways of detecting neural signals in microscopy data. I'm looking to pivot into the conservation tech/AI space once I've finished my degree, and I'm interested in CV and remote sensing in the field, with applications to conservation. And a thank you to @Sam Kelly for pointing me towards this group.

👍 Sam Kelly, Omiros Pantazis, Sara Beery, Manish Rai

Koustubh Sharma (koustubh@snowleopard.org)

2021-05-27 08:21:14

Possibly a naive question, but would someone know of possible resources that one can use to detect objects in aerial images taken from drones?

Ben Weinstein (benweinstein2010@gmail.com)

2021-05-27 10:13:07

Like training data? Or a prebuilt model? What kind of objects?

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-05-27 10:13:26

You might check out AIDE: https://github.com/microsoft/aerial_wildlife_detection

Elizabeth Bondi - DATASET

Tools for detecting wildlife in aerial images using active learning

Stars

118

Language

Python

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-05-27 10:14:01

Thread Reply:

Koustubh Sharma (koustubh@snowleopard.org)

2021-05-27 10:14:04

First train, then identify. Marmot burrows in high altitude pastures

Koustubh Sharma (koustubh@snowleopard.org)

2021-05-27 10:14:38

Thanks a ton for the resources Carly, will have a look

👍 Carly Batist

Koustubh Sharma (koustubh@snowleopard.org)

2021-05-27 10:32:33

Downloading the resources from GitHub as we speak. Excited 😊

😃 Carly Batist, Sara Beery

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2021-05-27 14:32:44

@Elizabeth Bondi has a series of work on detecting humans and animals from thermal infrared images!

Paper: https://projects.iq.harvard.edu/files/teamcore/files/2018_35_teamcore_spot_camera_ready.pdf
Dataset: https://sites.google.com/view/elizabethbondi/dataset

sites.google.com

BIRDSAI The Benchmarking IR Dataset for Surveillance with Aerial Intelligence (BIRDSAI, pronounced "bird's-eye") is a long-wave thermal infrared dataset containing nighttime images of animals and humans in Southern Africa. The dataset allows for benchmarking of algorithms for automatic detection

Original URL: https://sites.google.com/view/elizabethbondi/dataset

😊 Elizabeth Bondi, Sara Beery

Elizabeth Bondi (ebondi@g.harvard.edu)

2021-05-27 14:36:18

Thanks @Lily Xu - OLD ACCOUNT, DO NOT MESSAGE! Please let me know if you look into this and have questions, @Koustubh Sharma.

😊 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Koustubh Sharma (koustubh@snowleopard.org)

2021-05-27 21:05:46

Hi Elizabeth, I have never used the docker, and am getting some errors when trying to run it on my computer. If not much of a problem, can I bug you sometime this week or the next for a little assistance on running the tool?

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-05-27 10:35:05

WWF’s Dave Thau is doing a series on AI for conservation and managing planetary data -
https://medium.com/g-ai-a/g-ai-a-artificial-intelligence-and-planetary-scale-environmental-management-e34f2571761e

Medium

G.AI.A: Artificial Intelligence and Planetary-Scale Environmental Management

Can we build an AI-driven system for managing the planet? Perhaps — but let’s be realistic about the challenges & what it can achieve.

Reading time

8 min read

Original URL: https://medium.com/g-ai-a/g-ai-a-artificial-intelligence-and-planetary-scale-environmental-management-e34f2571761e

😍 Sara Beery, Talia Speaker

👍 Casey Youngflesh, Riccardo de Lutio, Bistra Dilkina, Monty Ammar

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-03 12:54:43

I've talked with a few people here about the bird detector for high-res airborne imagery. Here is a teaser with an early release model. It will continue to improve. All feedback welcome. https://colab.research.google.com/drive/1e9_pZM0n_v3MkZpSjVRjm55-LuCE2IYE?usp=sharing

colab.research.google.com

Google Colaboratory

Original URL: https://colab.research.google.com/drive/1e9_pZM0n_v3MkZpSjVRjm55-LuCE2IYE?usp=sharing

👍 Mikey Tabak, Gyri Reiersen

aruna (arunas@mit.edu)

2021-06-03 13:13:45

Beautiful stuff!

aruna (arunas@mit.edu)

2021-06-03 13:13:50

Where's the imagery from?

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-03 13:35:46

Antarctica, and Minnesota

aruna (arunas@mit.edu)

2021-06-03 14:10:10

I see, is it an open dataset?

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-03 14:10:25

not yet. perhaps in the future. this is demo only.

👍 aruna

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-03 12:56:47

[ MESSAGE TEXT EMPTY ]

👍 gvanhorn, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Sara Beery, Thijs, Barry Brook, David Will, Gyri Reiersen

🦉 Benjamin Hoffman, Sara Beery, Gyri Reiersen

:the_horns: Declan, Sara Beery, Mitch Fennell, Gyri Reiersen

🐦 Vienna Saccomanno, Tony Chang, Gyri Reiersen

Sara Beery (sara.beery@gmail.com)

2021-06-04 10:26:54

Symposium on ML for Biodiversity: https://poisotlab.github.io/ml-biodiv-symposium/

poisotlab.github.io

Machine Learning for Biodiversity

An online symposium to share tools, ideas, and knowledge

Original URL: https://poisotlab.github.io/ml-biodiv-symposium/

👍 Riccardo de Lutio, Ritwik, Omiros Pantazis, Jason Holmberg (Wild Me)

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-07 12:25:14

@Sarra Alqahtani any update on cross-scale training? I'm finding that my bird detection object detection pipeline (see above) is incredibly sensitive to the scale of the input images. For example, changing the input size of one of the datasets from 1100 px crops to 1300 px crops changes performance by 60%. Everything gets resized to 224 x 224 as it goes into the network, so the input scale is just adjusting how large a bird appears in an image. We start with giant orthomosaic tiles, so they have to be cut into crops. It is this step that appears to be very sensitive. In blue is prediction, in orange is annotation. Smaller representations work better. Thoughts from others? @Tony Chang @Elijah Cole (Deactivated) @Benjamin Kellenberger @John Brandt. I already do heavy data augmentation to crop and resize to zoom in and out during training, but I guess it is not enough. Added wrinkle is that this is 'zero shot' background object detection, so we are testing on a novel dataset that is withheld from training (like using south africa to predict australia, etc). Each dataset has different ground resolution.

Sara Beery (sara.beery@gmail.com)

2021-06-07 13:07:18

Have you tried more than those 2 resolutions? I'd be curious what the performance/input size curve looks like and whether there's any way to predict what resolution in your test data will do best based on your training data/model.

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-08 16:47:36

Just illustrating what the problem is. Here is with no data augmentation on a held-out dataset at new resolution. It makes mini ducks. Prediction in blue.

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-08 16:47:54

Working on even heavier zoom and cropping.

Sara Beery (sara.beery@gmail.com)

2021-06-08 16:48:21

Very interested, following along for more updates 🙂

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-14 19:37:08

if anyone would like to join, @Elijah Cole (Deactivated) and I are discussing his excellent paper tomorrow at 10PT on self-supervised learning https://arxiv.org/pdf/2105.05837.pdf. I am planning on a similar work for 80 species tree classification.

❤️ Sara Beery, Rebekah Loving

👍 gvanhorn, Ankita Shukla, Omiros Pantazis, Riccardo de Lutio

Elijah Cole (Deactivated) (ecole@caltech.edu)

2021-06-14 20:44:39

TLDR for the paper:

“When Does Contrastive Visual Representation Learning Work?”
https://arxiv.org/abs/2105.05837

Recent self-supervised learning techniques like SimCLR are able to learn good representations for ImageNet classification without using any labels. Do these techniques still work if you care about datasets that are not similar to ImageNet?

In the paper, we run a bunch of experiments to figure out:

How many images you need for both pretraining and downstream supervision;

How important it is to pretrain on images relevant to your downstream task;

Whether you can pretrain on datasets with noisy/degraded images; and

How well these methods perform on challenging fine-grained tasks e.g. iNat21.

Feel free to let me know if you have any questions!

arXiv.org

When Does Contrastive Visual Representation Learning Work?

Recent self-supervised representation learning techniques have largely closed the gap between supervised and unsupervised learning on ImageNet classification. While the particulars of pretraining...

Original URL: https://arxiv.org/abs/2105.05837

👍 Oisin Mac Aodha

❤️ Subhransu Maji

Elijah Cole (Deactivated) (ecole@caltech.edu)

2021-06-14 20:44:47

Some of my favorite results:

The gap between supervised and self-supervised performance on state-of-the-art fine-grained datasets like iNat21 is huge (~30% top-1).

There isn’t much benefit to pretraining on more than 500k images.

Adding more pretraining images from different domains does not necessarily lead to more general representations.

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2021-06-15 02:54:40

Awesome! Seems to be a great work indeed; I’m looking forward to reading it. Thanks @Elijah Cole (Deactivated)!

🙂 Elijah Cole (Deactivated)

Sara Beery (sara.beery@gmail.com)

2021-06-16 11:13:40

From the CCAI Newsletter: "TDWG 2021, the virtual conference of Biodiversity Information Standards (TDWG), which will be held 18–22 October 2021. Abstracts will be submitted to and published in Biodiversity Information Science and Standards (BISS). Symposium on applications of machine learning in biodiversity image analysis to be organized. The deadline for the submission of abstracts is August 2"

tdwg.org

TDWG 2021 - Virtual Annual Conference - TDWG

Biodiversity Information Standards (TDWG) is a non-profit organization and community dedicated to developing biodiversity information standards.

Original URL: https://climatechange.us3.list-manage.com/track/click?u=a5463f28627a77a4b2a79e7d0&id=3b116f8769&e=9746f73f39

tdwg.org

Biodiversity Information Standards (TDWG)

Biodiversity Information Standards (TDWG) is a non-profit organization and community dedicated to developing biodiversity information standards.

Original URL: https://climatechange.us3.list-manage.com/track/click?u=a5463f28627a77a4b2a79e7d0&id=019c70a176&e=9746f73f39

biss.pensoft.net

Biodiversity Information Science and Standards

description

Original URL: https://climatechange.us3.list-manage.com/track/click?u=a5463f28627a77a4b2a79e7d0&id=463e69a28d&e=9746f73f39

tdwg.org

Symposia, Workshops, & Discussion Sessions - TDWG

Biodiversity Information Standards (TDWG) is a non-profit organization and community dedicated to developing biodiversity information standards.

Original URL: https://climatechange.us3.list-manage.com/track/click?u=a5463f28627a77a4b2a79e7d0&id=4ddfd105b8&e=9746f73f39

👍 Oisin Mac Aodha, Omiros Pantazis, Beckett Sterner, Nico Franz, Jason Holmberg (Wild Me), Atriya Sen, David Russell

Ritwik (rittyun@yahoo.com)

2021-06-17 04:24:42

Hi, I was wondering if someone has experience about dealing with watermarks in images while training a classifier. I'm using google images to build training data for certain specific species but many times there are watermarks on the images. The models will mostly likely be sensitive to the watermarks for classification. Has anyone encountered this issue?

Elijah Cole (Deactivated) (ecole@caltech.edu)

2021-06-17 09:46:08

I don’t know how prevalent they are, but I know there are some images with various watermarks in standard datasets like ImageNet and Places365. People don’t seem to worry about them too much in practice.

@Sara Beery didn’t you look into this for camera trap images once?

Sara Beery (sara.beery@gmail.com)

2021-06-17 10:11:37

A bit! camera traps frequently have these bars at the top and bottom that are distinct to camera type. I've experimented with cropping them out and with leaving them in and found it didn't make much of a difference.

Ritwik (rittyun@yahoo.com)

2021-06-17 10:40:38

Thanks a lot for your inputs. I'm assuming if the frequency, appearance and location of the marks, are more or less equally distributed across classes it will not affect global evaluation metrics. Or if their frequency is so low that it is drowned in the standard error margin. In my case, I fear, they might end up being discerning features. Will try some experiments..

Sara Beery (sara.beery@gmail.com)

2021-06-17 10:46:16

In my case, I think they are discerning features, but that there are a bunch of other discerning but correlative features the model can focus on just in the static backgrounds.

Sara Beery (sara.beery@gmail.com)

2021-06-17 10:46:31

So with or without them, the model finds it easy to overfit

Ritwik (rittyun@yahoo.com)

2021-06-17 11:12:24

true, a lot of overfitting opportunities for the model 🙂

Ben Augustine (ben.augustine@cornell.edu)

2021-06-17 12:07:27

Interesting, Sara. I was concerned about these on camera trap images, but have just ignored them rather than try to crop them out. In my case, different camera brands are deployed at different sites and different sites have a somewhat different suite of species, so I thought this may increase overfitting. Irrelevant camera information on the photos correlates with species occupancy/prevalence.

👍 Sara Beery

Sara Beery (sara.beery@gmail.com)

2021-06-17 12:12:12

yes, the bars correlate with species prevalence, but so does the entire background of the image and underlying hardware-specific image statistics? So I guess my point is that yes, models can focus on the bars but without the bars they still have a lot of opportunity to overfit - which is my hypothesis as to why removing the bars didn't do much

Ben Augustine (ben.augustine@cornell.edu)

2021-06-18 13:47:36

Makes sense.

Sara Beery (sara.beery@gmail.com)

2021-07-15 10:06:00

Ok, I take it back, if you add a project that only has images of one species, and it's a camera type that isn't represented previously, then yep. Massive overfitting to the logo. I think probably in previous experiments I was doing there was a lot more variability in species per camera type so it was less of a problem.

Ritwik (rittyun@yahoo.com)

2021-07-15 12:10:42

thanks @Sara Beery makes sense.. if there's a constant signal for a particular category it will be picked up by filter weights.. i'll update this space once I have some curated data (long task)

👍 Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-18 14:21:10

@gvanhorn or @Holger Klinck, just a thought, you've probably already done this. I'm prepping for a trip to S. ecuador and been listening to local recordings. There are many recordings that lack 'additional' bird info metadata, but whose sonograms clearly show other species. Like this (bay wren?) embedded in this recording. Have you explored semi-supervised weak labels to try to draw info from these unlabeled, but high quality recordings?

gvanhorn (grv22@cornell.edu)

2021-06-18 14:22:01

Not specifically, but some of that info is captured in the checklist associated with the audio.

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-18 14:24:24

k, sorry if that was naive idea, but its interesting how often I feel like the unlabeled auxiliary species is actually better quality than the, often rarer, labeled, target species. Like this one where, i'm not sure the target is, but I hear glossy backed thrush? and a tapaculo?

gvanhorn (grv22@cornell.edu)

2021-06-18 14:25:05

Yeah, welcome to audio!

😅 Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-18 14:25:34

lol, sorry, didn't mean to :https://xkcd.com/1831/ i'll go back to images.

xkcd (https://xkcd.com/)

Here to Help

[Title text] ""We TOLD you it was hard." "Yeah, but now that I'VE tried, we KNOW it's hard.""

Original URL: https://xkcd.com/1831/

😅 gvanhorn, Carly Batist, Sara Beery

gvanhorn (grv22@cornell.edu)

2021-06-18 14:25:47

We’ve actually been hard at work on this problem! Stay tuned for some cool stuff rolling out over the next few weeks and months

👍 Ben Weinstein, Carly Batist, Sara Beery

gvanhorn (grv22@cornell.edu)

2021-06-18 14:23:47

And the recent work of @Elijah Cole (Deactivated) is relevant for this particular example (i.e multilabel problems where you only have 1 positive label for each training sample)

👍 Sara Beery, Oisin Mac Aodha

Elijah Cole (Deactivated) (ecole@caltech.edu)

2021-06-18 15:19:45

Here’s a link:

https://arxiv.org/abs/2106.09708

Happy to discuss!

arXiv.org

Multi-Label Learning from Single Positive Labels

Predicting all applicable labels for a given image is known as multi-label classification. Compared to the standard multi-class case (where each image has only one label), it is considerably more...

Original URL: https://arxiv.org/abs/2106.09708

🙌 Sara Beery, Oisin Mac Aodha

👍 Mikey Tabak

Elijah Cole (Deactivated) (ecole@caltech.edu)

2021-07-15 10:48:29

The code is now available as well if anyone wants to play around with it:

https://github.com/elijahcole/single-positive-multi-label

elijahcole/single-positive-multi-label

Multi-Label Learning from Single Positive Labels - CVPR 2021

Website

<a href="https://arxiv.org/abs/2106.09708">https://arxiv.org/abs/2106.09708</a>

Stars

9

gvanhorn (grv22@cornell.edu)

2021-06-23 13:06:29

Super excited to announce the launch of Sound ID for Merlin. Simply hold your phone up, press record, and Merlin will help you ID who’s singing in real time! If you happen to be in the US or Canada, download it today and try it out in your own backyard. Over the next months (and years!) we’ll be expanding the feature to cover more species and regions. https://merlinbirdid.page.link/sound-id

merlinbirdid.page.link

Merlin Bird ID, now with Sound ID!

Merlin helps you learn bird songs by showing realtime suggestions as you record the birds around you.

Original URL: https://merlinbirdid.page.link/sound-id

😍 Sara Beery, Ben Weinstein, Talia Speaker, Declan, Beckett Sterner, Casey Youngflesh, Carly Batist, Stefan Schneider, Subhransu Maji, Lauren E. Gillespie, Justin Kay, Dan Morris, Riccardo de Lutio, Gyri Reiersen, Cody Kupferschmidt, Michael Wethington, Ayan Mukhopadhyay

🐦 Ed Miller, Thijs, Stefan Schneider, Sara Beery, Michael Wethington

😎 Jon Van Oast, Sara Beery, Ritwik, Michael Wethington

gvanhorn (grv22@cornell.edu)

2021-06-23 13:07:47

@Ben Weinstein it doesn’t cover S. ecuador yet but we’ll get there soon!

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:08:10

awesome. I'm going outside to try it now.

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:09:21

i'm also trying to do my saw-whet call to see what happens.

🤓 Casey Youngflesh, Sara Beery

Sara Beery (sara.beery@gmail.com)

2021-06-23 13:10:34

lol BENNNNN we're in a workshop 🤣

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:10:44

i'm already outside.

🤩 Sara Beery

Sara Beery (sara.beery@gmail.com)

2021-06-23 13:10:47

(ie I want to go outside and try it too)

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:16:30

okay @gvanhorn I have some questions. First, amazing, congrats, truly a remarkable accomplishment.

😊 gvanhorn

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:16:45

Can I opt in to let Cornell save my recordings to improve future models?

➕ Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:17:08

Could I annotate them here, like, select 'yes' or no for a given species?

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:17:54

like it was perfect on lesser goldfinch, but then tossed a hutton's vireo in there (it was the same call), you can imagine users contributing in a weak supervised way.

Ed Miller (ed@hypraptive.com)

2021-06-23 13:18:21

@gvanhorn Is this feature of Merlin related to BirdNET?

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:18:24

but watching it click on a song sparrow (nice and close) is a warm and fuzzy feeling.

gvanhorn (grv22@cornell.edu)

2021-06-23 13:19:02

Uploading recordings to the Macaulay Library will hopefully be available by this fall. I think it will also support uploading photographs, and could have tighter integration with eBird.

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:19:26

brilliant. The data gathering aspect of this is so valuable, like tesla watching you drive.

👍 Sara Beery, gvanhorn

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:20:13

it did a pretty good job at a distant set of species, i was impressed with the western scrub jay it grabbed from several houses over.

👍 gvanhorn

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:21:03

what was the key innovation that made this happen, $ for developers? More clean data? Algorithmic improvements?

gvanhorn (grv22@cornell.edu)

2021-06-23 13:21:16

And then plenty of headroom for improved performance. One limitation that I’ve been trying to work around within Merlin is that taxonomic predictions isn’t well supported by the app UI. So we’ll hopefully keep improving upon egregious mistakes, but then more fine-grained mistakes will require a bigger UI update that let’s me return taxonomic probabilities.

gvanhorn (grv22@cornell.edu)

2021-06-23 13:23:49

Sound ID benefitted from some fore-runners like BirdNet and BirdVox (@Ed Miller this sort of answers your question), so that got us up and running pretty quickly. Then it was experience with dataset building and working with experts to refine the annotations that got us to a point where we were reasonably happy with the live experience. Certainly the experience with Seek helped out here

👍 Sara Beery

gvanhorn (grv22@cornell.edu)

2021-06-23 13:27:13

@Ben Weinstein I guess the answer is: right team and the right data.

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:33:10

I think that's a good answer. You didn't say, we need new and interesting and complex models.

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:33:33

when I give talks I always ask, do we need new models, new theory or new data.

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:33:40

and clearly all three are nice.

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 13:33:49

but where should we focus/

Sara Beery (sara.beery@gmail.com)

2021-06-23 13:42:12

I feel like Grant in particular is really fantastic at curating accurate labeled data with experts in the loop, and not just the data collection but getting these difficult labels to be accurate was key.

😊 gvanhorn

gvanhorn (grv22@cornell.edu)

2021-06-23 13:53:42

A teaser blog post (about future blog posts that will have more details….) can be found here: https://www.macaulaylibrary.org/2021/06/22/behind-the-scenes-of-sound-id-in-merlin/

macaulaylibrary.org

Behind the Scenes of Sound ID in Merlin

What is Sound ID? Today we announced one of our biggest breakthroughs—Sound ID, a new feature in the Merlin Bird ID app—and a major leap forward in sound identification and machine learning

Original URL: https://www.macaulaylibrary.org/2021/06/22/behind-the-scenes-of-sound-id-in-merlin/

❤️ Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-23 15:42:41

@gvanhorn if there is a way for me to demo a pack for the tropics when I go, happy to give feedback. Not sure if that's feasible/useful.

gvanhorn (grv22@cornell.edu)

2021-06-23 16:02:21

It is possible, you can join the Merlin Beta here: https://merlin.allaboutbirds.org/beta/
The Beta testers will get the upcoming packs first, but currently the model in the Beta and the Production app are the same.

Merlin Bird ID - Free, instant bird identification help and guide for thousands of birds

Join the Merlin Beta Test Team

Interested in helping us make Merlin better? Sign up to be a beta tester! Explore new Merlin features before they're available to the public and help us improve Merlin by catching bugs and issues. Android Apply for membership in this group: <https://groups.google.com/forum/#!forum/merlin-bird-i>

Original URL: https://merlin.allaboutbirds.org/beta/

👍 Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2021-06-28 15:49:42

i've been playing with this a bit more, really interesting. Is there a long term interest in making an API so that scientists can query recordings against the model?

🤞 Sara Beery

gvanhorn (grv22@cornell.edu)

2021-06-29 12:36:08

Good idea! Yes and no. Yes: definitely interested and I can see the benefit it would provide to scientists. No: I don’t have the funds queued up to support that kind of API, although I don’t think it would cost too much run a bare bones / no fuss api. But we’ll see how things shake out this summer.

👍 Ben Weinstein

gvanhorn (grv22@cornell.edu)

2021-06-23 13:36:27

Pro tip for folks that want to squeeze more out of the Sound ID model: pickup an external mic! We’ll keep optimizing the model for built-in microphones, but a pretty big boost in average precision can be achieved by just using a better mic. I’ve been playing around with the following setup: Rode VideoMic GO, TRS to TRRS cable (male to male), and the head phone jack dongle for an iPhone.

👍 Sara Beery

Ed Miller (ed@hypraptive.com)

2021-06-24 21:30:16

Is anyone attending the CV4Animals Workshop at virtual CVPR tomorrow? If so, drop by the BearID poster (Paper ID 32) and say "hi'!

❤️ Sara Beery, Elizabeth Bondi, Ankita Shukla

🎉 Jon Van Oast

:bearid: Omiros Pantazis, Carly Batist

Sara Beery (sara.beery@gmail.com)

2021-06-24 21:31:37

I'll be there with @Elizabeth Bondi!!

🎉 Elizabeth Bondi, Jon Van Oast

Ed Miller (ed@hypraptive.com)

2021-06-24 21:34:03

I've never done a poster session. Let alone a virtual session on Gatherly! 😧

Elizabeth Bondi (ebondi@g.harvard.edu)

2021-06-24 21:56:30

We aren't totally sure what to expect either, but it should be fun 🙂

❤️ Sara Beery

:bearid: Ed Miller

Silvia Zuffi (silvia@mi.imati.cnr.it)

2021-06-25 06:06:18

I will also attend, no idea how this virtual poster session could go, let’s see!

❤️ Sara Beery, Elizabeth Bondi

:bearid: Ed Miller

Sara Beery (sara.beery@gmail.com)

2021-06-25 16:25:33

FANTASTIC talk @Silvia Zuffi!!!

🎉 Elizabeth Bondi

👏 Ed Miller

Silvia Zuffi (silvia@mi.imati.cnr.it)

2021-06-25 16:44:20

Thanks!!

Yuval Boss (yuval@yuvalboss.com)

2021-06-25 14:02:56

Hi Everyone,

I work for NOAA's Polar Ecosystems Program on aerial object detection/classification of ice-associated seal species and polar bears. I also collaborate with other projects here in the Marine Mammal Lab trying to implement ML for different detection/classification/tracking tasks in aerial or satellite image data. All the work on my end leads to aiding researchers in population assessment studies.

Some of the interesting areas of my work have to do with the challenges of:
- small object detection in real-time
- challenges with class imbalances and small number of labels for some classes(200)
- how the results of these models can be used by statisticians in their assessment/modeling of abundance
- multimodal approaches for localization/classification in aligned thermal/color(and soon UV) imagery.
- object tracking across frames
- image quality assessment

Here is a short article from a year ago sharing what our project is doing at a high level: fisheries.noaa.gov/feature-story/developing-artificial-intelligence-find-ice-seals-and-polar-bears-sky

My background is in CS and would love to collaborate more with other developers working on similar problems on the technical side. Please reach out if you are working on similar problems and want to chat about them!

Thank you Sara for setting this up, this community seems awesome from what I've seen so far and I'm really glad to be a part of it.

Yuval Boss - yuval.boss@noaa.gov

NOAA

Developing Artificial Intelligence to Find Ice Seals and Polar Bears from the Sky

Several million photos that might contain polar bears; a call to jury duty; and a Microsoft Hackathon. This series of events led NOAA Fisheries scientists at the Alaska Fisheries Science Center to team up with tech industry talent to develop Artificial Intelligence (AI) for Arctic marine mammal research.

Original URL: https://www.fisheries.noaa.gov/feature-story/developing-artificial-intelligence-find-ice-seals-and-polar-bears-sky

👋 gvanhorn, Stefan Schneider, Sara Beery, Declan, Ben Weinstein, Lloyd Hughes, Dan Morris, Medhini Gulganjalli Narasimhan, Gavin Kyte, David Russell

🐻‍❄️ Stefan Schneider, Carly Batist, Sara Beery, Jason Holmberg (Wild Me), Gavin Kyte

🦭 Carly Batist, Sara Beery, Caleb Robinson, Jason Holmberg (Wild Me), Tony Chang

👍 Benjamin Kellenberger, Jon Van Oast, Caleb Robinson, Jason Holmberg (Wild Me)

Gyri Reiersen (gyri.reiersen@tum.de)

2021-07-01 04:56:02

Came across this project: https://seabee.no/
Using drones and AI to monitor and map the biodiversity of the Norwegian coastal zone, specifically used for bird counting. Super cool!

seabee.no

Seabee Norway

Norwegian Infrastructure for Drone-based Research, Mapping and Monitoring in the Coastal ZoneInnovation in the world of drone-based research.

Estimated reading time

10 minutes

Original URL: https://seabee.no/

🐥 Sara Beery, Jason Holmberg (Wild Me), David

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-07-01 12:08:30

Cool AI initiative I came across - http://www.namethatfish.com/

Name That Fish

Fish Identification | Name That Fish

Want to know what you saw underwater? <a href="http://namethatfish.com">namethatfish.com</a> uses artificial intelligence to help you find out!

Original URL: http://www.namethatfish.com/

❤️ Sara Beery, Jason Holmberg (Wild Me), Gracie Ermi

👍 Justin Kay, Jason Holmberg (Wild Me)

Ben Weinstein (benweinstein2010@gmail.com)

2021-07-01 13:22:10

any idea who is behind this? No contact on the website? did i miss it?

Holger Klinck (hk829@cornell.edu)

2021-07-01 13:25:43

https://www.linkedin.com/in/jake-scubotics/

Ben Weinstein (benweinstein2010@gmail.com)

2021-07-01 13:26:21

funny, that's my institution, never heard of it.

Mike C (mike@mikecee.solutions)

2021-07-06 13:39:28

Can our datasets be of value to you? A few project updates

The University of Hong Kong has just developed an app called ‘Saving Face’ for i-id of napoleon wrasse:

https://onlinelibrary.wiley.com/doi/abs/10.1002/aqc.3199
https://www.scmp.com/magazines/post-magazine/short-reads/article/3134123/facial-recognition-app-used-protect-endangered

The team has developed other ML applications in conservation such as:
https://www.clearbot.org/
https://openoceancam.com/

South China Morning Post

Facial recognition app helps protect endangered fish species

A facial recognition app created in Hong Kong is being used to protect an endangered species of coral reef fish called the humphead wrasse.

Original URL: https://www.scmp.com/magazines/post-magazine/short-reads/article/3134123/facial-recognition-app-used-protect-endangered

Clearbot

Clearbot | Hong Kong

Building fleets of robots that clean our oceans.

Original URL: https://www.clearbot.org/

Ben Weinstein (benweinstein2010@gmail.com)

2021-07-08 13:00:35

I had a researcher send me annotations in photoshop. Any one want to guess about how I might try to extract them. Some of sort of export vector graphics as svg?

🙀 Sara Beery

Jes Lefcourt (jesl@vulcan.com)

2021-07-08 13:15:28

Yep. If you've got the PSD file, exporting to SVG and then parsing shouldn't be too hard.

Ben Weinstein (benweinstein2010@gmail.com)

2021-07-08 13:16:10

thanks. i'll update here. Right now i've got the JPG, but requesting the original.

Ben Weinstein (benweinstein2010@gmail.com)

2021-07-08 14:51:30

I made a tiny clip on annotating images in qgis if its useful to others. This was a small addition to the meeting we had with @Vienna Saccomanno @Benjamin Kellenberger @John Payne https://deepforest.readthedocs.io/en/latest/bird_detector.html

👍 Sara Beery, David Will

Jonathan Crall (erotemic@gmail.com)

2021-07-09 14:33:20

@Ben Weinstein I've done something similar before (I think on that same - or at least a similar dataset.

🎉 Jon Van Oast, Sara Beery

Jonathan Crall (erotemic@gmail.com)

2021-07-09 14:33:21

https://gitlab.kitware.com/viame/bioharn/-/blob/master/dev/data_tools/read_ps_count_jpeg.py

GitLab

dev/data_tools/read_ps_count_jpeg.py · master · viame / bioharn

GitLab Community Edition

Original URL: https://gitlab.kitware.com/viame/bioharn/-/blob/master/dev/data_tools/read_ps_count_jpeg.py

Jonathan Crall (erotemic@gmail.com)

2021-07-09 14:36:43

Look at the `parsephotoshopcount_annots function, photoshop literally has a special annotation layer. In my case it was dots, but you might have circles. Figuring all this out took me way too much time, hopefully my initial legwork helps you out.

Jon Van Oast (jon@wildme.org)

2021-07-09 14:48:56

slightly off-topic, but i often think about putting the annotation data we have in an image as part of a comment in jpeg. is it crazy to imagine trying to develop a "standard" for this? (e.g. some kind of flexible chunk of json -- maybe stored as a jpeg segment or exif MakerNote )

Jonathan Crall (erotemic@gmail.com)

2021-07-09 15:05:20

Hey Jon 🙂, I think storing annotation data inside the image itself it probably not a great idea. But I do think existing annotation formats are lacking. I do like the general idea behind MS-COCO because it provides a relatively intuitive way to specify annotations, and I've been working on an extension of it called KW-COCO, which adds support for images-sequences, multispectral imagery, and richer annotations (like polygons with holes):

https://gitlab.kitware.com/computer-vision/kwcoco

GitLab

computer-vision / kwcoco

The Kitware COCO Image Annotation Module

Original URL: https://gitlab.kitware.com/computer-vision/kwcoco

Jonathan Crall (erotemic@gmail.com)

2021-07-09 15:08:50

I also saw there is a CameraTraps coco format: https://github.com/microsoft/CameraTraps/tree/master/data_management

There is mostly a 1-to-1 conversion between the extensions there and the extensions in the kwcoco spec. (e.g. "seqid" -> "videoid", "framenum" -> "frameindex")

👍 Sara Beery

👋 Jon Van Oast

😎 Jon Van Oast, David Russell

Jon Van Oast (jon@wildme.org)

2021-07-09 15:11:09

yeah, it feels a bit hacky - especially for large datasets where it makes way more sense to have an external set of annotation data. but as a "why not?" kinda extra layer, might be kinda fun. 🙂 thanks much for the pointers on just the general topic of format of data!

Jonathan Crall (erotemic@gmail.com)

2021-07-09 15:13:11

The kwcoco CLI has a command that is supposed to help grab data and convert it into kwcoco. Currently it support for cifar, domainnet, sapcenet7, and camvid, but I'm looking into adding datasets from lila.science. I also recently found out about a library that scikit-image uses for fetching data files called "pooch": https://github.com/fatiando/pooch . Ultimately, it would be really cool to have an IPFS solution for all of these public / benchmark / challenge datasets.

fatiando/pooch

A friend to fetch your data files

Website

<a href="https://www.fatiando.org/pooch">https://www.fatiando.org/pooch</a>

Stars

125

Jon Van Oast (jon@wildme.org)

2021-07-09 15:14:32

i am not sure if there is a strict definition of "annotation" that i am breaking here, but i wonder if this is adaptable to some of the types of features we have historically been interested in, like edges (fluke / fin) and sets of points.

Jonathan Crall (erotemic@gmail.com)

2021-07-09 15:17:46

kwcoco does have support for keypoints, and in general you can add any extra data you want, as long as there is some bounding box to localize where the annotation roughly is. At one point I did have line annotations for scallop radii, but I dropped them in favor of converting them to a polygon. The fluke use-case is a good example of a different sort of geometry that really should be supported and doesn't fit in cleanly with the existing keypoint / box / polygon paradigms.

👍 Sara Beery

Mike C (mike@mikecee.solutions)

2021-07-10 10:57:30

Notion’s ‘Sessions’ will be available free to all Beta users who join and complete below checklist once Sessions launches July 12 2021 🙂

https://www.notion.so/Sessions-is-forever-free-for-its-active-beta-users-a7f6502c63eb40009add29cf14b03d09

Sessions Community on Notion

Sessions is forever free for its active beta users

👀 Sessions, powered by Flow OS, is a web-based platform that offers the one-window solution to hyper-optimize your meetings.

Original URL: https://www.notion.so/Sessions-is-forever-free-for-its-active-beta-users-a7f6502c63eb40009add29cf14b03d09

Eric Greenlee (Eric.greenlee.96@gmail.com)

2021-07-12 09:42:18

Hi everyone! Thanks to @Lily Xu - OLD ACCOUNT, DO NOT MESSAGE for adding me to this group. I'm an EE looking to work on problems in conservation and am hoping for recommendations for grad programs (or companies). I've worked as an RF engineer for the past 3 years designing a variety of wireless systems and ideally want to continue doing this kind of work, but I'm open to other hardware projects. I've had trouble finding PhD programs that apply engineering to problems in conservation (rather than doing longer-term EE research) and was hoping you software people might know of cool hardware things in the works. Also open to jobs in private industry. Thanks in advance!

👋 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Sara Beery, Elizabeth Bondi, Jason Holmberg (Wild Me), Mike C

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2021-07-12 09:43:08

Welcome Eric!!

❤️ Eric Greenlee

Sara Beery (sara.beery@gmail.com)

2021-07-12 09:51:47

Welcome!! One way you can do this, and I think what a lot of folks here have done, is find a PhD advisor who will support this kind of interdisciplinary work within a traditional EE/CS type program

Sara Beery (sara.beery@gmail.com)

2021-07-12 10:02:36

For example, @Elijah Cole (Deactivated), @gvanhorn and I all did/are doing our PhD's with @Pietro Perona (who is the best, thanks Pietro!!) who is supportive of our work on different aspects of real-world problems in the natural world and other social good domains like medicine. Milind Tambe advises @Lily Xu - OLD ACCOUNT, DO NOT MESSAGE and @Elizabeth Bondi and has a similar passion for work that has positive societal impact. If you're more on the hardware side one person you could reach out to to get a better sense of the lay of the land might be Andrew Schultz, who is a PhD student in MechE at Georgia Tech and does awesome conservation-focused research and education (https://streamerlinks.com/StreamingScience)

❤️ Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Elizabeth Bondi, aruna, Eric Greenlee

Suzanne Stathatos (suzanne.stathatos@gmail.com)

2021-07-12 18:21:55

Hi Eric! I am just starting the transition from working as a software engineer to becoming a PhD student (with Pietro 🎉 !). The process of finding programs and debating PhD vs. private industry work is very fresh in my mind. While not entirely up to date, this site has a number of researchers in this intersection. Alternatively, in the private industry, the Work On Climate slack channel is a good place to lurk for companies trying to solve climate issues (not conservation-focused, but there could be some overlap in interests).

workonclimate.org

Work on Climate

We are an action-oriented Slack community for people serious about climate work. Find climate jobs. Build climate companies. Find your people.

Original URL: https://workonclimate.org/

🙌 Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

❤️ Eric Greenlee

Eric Greenlee (Eric.greenlee.96@gmail.com)

2021-07-13 09:55:52

Thank you so much @Sara Beery and @Suzanne Stathatos! All your resources are fantastic, and CompSustNet in particular looks promising!

❤️ Sara Beery, Suzanne Stathatos, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Mike C (mike@mikecee.solutions)

2021-07-18 05:28:51

Welcome @Eric Greenlee - I believe @Ben Seleb is doing a graduate project in conservation hardware. Worth giving him a shout.

Also, Shah Selbe is into field IOT hardware. Maybe his work can give you some leads

https://medium.com/conservify/bringing-conservation-technology-to-life-eadd93bfa3af

https://medium.com/conservify

Medium

Bringing Conservation Technology to Life

An abridged version of this was posted to the Virgin Unite blog as part of their 2016 Davos (World Economic Forum) coverage.

Reading time

7 min read

Original URL: https://medium.com/conservify/bringing-conservation-technology-to-life-eadd93bfa3af

Medium

Conservify – Medium

Building open source hardware and software for conservation.

Original URL: https://medium.com/conservify

❤️ Eric Greenlee, Sara Beery

😊 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Eric Greenlee (Eric.greenlee.96@gmail.com)

2021-07-18 08:29:51

Thanks @Mike C! I reached out to Ben and I actually to Shah 6 months ago about FieldKit

Stefan Schneider (sschne01@uoguelph.ca)

2021-07-12 12:16:10

Hi Everyone! Would anyone happen to know of a public/available insect/arthropod camera trap dataset? Something with multiple instances of species/individuals on a semi-standardized background. Akin to something like this:

Sara Beery (sara.beery@gmail.com)

2021-07-12 12:17:36

I don't know of a public one, but I was JUST talking to someone who is interested in this on Friday for Ag applications (detecting invasives)

Stefan Schneider (sschne01@uoguelph.ca)

2021-07-12 12:18:57

yeah! That's pretty much the exact same idea. If something comes through the grapevine I'd love to hear about it

👍 Sara Beery

Sara Beery (sara.beery@gmail.com)

2021-07-12 12:26:18

@Chandra Krintz and @rich wolski this is similar to what we were talking about Friday, right? Do you guys know of any public datasets?

Ben Weinstein (benweinstein2010@gmail.com)

2021-07-12 12:28:34

let me know too, I have collaborators in switzerland working on microscope images.

Ben Weinstein (benweinstein2010@gmail.com)

2021-07-12 12:28:51

I don't anything annotated and ready to share yet, but I'll update (probably be like a year)

Ben Weinstein (benweinstein2010@gmail.com)

2021-07-12 12:29:43

you should email https://www.luerig.net/

luerig.net

Moritz D. Lürig

Original URL: https://www.luerig.net/

👍 Sara Beery, Stefan Schneider

Stefan Schneider (sschne01@uoguelph.ca)

2021-07-12 13:40:24

I'll do that! Thanks 🙂 The dataset doesn't actually have to be labeled

rich wolski (rich@cs.ucsb.edu)

2021-07-12 14:17:33

@Sara Beery @Stefan Schneider @Ben Weinstein Hi All. Sadly, while we know of a research team that is working on insect identification I don't think they have made an image data set public. We will check to see if they have an interest in doing so.

Jonathan Crall (erotemic@gmail.com)

2021-07-13 10:33:41

I know of this honeybee tracking dataset: https://zenodo.org/record/4400651#.YO2kE3VKi54 which corresponds to the paper/code described on this github page: https://github.com/vladan-stojnic/Detection-of-Small-Flying-Objects-in-UAV-Videos

vladan-stojnic/Detection-of-Small-Flying-Objects-in-UAV-Videos

Code for paper "Detection of Flying Honeybees in UAV Videos"

Stars

15

Language

Python

🐝 Stefan Schneider, Jonathan Crall

Stefan Schneider (sschne01@uoguelph.ca)

2021-07-13 10:35:46

thanks for pointing this one out!

Chris Yeh (chrisyeh96@gmail.com)

2021-07-21 01:52:40

Came across this postdoc opportunity which might be interesting for certain members in this Slack: https://earth.stanford.edu/job/two-postdoctoral-positions-stanford-university-remote-sensing-and-georeferencing-global-methane

earth.stanford.edu

Two postdoctoral positions, Stanford University: Remote sensing and georeferencing of global methane-source infrastructure | Stanford Earth

Two postdoctoral positions, Stanford University: Remote sensing and georeferencing of global methane-source infrastructure AI/Computer vision + remote sensing Geospatial data integration

Original URL: https://earth.stanford.edu/job/two-postdoctoral-positions-stanford-university-remote-sensing-and-georeferencing-global-methane

❤️ Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Siyu Yang

🌲 Suzanne Stathatos

Ayan Mukhopadhyay (ayanmukg@gmail.com)

2021-07-24 06:09:50

Hey everyone! Happy to be a part of this community. I wanted to share a project we are doing about conserving monarch butterflies. These tiny creatures travel over 3000 miles every year to their overwintering sites in Mexico (and some head to California). As monarchs stay in extremely dense clusters, it is very difficult to count them. Currently, people simply estimate size of the population every year by looking at the total hectares occupied. Unfortunately, this can be extremely error-prone. We collected images of monarchs and showed how simple crowd counting techniques using deep nets can be used to accurately estimate counts of butterflies in dense clusters. Read the paper here: https://www.biorxiv.org/content/10.1101/2021.07.23.453502v1 Our dataset is open for anyone to use so if you are interested, please reach out. Also, if you know anyone in CA or Mexico who might be interested in trying this out in the field, please let us know and we will be happy to help!

bioRxiv

Using Deep Learning to Count Monarch Butterflies in Dense Clusters

Monarch butterflies display one of the most fascinating migration patterns of all species, traveling over 3000 miles from their North American breeding grounds to reach overwintering sites in Central Mexico. Recent studies have suggested that monarchs have experienced an alarming decline in population size due to a combination of deforestation, loss of native milkweed and nectaring plants, and climate change. An issue that conservation efforts face is the lack of principled mechanisms to accurately estimate and count the population size of monarchs. This difficulty occurs due to their small size and existence in dense overwintering clusters in forests. We create an open-source tool to aid conservationists estimate the count of monarch butterflies from images automatically. To the best of our knowledge, our approach, based on deep convolutional neural networks, is the first automated application that can count small insects like monarch butterflies in dense clusters. We demonstrate that our approach achieves high accuracy in counting the number of butterflies even in the presence of occlusion. We also release an open-source dataset containing high resolution images of monarch butterflies along with human annotations for each butterfly's position. Our open-source implementation can be readily used by scientists to estimate monarch numbers in overwintering clusters. ### Competing Interest Statement The authors have declared no competing interest.

Original URL: https://www.biorxiv.org/content/10.1101/2021.07.23.453502v1

❤️ Sara Beery, Emilio Luz-Ricca, Armin Bazarjani, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Barry Brook, David Russell

👍 Dan Morris, Mikey Tabak

Barry Brook (barry.brook@utas.edu.au)

2021-07-27 07:38:41

@Sara Beery Your new paper on SDMs for ML practitioners (https://arxiv.org/pdf/2107.10400.pdf) is excellent!

❤️ Sara Beery, Kevin Winner, Vienna Saccomanno, Elijah Cole (Deactivated), Cody Kupferschmidt, Arjun Subramonian (they/them), Jason Holmberg (Wild Me)

👏 gvanhorn, Nico Lang, Riccardo de Lutio, Talia Speaker, Ayan Mukhopadhyay, Beckett Sterner, Declan, Emilio Luz-Ricca, Mitch Fennell, Armin Bazarjani, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Mikey Tabak

Sara Beery (sara.beery@gmail.com)

2021-07-27 07:44:02

Thank you!!! This was a joint effort with @Elijah Cole (Deactivated) and was wonderfully advised by @Kevin Winner, I learned a TON while writing it 🙂

❤️ Barry Brook, Kevin Winner

David Rolnick (dsrolnick@gmail.com)

2021-07-28 21:43:03

Do you know about an exciting project involving AI & climate change? Climate Change AI and the Centre for AI & Climate are soliciting case studies for a report by the Global Partnership on AI (GPAI) to be launched alongside COP26. We are most interested in projects that have either been deployed or are aimed towards deployment.

Examples of what case studies might look like:
- A company using AI to increase the energy efficiency of factories.
- An NGO tracking deforestation using AI.
- A research team incorporating AI in climate modeling.
- A governmental agency using AI for disaster response.

Suggest projects by August 12 at: https://www.climatechange.ai/gpai-case-studies

👏 Sara Beery

👍 Carly Batist, Bistra Dilkina

Barry Brook (barry.brook@utas.edu.au)

2021-07-31 21:56:17

@Siyu Yang I enjoyed your recent presentation and Q&A on the MegaDetector! Very useful. https://youtu.be/LUkQVARAVFI

YouTube

WILDLABS.NET (https://www.youtube.com/channel/UCrxw8iiyFalKHFNAhZYCAYA)

Siyu Yang: How do I get started with Megadetector?

Original URL: https://youtu.be/LUkQVARAVFI

❤️ Sara Beery, Olivier Gimenez, Jason Holmberg (Wild Me), Talia Speaker

Siyu Yang (yasiyu@microsoft.com)

2021-08-02 16:14:03

Thanks! Glad people find it helpful

Petar Gyurov (pgyurov93@gmail.com)

2021-08-02 12:23:44

Hi guys. Does anyone where I can find any datasets of polar bear images? Cheers 🐻‍❄️

Alasdair Davies (alasdair@shuttleworthfoundation.org)

2021-08-03 06:49:51

Only thermal from this project (if useful)

Yuval Boss (yuval@yuvalboss.com)

2021-08-09 15:06:11

http://lila.science/datasets/arcticseals and http://lila.science/datasets/noaa-arctic-seals-2019/. I'm not certain if both of these have polar bears, the first one should but unfortunately not many and you would need to download a lot more data to access them. Send me a message and I can package up some polar bear data seperately. Also these are aerial images so no close ups of polar bears. Would love to hear what you are working on if you want to share! 🙂

LILA BC

NOAA Arctic Seals - LILA BC

This data set contains about one million thermal/RGB image pairs, representing a 2016 aerial survey of the Alaksan coastline, conducted by NOAA fisheries.

Written by

lilawp

Est. reading time

2 minutes

Original URL: http://lila.science/datasets/arcticseals

LILA BC

NOAA Arctic Seals 2019 - LILA BC

Written by

lilawp

Est. reading time

2 minutes

Original URL: http://lila.science/datasets/noaa-arctic-seals-2019/

Petar Gyurov (pgyurov93@gmail.com)

2021-08-09 16:40:14

@Yuval Boss Ah, sounds great but unfortunately aerial photos won't do. I am working on face detection and recognition so need some close up shots! Similar work to the BearID project. I did find this dataset which contained ~900 images of polar bears so that's a good starting point 🙂

cvml.ist.ac.at

Animals with Attributes 2

Computer Vision and Machine Learning at IST Austria: curiosity-driven, striving for excellence

Original URL: https://cvml.ist.ac.at/AwA2/

Yuval Boss (yuval@yuvalboss.com)

2021-08-09 17:04:34

Awesome! If you come across any aerial polar bear data don't hesitate to share 🙂 we only have a few hundred examples and could really benefit from more

👍 Petar Gyurov

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2021-08-02 12:58:07

Professor Jane Waterman at the University of Manitoba did some whiskerprint photo ID work for polar bears. May have good imagery.

👍 Sara Beery, Petar Gyurov, Jon Van Oast

Petar Gyurov (pgyurov93@gmail.com)

2021-08-02 14:24:09

Thank you, I will get in touch!

Ben Weinstein (benweinstein2010@gmail.com)

2021-08-02 15:29:21

dominique chabot wrote a paper that I declined to review in the last few months on scraping web images for polar bears, you can ask him dominique.chabot@mail.mcgill.ca

👍 Petar Gyurov, Sara Beery

Remi Gonety (gonetyremi@outlook.com)

2021-08-05 14:33:30

Hey everyone, I am new here 👋. Thanks to @Dan Morris for inviting me. I will be starting a master in data science at the University of Edinburgh. I have a background in environmental engineering, and I am interested in water and conservation. I am excited to contribute and learn from you all.

👍 Oisin Mac Aodha, Benjamin Kellenberger, Jason Holmberg (Wild Me), Sara Beery

👋 Jon Van Oast, Daniel Grzenda, Declan, Jason Holmberg (Wild Me), Sara Beery, Stefan Schneider, Ed Miller, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2021-08-05 15:16:59

Awesome! Welcome to absolutely beautiful old Dunedin! 😄

😁 Remi Gonety

Remi Gonety (gonetyremi@outlook.com)

2021-08-05 19:01:55

thanks

HM van Zyl (senseivzyl@gmail.com)

2021-08-09 13:45:58

Hey all! Thanks @Sara Beery for the invite.

Quick Intro,

Henk van Zyl, from Cape Town, South Africa. I’m the Cofounder of Symbyte, where we help companies deploy useful AI. Cool things we’ve worked on : Active Shooter Detection and Triangulation, and saving youth from a life of Gangsterism. Our bread and butter is mostly MLops and cloud and data engineering.

I'm super interested in this field and would love to volunteer some time if anyone ever needs a grunt to help.

Cheers

👍 Jason Holmberg (Wild Me), Omiros Pantazis, Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Emmanuel Dufourq (edufourq@gmail.com)

2021-08-19 12:24:30

hey @HM van Zyl - also from CT here! Are you interested in collaborations with academics? I've recently joined stellenbosch uni, let me know if you'd be keen and there should be a new cohort of MSc students in engineering looking for projects soon

😍 Sara Beery

HM van Zyl (senseivzyl@gmail.com)

2021-08-19 15:25:28

Hey Emmanuel! I'd love to do a quick meet. I'll pm you my email and number then we can get started! Perhaps someone in the channel could give us an idea of something cheap and dirty we could attempt. I think the Stellenbosch campus is ripe for some bird monitoring! I'll take this over to random.

Emmanuel Dufourq (edufourq@gmail.com)

2021-08-20 03:13:07

I have tons of ideas, literally a whole list of potential MSc topics! 🙂 So yes, I think collaborating with the zoological department at SU and the data science school which I work in would be a great 3-way collaboration. I have a bunch of audiomoths and field recorders. So yes, please do PM me your details

❤️ Sara Beery

Daniel Davila (daniel.davila@kitware.com)

2021-08-09 13:52:54

Hello, I just got back from an excellent talk given by @Sara Beery on biodiversity model robustification (totally a word). Im Dan, I work at Kitware which is an open source scientific computing company with a huge focus on AI/ML systems. I have a background in deployed multiplatform, multisensor CV systems for environmental monitoring. It looks like Jon Crall has been contributing here for a bit, brilliant colleague of mine here at Kitware. Also looks like some of our ecological work has been featured, thanks to @Yuval Boss , we are heavily involved in the NOAA seal detection work among other things. Nice to meet yall!

👋 Sara Beery, Jason Holmberg (Wild Me), HM van Zyl, Omiros Pantazis, Jon Van Oast, Jan Kees, Ed Miller, Talia Speaker, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Frederic Fol Leymarie (ffl@dynaikon.com)

2021-08-10 06:14:14

Hi; I too joined after the excellent interactive presentation by @Sara Beery yesterday. I work with DynAIkon which focuses on AI solutions for the camera trap world. We are part of a large EU consortium which covers different facets of conservation (focused on observations of plants and animals) and citizen science : https://cos4cloud-eosc.eu/

Cos4cloud

home

Sign up and stay tuned! Via our newsletterwe share all our news & events! Subscribe now! Cos4Cloud, a European Horizon 2020 project to boost citizen science technologies We are developing eleven technological services to improve citizen

Original URL: https://cos4cloud-eosc.eu/

👋 Kai Waddington, Sara Beery, Ross Gardiner, Talia Speaker, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Emily Charry Tissier (hello@whaleseeker.com)

2021-08-12 14:59:58

Hello! Our group (Whale Seeker) detects marine mammals from imagery using AI. We are trying to create educative and thoughtful content about AI so non-experts can better understand the how and why. We frequently come across the attitude from possible clients that “AI is too complicated and expensive so we’ll substitute crowdsourcing/citizen science instead.” We wrote a blog on this topic that we’d love your thoughts on if you have time. https://www.whaleseeker.com/blog/widening-the-bottleneck-can-citizen-science-accelerate-conservation. Thank you in advance!

Whale seeker

Widening the bottleneck: can citizen science accelerate conservation? — Whale Seeker

Ecological monitoring is a key part of any conservation work : if you don’t know what’s out there, you can’t protect it. But that’s easier said than done: using traditional survey methods, this work is costly, logistically challenging, slow and limited in its geographic scale. In the last couple of

Original URL: https://www.whaleseeker.com/blog/widening-the-bottleneck-can-citizen-science-accelerate-conservation

❤️ Sara Beery, aruna, Carly Batist, Malcolm Kennedy, Rebekah Loving

😎 Jon Van Oast, Mitch Fennell, Omiros Pantazis

👏 Ștefan Istrate, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

🐳 Cameron Trotter, Mike C

👀 Cameron Trotter

Mike C (mike@mikecee.solutions)

2021-08-13 12:33:52

Love your blog posts. I would love to share a summary version linking to the full article (backlinked and attributed to you) on the Open Ocean Camera Medium publication - can we get permission to? 🙂 @Emily Charry Tissier

👍 Emily Charry Tissier, Malcolm Kennedy

Emily Charry Tissier (hello@whaleseeker.com)

2021-08-13 12:39:31

Absolutely!

👍 Malcolm Kennedy

Mike C (mike@mikecee.solutions)

2021-08-13 12:34:48

Stumbled upon the ARM AI tools catalogue while browsing ConservationX. Might be a resource for those looking for tools, and an advertising channel for those looking to spread word about their solutions?

https://www.arm.com/why-arm/partner-ecosystem/ai-ecosystem-catalog/conservation-x-labs

Arm | The Architecture for the Digital World

Conservation X Labs

Conservation X Labs addresses human-induced drivers of wildlife extinction through the development of breakthrough technologies.

Original URL: https://www.arm.com/why-arm/partner-ecosystem/ai-ecosystem-catalog/conservation-x-labs

👍 Ed Miller

Ed Miller (ed@hypraptive.com)

2021-08-13 20:48:36

I work for Arm. Feel free to message me if you have any questions!

❤️ Jason Holmberg (Wild Me)

Caleb Robinson (calebrob6@gmail.com)

2021-08-14 22:44:27

Awesome to see the work on tree crown detection from @Rebekah Loving @Arushi Agarwal at KDDs Fragile Earth workshop! ("A network fusion model pipeline for multi-modal, deep learning for tree crown detection" - https://ai4good.org/fragile-earth-2021/)

🎉 Jon Van Oast, Elijah Cole (Deactivated), Ben Weinstein, Justin Kay, Aaron Ferber, Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Björn Lütjens

😍 Sara Beery

Marc Grimson (mg2425@cornell.edu)

2021-08-15 17:23:56

Hi everyone! I'm relatively new here to the Slack and recently started my PhD in CS and I'm really excited to work on problems in Sustainability and Conservation. I'm coming from a number of years as a software engineer where unfortunately there wasn't much ML. I'm hoping to improve on my ML knowledge, and was wondering if anyone knew of any good intermediate level online resources for ML, or if there were any interesting datasets in Conservation that would be relatively easy to play around with and develop models on for practice. Thanks in advance!

👋 Sara Beery, Nico Franz, Chris Yeh, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Björn Lütjens

Sara Beery (sara.beery@gmail.com)

2021-08-16 10:40:39

There are a lot of accessible datasets/challenges on kaggle via the FGVC workshop competitions! https://sites.google.com/view/fgvc8/competitions

sites.google.com

FGVC8 - Competitions

GeoLifeCLEF2021 Predict the presence of species using observations paired with aerial images and environmental features. <https://www.kaggle.com/c/geolifeclef-2021>

Original URL: https://sites.google.com/view/fgvc8/competitions

George Ore (gore@caltech.edu)

2021-08-17 01:30:19

Hello everyone, I am an incoming college freshman assigned to check out your cool work. I hope I can learn a lot from everyone and get some essential skills under my belt.

👋 Declan, Sara Beery, Suzanne Stathatos, Alex Borowicz, Carly Batist, Emmanuel Dufourq, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-08-20 13:22:50

Having wished for a long time that there was a central database-type thing for #tech4wildlife, @Gracie Ermi & I created it ourselves! Here’s a directory of conservation tech org’s, companies, collabs, projects, etc. so the community has a centralized place to search for resources --> https://sites.google.com/view/conservation-tech-directory.

We’ve posted more info on Twitter if you want to check that out. There’s a link there to a form where you can suggest a resource if you know of something that belongs here but currently isn’t! Hope it can be a useful resource for you all. If you have any questions or suggestions, do feel free to reach out to us 🙂

sites.google.com

A list of organizations, communities, companies, open-source projects, and other resources within the conservation technology space.

Original URL: https://sites.google.com/view/conservation-tech-directory

😎 Jon Van Oast, Thijs, Howard L Frederick, HM van Zyl, Ted Schmitt, Jason Holmberg (Wild Me), Ritwik, Björn Lütjens

🎉 Gracie Ermi, aruna, Emilio Luz-Ricca, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, HM van Zyl, Jason Holmberg (Wild Me), Talia Speaker, Tarun, Rebekah Loving, Björn Lütjens

❤️ Justin Kay, Howard L Frederick, HM van Zyl, Jason Holmberg (Wild Me), Ankita Shukla, Suzanne Stathatos, Eric Greenlee, Yuval Boss

Monty Ammar (montyx23@gmail.com)

2021-12-21 17:34:07

Hey Carly this sounds brilliant! The link does not work for me for some reason though. Wondering if the link is deprecated or whether there is a new one at all? Many thanks

Gracie Ermi (gracieermiifthen@gmail.com)

2021-12-21 17:47:38

Hi @Monty Ammar! The directory now lives at https://conservationtech.directory

conservationtech.directory

Deep Chimpact: Depth Estimation for Wildlife Conservation

Visit the post for more.

Original URL: https://conservationtech.directory

👍 Carly Batist

Monty Ammar (montyx23@gmail.com)

2021-12-21 18:45:10

Hey, thank you very much! ☺️

❤️ Carly Batist

Sara Beery (sara.beery@gmail.com)

2021-09-09 15:33:39

Awesome!! @Oisin Mac Aodha 🙂

Sara Beery (sara.beery@gmail.com)

2021-09-09 15:37:00

@Greg Lipstein how was the ground truth collected?

Sara Beery (sara.beery@gmail.com)

2021-09-09 15:38:52

Oh nevermind, I found it 🙂 https://www.drivendata.org/competitions/82/competition-wildlife-video-depth-estimation/page/392/#about-the-data

DrivenData

Help conservationists monitor wildlife populations. In this challenge, you will estimate the depth of animals in camera trap videos. Better distance estimation models can rapidly accelerate the availability of critical information for wildlife monitoring and conservation.

Original URL: https://www.drivendata.org/competitions/82/competition-wildlife-video-depth-estimation/page/392/#about-the-data

👍 Greg Lipstein

Greg Lipstein (greg@drivendata.org)

2021-09-09 15:38:58

Thanks @Sara Beery! Check out About the Data section here https://www.drivendata.org/competitions/82/competition-wildlife-video-depth-estimation/page/392/

DrivenData

Deep Chimpact: Depth Estimation for Wildlife Conservation

Original URL: https://www.drivendata.org/competitions/82/competition-wildlife-video-depth-estimation/page/392/

Greg Lipstein (greg@drivendata.org)

2021-09-09 15:39:03

jinx

😁 Sara Beery, Jon Van Oast

Sara Beery (sara.beery@gmail.com)

2021-09-09 15:39:13

So, along those lines, any sense of what the accuracy in this manual labeling is?

Sara Beery (sara.beery@gmail.com)

2021-09-09 15:39:54

@Elijah Cole (Deactivated) and I captured a similar thing for some of our cameras, but I got the sense that there was a lot of room for human error/subjectivity to creep in.

Greg Lipstein (greg@drivendata.org)

2021-09-09 15:50:10

That would be a good question for the MPI-EVA team. From what we’ve heard, that's right; it's often a time-consuming, manual, and error-prone process to attach these distance labels. That said, some noise in labels is to be expected, and the difficulty labeling provides some of the value from having ML solutions that can help. The scale of the data also helps here, especially to the extent the noise is random

Sara Beery (sara.beery@gmail.com)

2021-09-09 15:54:27

Absolutely! I'm curious if there's an upper bound on performance based on label error, and/or if the error is biased for a given labeler (ie maybe I'm likely to always label closer than reality, for example)

Sara Beery (sara.beery@gmail.com)

2021-09-09 15:56:02

Another interesting thing to look at would be agreement between labelers, did multiple experts label each image?

Sara Beery (sara.beery@gmail.com)

2021-09-09 15:56:26

(I realize these are not necessarily things you have answers to, just thinking out loud 🙂 )

Oisin Mac Aodha (macaodha@caltech.edu)

2021-09-09 15:57:36

Looks cool.

Greg Lipstein (greg@drivendata.org)

2021-09-09 16:02:51

I realize these are not necessarily things you have answers to, just thinking out loud 🙂

Totally! Those are all good thoughts

Oisin Mac Aodha (macaodha@caltech.edu)

2021-09-09 16:57:38

@Greg Lipstein I see that you tried our monodepth2 code in the example image - very cool. Have you tried Midas, it is trained on much more diverse data?
https://github.com/isl-org/MiDaS

❤️ Sara Beery

Greg Lipstein (greg@drivendata.org)

2021-09-09 17:20:13

We did! I haven't seen Midas but I'm also not the right one to ask simple_smile . Much better to check with our (fantastic) data scientist @Emily Dorne

Emily Dorne (emily@drivendata.org)

2021-09-09 17:33:32

@Oisin Mac Aodha I have seen the paper but haven't yet tried out the code! Very cool to see transformers being used in this space

👍 Oisin Mac Aodha

Ștefan Istrate (stefan.istrate@gmail.com)

2021-10-11 08:36:34

Hi @Greg Lipstein! What does the ground truth distance mean when multiple animals are present in the image? I couldn't find any reference in the resources listed on DrivenData.

Katie Wetstone (she, her) (katie@drivendata.org)

2021-10-11 12:26:38

@Ștefan Istrate Wherever possible, frames with multiple animals present have been excluded from the labels that participants have to predict. However, the hand-labeled ground truth is not 100% comprehensive and there are some unlabeled animals. We used the megadetector model to filter out any frames with multiple high-probability bounding boxes, but the predictions are not perfect and a small number may have slipped through - looks like you found a good example of this. The estimated bounding box given is the one with the highest probability, but may not correspond exactly to the label. These cases should be very rare in the data!

👍 Sara Beery

Ștefan Istrate (stefan.istrate@gmail.com)

2021-10-11 12:29:39

Makes sense, thanks!

Katie Wetstone (she, her) (katie@drivendata.org)

2021-10-11 12:30:10

happy to help!

Sara Beery (sara.beery@gmail.com)

2021-09-10 08:03:18

IJCV Special Issue focusing on Animal Tracking and Modeling.

https://www.springer.com/journal/11263/updates/19611514?gclid=Cj0KCQjw4eaJBhDMARIsANhrQADTS1NAfl_1ZXtPL12_n_oP3x7tzZTV6Ognaw7At_BKmmogofes61waAvY6EALw_wcB

Springer

International Journal of Computer Vision

Original URL: https://www.springer.com/journal/11263/updates/19611514?gclid=Cj0KCQjw4eaJBhDMARIsANhrQADTS1NAfl_1ZXtPL12_n_oP3x7tzZTV6Ognaw7At_BKmmogofes61waAvY6EALw_wcB

👍 Benjamin Kellenberger, Emily Charry Tissier, Chris Yeh, Jason Holmberg (Wild Me)

😎 Jon Van Oast

Petar Gyurov (pgyurov93@gmail.com)

2021-09-13 06:42:28

Anyone know of any libraries/software that can help with identifying duplicate images? I want to make sure my training and test sets don't share any images. A quick search returned imgdiff but curious if there's anything purpose built. Cheers

Daniel Davila (daniel.davila@kitware.com)

2021-09-13 10:48:06

My group is working on this problem a bit, would be happy to share if interested. There are a few interesting papers in this area as well:

Semantic Redundancies in Image-Classification Datasets: The 10% You Don’t Need
https://arxiv.org/pdf/1901.11409.pdf
A critical look at the current train/test split in machine learning
https://arxiv.org/pdf/2106.04525.pdf
Learning From Less Data: A Unified Data Subset Selection and Active Learning Framework for Computer Vision
https://arxiv.org/pdf/1901.01151.pdf
Do we train on test data? Purging CIFAR of near-duplicates
https://arxiv.org/pdf/1902.00423.pdf
Apple Neural Hash (CASM) Report
https://www.apple.com/child-safety/pdf/CSAM_Detection_Technical_Summary.pdf

Also a few commercial options:

https://www.lightly.ai/post/how-redundant-is-your-dataset

👍 Justin Kay, Petar Gyurov

Petar Gyurov (pgyurov93@gmail.com)

2021-09-13 11:18:15

Thanks! Lots of good stuff in there. I thought about using Apple's Neural Hash for this... I wonder how effective it would be.
Do share your work! I am interested to learn more about this.

aruna (arunas@mit.edu)

2021-09-13 11:31:15

I have used the perceptual hash for this, with decent though not 100% perfect results: https://github.com/knjcode/imgdupes

knjcode/imgdupes

Finding and deleting near-duplicate images based on perceptual hash.

Stars

193

Language

Python

👍 Petar Gyurov

Petar Gyurov (pgyurov93@gmail.com)

2021-09-13 12:42:32

That looks great, thanks! Will give it a go.

Daniel Davila (daniel.davila@kitware.com)

2021-09-13 12:56:33

When you say "share" images between train/test, what exactly do you mean? It's a really interesting question because there is a big difference between finding exact duplicates, near duplicates (smooth augmentations such as color shifts, translations, crops, rotations...), and finding similar images which are not reachable at all in augmentation space all but are so semantically similar that you would never want to both train/test on them.

Petar Gyurov (pgyurov93@gmail.com)

2021-09-13 13:06:49

I was initially only considering exact duplicates but now you've opened my eyes! I'd love to get a measure of near duplicates and similiar images but I wouldn't consider it a priority for what I am doing just yet.

Caleb Powell (cpowel21@asu.edu)

2021-09-13 14:51:38

My first instinct would be to hash them. I've never used it, but this might be worth checking out:
https://github.com/JohannesBuchner/imagehash

JohannesBuchner/imagehash

A Python Perceptual Image Hashing Module

Stars

2092

Language

Python

👍 Petar Gyurov, Yuval Boss

Suhail Alnahari (alnah005@umn.edu)

2021-09-14 13:56:13

If you're using image hashes, specifically apple's neural hash, be aware of this https://blog.roboflow.com/neuralhash-collision/

Roboflow Blog

ImageNet contains naturally occurring NeuralHash collisions

NeuralHash is the perceptual hashing model that back's Apple's new CSAM (child sexual abuse material) reporting mechanism. It's an algorithm that takes an image as input and returns a 96-bit unique identifier (a hash) that should match for two images that are "the same" (besides some minor perturbations like JPEG

Written by

Brad Dwyer

Filed under

News

Original URL: https://blog.roboflow.com/neuralhash-collision/

😅 Caleb Powell

David Rolnick (dsrolnick@gmail.com)

2021-09-15 20:42:49

Focus issue from AIHub on "Life on Land": https://aihub.org/2021/09/10/focus-on-life-on-land-call-for-contributions/

AIhub.org connecting the AI community and the world. - Association for the Understanding of Artificial Intelligence

Focus on life on land: call for contributions - AIhub.org connecting the AI community and the world. - Association for the Understanding of Artificial Intelligence

Original URL: https://aihub.org/2021/09/10/focus-on-life-on-land-call-for-contributions/

👍 Sara Beery, Jason Holmberg (Wild Me)

😎 Jon Van Oast, Jason Holmberg (Wild Me)

Sara Beery (sara.beery@gmail.com)

2021-09-17 10:21:25

We are thrilled to announce the call for applicants for the first annual Resnick Sustainability Institute Summer School on Computer Vision Methods for Ecology (https://cv4ecology.caltech.edu/ ). This intensive, three-week program will teach applied computer vision methods to senior ecology graduate students and postdocs, and will be hosted at Caltech from August 1-20, 2022. Students will develop hands-on computer vision systems to help answer their own ecological research questions, using their own data. They will receive daily mentorship from a passionate team of computer vision experts with a track record of impact in conservation and sustainability. Each student will be provided with $2500 in cloud credits to facilitate their project development sponsored by Microsoft AI for Earth and Amazon AWS. Our team of instructors will work with applicants leading up to the intensive to curate computer-vision-ready labels for their data that will be used to prototype systems for their research questions during the class. Students will leave the course empowered to build their own computer vision models for ecological applications, and gain skills in problem formulation, dataset curation, model training, model evaluation, and hosting models for inference. Acceptance to the 2022 school will be competitive, we plan to accept only 20 students who will be evaluated on their past work as well as their proposed projects. Applications are due December 1st, and we strongly encourage applicants from minoritized groups in academia.

https://twitter.com/cv4ecology/status/1438867424078557187

👍 Benjamin Kellenberger, Jason Holmberg (Wild Me), Riccardo de Lutio, Carly Batist, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Nico Franz, Ted Schmitt, Stefan Schneider, Casey Youngflesh, Ayan Mukhopadhyay, Chris Yeh, David, Björn Lütjens

🎉 Suzanne Stathatos, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Oisin Mac Aodha, Nico Franz, Omiros Pantazis, Gracie Ermi, Carly Batist, Stefan Schneider, Mitch Fennell, Olivier Gimenez, David, Talia Speaker

❤️ Jon Van Oast, Ben Weinstein, Justin Kay, Stefan Schneider, Hannah Yin, Mitch Fennell

Sara Beery (sara.beery@gmail.com)

2021-09-17 10:21:50

Please share widely!!

👍 Jon Van Oast

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2021-09-17 11:03:12

this is a wonderful initiative, Sara! so glad that you're doing this!

❤️ Sara Beery

Gracie Ermi (gracieermiifthen@gmail.com)

2021-09-17 11:37:23

This is so awesome, Sara! Will definitely spread the word!

❤️ Sara Beery

Jon Van Oast (jon@wildme.org)

2021-09-17 11:53:37

this is really great!

❤️ Sara Beery

Casey Youngflesh (caseyyoungflesh@gmail.com)

2021-09-17 13:48:26

So awesome!!

❤️ Sara Beery

Sara Beery (sara.beery@gmail.com)

2021-09-22 09:47:29

"The annual LifeCLEF workshop will take place tomorrow by videoconference during the CLEF 2021 conference.
As every year, we will present the main results of the challenges organized and we will also have 3 invited speakers who will talk more broadly about the application of AI for citizen science, biodiversity and land management. The detailed programme of the workshop and the links to zoom videoconferences can be found on LifeCLEF website:
https://www.imageclef.org/LifeCLEF2021"

👍 Oisin Mac Aodha, Ștefan Istrate, Justin Kay

:thumbsup_all: Frederic Fol Leymarie

😎 Jon Van Oast

Sachith Seneviratne (sachith.seneviratne@unimelb.edu.au)

2021-09-23 10:53:44

Hello everyone, I participated in one the LifeCLEF challenges and found my way here after Sara's very interesting talk. Greetings from down under!

👋 Stefan Schneider, Benjamin Kellenberger, Caleb Powell, Jon Van Oast, Sara Beery, Ben Weinstein, Elijah Cole (Deactivated), Jason Holmberg (Wild Me)

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-09-24 11:41:18

@Gracie Ermi and I just migrated our Conservation Tech Directory site to a new platform that is more user-friendly and aesthetic. It has a snazzy link title now too! - conservationtech.directory. Check it out!

This update also comes with ~30 new entries as well, bringing the total count to 447 now. As before, use this Google form to submit resources that should be on here but aren’t yet. And downloadable PDF of the directory can be found on FigShare.

conservationtech.directory

Forest Monitoring, Land Use & Deforestation Trends | Global Forest Watch

Visit the post for more.

Original URL: https://conservationtech.directory/

🙌 Jes Lefcourt, Jason Holmberg (Wild Me), Jon Van Oast, Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Talia Speaker, Ștefan Istrate, Suzanne Stathatos, Mitch Fennell, Megan Cromp, Kirk Larsen, Petar Gyurov

🙌:skin_tone_3: Hannah Yin

Gracie Ermi (gracieermiifthen@gmail.com)

2021-09-24 13:29:17

Also, please feel free to use the google form to submit corrections to resources that are already in the list. We want this to be as accurate as possible, so we’re all ears if you have a correction or a piece of info to fill in!

💯 Carly Batist, Jason Holmberg (Wild Me), Ted Schmitt, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Chris Yeh (chrisyeh96@gmail.com)

2021-09-28 16:28:20

Does anyone know any high-quality ML-friendly datasets for any of the following:

measuring carbon stocks of forests

counting / identifying solar PV and/or wind farm installation

mapping CO2, methane, and/or other GHG emissions

estimating local-level industrial carbon intensity

detection / classification of deforestation activity

By "ML-friendly," I mean that the dataset must have well-defined train/test splits and evaluation metrics. Basically, I'm looking for examples of good ML datasets for monitoring indicators that are commonly associated with climate change. Many thanks in advance!

👍 Björn Lütjens, Sara Beery

Sara Beery (sara.beery@gmail.com)

2021-09-28 16:28:55

@Björn Lütjens @Ben Weinstein

👍 Björn Lütjens

Ben Weinstein (benweinstein2010@gmail.com)

2021-09-28 16:31:38

I do not know of any ML ready datasets for these metrics.

Ben Weinstein (benweinstein2010@gmail.com)

2021-09-28 16:32:01

I could recommend about how one could be created, but I would be surprised if they existed.

Ben Weinstein (benweinstein2010@gmail.com)

2021-09-28 16:33:44

there was a student from geohackweek I taught at UW who was using images from industrial traffic from west africa to try to estimate carbon intensity and emissions. I could find her name.

Björn Lütjens (bjoern.luetjens@gmail.com)

2021-09-28 16:47:15

Hi Chris! Agreed with Ben - there doesn't exist an ML ready carbon dataset ,but some people are working on it. This community has assembled a GitHub list on processed and raw datasets in forests which be helpful for you: https://github.com/blutjens/awesome-forests

If you're just looking for nice climate-tangent ML ready datasets ,the Kaggle datasets in this list would be the perfect fit for you. Lmk if that helps

blutjens/awesome-forests

🌳 A curated list of ground-truth forest datasets for the machine learning and forestry community.

Stars

Last updated

a month ago

Daniel Davila (daniel.davila@kitware.com)

2021-09-28 16:49:14

Which scale are we talking about? Handheld cameras or satellites?

Chris Yeh (chrisyeh96@gmail.com)

2021-09-28 16:58:18

@Ben Weinstein: Thanks for confirming my hunch that such datasets are hard to come by.

@Björn Lütjens: Thanks for that GitHub link! A couple of the deforestation tracking ones look relevant.

@Daniel Davila: either!

Daniel Davila (daniel.davila@kitware.com)

2021-09-28 17:04:43

For emissions in particular, I can think of a few leads... all commercial companies though. There are a few groups working on co2 and methane plume detection. The two leaders I am familiar with (having worked for the former) are SwRI with SLED and FLIR systems with their OGI line of cameras. They both have extensive methane plume datasets, but whether they would allow you access to them is a different story. There are some air/space-borne flavors of this tech offered by other (mostly proprietary) sources like GHGSat and the EDF's upcoming MethaneSAT mission, which we'll have to see if they make that data public and annotated.

Chris Yeh (chrisyeh96@gmail.com)

2021-09-28 17:08:00

@Daniel Davila: Yeah, I'm aware that many commercial companies working on this space. Just to confirm - you aren't aware of any publicly + freely accessible ML datasets for emissions monitoring and/or plume detection, right?

Daniel Davila (daniel.davila@kitware.com)

2021-09-28 17:08:30

No, sorry! Somebody should fund the creation of one

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-09-28 18:24:42

Global Forest Watch tracks deforestation through ML analyses of geospatial data. I’m not sure if the actual training data is public but you might try reaching out the them? If anything the folks at WRI (makers of GFW) might have leads on the types of datasets you’re looking for.

globalforestwatch.org

Global Forest Watch offers free, real-time data, technology and tools for monitoring the world’s forests, enabling better protection against illegal deforestation and unsustainable practices.

Original URL: https://www.globalforestwatch.org/

Ben Weinstein (benweinstein2010@gmail.com)

2021-09-28 18:48:58

@John Brandt is connected to this project.

Nico Lang (nila@di.ku.dk)

2021-09-30 07:28:48

Hi Chris,

ESA’s AI4EO initiative organised its first challenge to improve emissions monitoring. It was a super-resolution task. The challenge is already closed, but maybe you can find the dataset somewhere or contact the organisers.
https://ai4eo.eu/ai4eo-challenge1
https://platform.ai4eo.eu/air-quality-and-health/data

AIREO is another ESA initiative with the aim of providing “AI-ready” datasets. I think the initiative is still in its early stages. Therefore, the prototype datasets currently available are rather small and not really suitable for training deep models.
https://www.aireo.net/

cheers

👍 Chris Yeh

Ben Weinstein (benweinstein2010@gmail.com)

2021-09-30 12:40:00

just clarifying here esa -> is european space agency, and not the ecological society of america. confused me.

👍 Nico Lang

Chris Yeh (chrisyeh96@gmail.com)

2021-10-02 20:53:02

Thanks @Carly Batist - I am aware of Global Forest Watch. However, yeah, I'm concerned that the publicly accessible maps are not the original "ground truth." I may consider reaching out to WRI.

@Nico Lang - Thanks for sharing the AI4EO and AIREO datasets. Will take a look.

Sara Beery (sara.beery@gmail.com)

2021-10-04 17:56:47

Relevant to the above thread: https://www.frontiersin.org/research-topics/26080/forest-carbon-monitoring-and-artificial-intelligence

Frontiers

Forest Carbon Monitoring and Artificial Intelligence

Forest carbon monitoring is the collection and archiving of all data such as land use and land cover change, forest inventory, carbon stocks, carbon emission factors, and other related data necessary for measuring, reporting, and verifying carbon emissions or sources, emission reductions, carbon removals or sinks, within the targeted locations at any scale temporarily and spatially. As part of the climate change mitigation measures, for achieving the target of the Paris Agreement, a key focus is on reducing emissions from; deforestation, forest degradation, conservation of forests, sustainable management of forests, and enhancement of forest carbon stocks. This is collectively known as REDD+ of the United Nations Framework Convention on Climate Change and is a result-based payment for developing countries, where the implementation of the REDD+ activities results in carbon emission reductions and/or carbon removals. Therefore, forest carbon monitoring plays a vital role in ensuring that carbon emissions or removals can be measured, reported, and verified. This Research Topic focuses on applications of AI, machine learning, and deep learning for solving the forest carbon monitoring, measurement, reporting, and verification challenges in various disciplines such as land cover, land-use change, illegal logging, forest degradation, deforestation, and forest restoration. This collection also considers papers on data-driven approaches and the development of online applications f...

Original URL: https://www.frontiersin.org/research-topics/26080/forest-carbon-monitoring-and-artificial-intelligence

👍 Nico Lang, David, Jan Kees, Chris Yeh

Stuart Neilon (stuartneilon@gmail.com)

2021-10-13 09:50:23

Hello Everyone,

Stuart Neilon (stuartneilon@gmail.com)

2021-10-13 09:50:58

Does anyone know of a method to track individuals within a matrix of camera traps?

Sara Beery (sara.beery@gmail.com)

2021-10-13 09:53:44

Do you mean re-identify specific animals across images seen at different camera traps in a grid?

Stuart Neilon (stuartneilon@gmail.com)

2021-10-13 10:02:57

Yes. I apologise for the vague question. I am looking for a method of identifying indivduals of a particular species, that have been previously detected(by megadetector), to map their movements.

Sara Beery (sara.beery@gmail.com)

2021-10-13 10:04:39

What species? There are methods that work well for certain types of patterned species (see https://www.wildme.org/#/ and https://sites.google.com/corp/view/wacv2020animalreid/home)

sites.google.com

WACV2020 AI for Animal Re-ID

Summary from Deep Learning for Animal Re-Identification Workshop We'd like to thank everyone who presented and made it out to the WACV 2020 Animal Re-ID Workshop. We've collected the talks below for your future endevours.

Original URL: https://sites.google.com/corp/view/wacv2020animalreid/home

Sara Beery (sara.beery@gmail.com)

2021-10-13 10:05:22

@Jason Parham, @Maxime Vidal and @Stefan Schneider are the experts :) The first challenge is curating ground truth in order to robustly evaluate one of these methods on your specific data/population

Stuart Neilon (stuartneilon@gmail.com)

2021-10-13 10:08:46

The family of Suidae is of interest, with Warthog being the most important. They do lack patterns, but their facial features(eg. tusk length) can sometimes be used to identify individuals.

Sara Beery (sara.beery@gmail.com)

2021-10-13 10:10:30

Cool. I'm working with elephants currently, and we're exploring tusk length and shape as a contributing identifier but it's still work in progress. Is this something humans are able to do reasonably well for warthogs?

Stuart Neilon (stuartneilon@gmail.com)

2021-10-13 10:19:38

I am no expert in identifying Warthog, but the size of the tusk and overall body size can give a resonable indication of age, which can help in identifying individuals.

Stuart Neilon (stuartneilon@gmail.com)

2021-10-13 10:39:29

Would it be feasible to train from a checkpoint(efficientnetb0-7). If it is feasable, how many images would be required, per individual, to achieve reasonable results.

Stefan Schneider (sschne01@uoguelph.ca)

2021-10-13 11:07:22

Hey Stuart. Here's a paper I wrote that gives a 1000ft view of the field. There may be one or two techniques that standout to you from here.

https://besjournals.onlinelibrary.wiley.com/doi/epdf/10.1111/2041-210X.13133

If you're interested in state-of-the-art, you have a question of if the population is closed vs open. If you have a closed population where you can get images of every individual, you can use a traditional classifier (Conv Neural Network kind of thing). If the population is open, you'll need something like a similarity comparison network.

https://openaccess.thecvf.com/content_WACVW_2020/papers/w2/Schneider_Similarity_Learnin[…]vidual_Re-Identification_-_Beyond_the_WACVW_2020_paper.pdf

I have a version of this paper catered towards ecologists coming this Winter sometime. So if this one is a little too abstract, stay tuned for that.

👍 Sara Beery, Stuart Neilon, Mitch Fennell

Stuart Neilon (stuartneilon@gmail.com)

2021-10-13 11:31:22

Awesome, thank you for your help @Sara Beery @Stefan Schneider !

Stuart Neilon (stuartneilon@gmail.com)

2021-10-13 12:46:37

@Stefan Schneider Hi Stefan, in your paper it states that ideally, the dataset for training a re-ID system(similarity comparison), >500 individuals are needed. Would a triplet loss network still be advantageous over a CNN, in a smaller population(<100) therefore containing a higher number of images per individual.

Stefan Schneider (sschne01@uoguelph.ca)

2021-10-13 12:54:00

Just so we're on the same terms, triplet-loss describes the loss used by a similarity comparison network, which is a form of Metric Learning. You could train a similarity comparison network using a number of different losses, triplet and contrastive are the two I use in the paper.

CNN and similarity comparison networks need approximately the same number of images. The advantage of similarity comparison networks is their accuracy will transfer to previously unseen individuals, and you only need ~500 images of a few individuals, whereas a CNN ideally has 500 images for every individual and will erroneously classify previously unseen individuals as one with data

500 is a rough ballpark number. You could get away with less if you go heavy on augmentations.

If animal re-ID is what you require, best bet would be to get started. Try training a similarity comparison network using a contrastive loss (it's easier) and just see what accuracies you get. You can go from there to determine if you need additional data, more augmentation, use of the triplet-loss, etc. If you've never trained a machine learning model before, start with a basic CNN and build from there. You'll need the CNN anyways for the similarity comparison network

🙌 Stuart Neilon

Stuart Neilon (stuartneilon@gmail.com)

2021-10-13 13:02:48

Thank you very much, that is very informative.

Alina Zare (azare@ufl.edu)

2021-10-13 10:24:22

Hi Everyone - I am new to this Slack group and would just like to introduce myself. My name is Alina Zare. I am Professor in Electrical and Computer Engineering at the University of Florida. I lead the Machine Learning and Sensing lab at UF where we develop new AI/ML approaches with application to agriculture, ecology, and plant science (among other areas) - more here: https://faculty.eng.ufl.edu/machine-learning/. Looking forward to meeting/interacting with ya’ll.

👋 Sara Beery, Omiros Pantazis, Stuart Neilon, Declan, Lukas Picek, Elijah Cole (Deactivated), Jason Holmberg (Wild Me), Chris Yeh

👍 Lukas Picek, Jason Holmberg (Wild Me)

Scott Hosking (jshosking@gmail.com)

2021-10-13 11:32:10

The Alan Turing Institute are collaborating with CEFAS to run a two-phase Data Study Group (DSG) to explore and build toolkits for rapid identification of plankton using machine learning

Please feel free to pass on to your networks 😊

https://www.turing.ac.uk/events/data-study-group-november-2021

The Alan Turing Institute

Data Study Group – November – 2021

Original URL: https://www.turing.ac.uk/events/data-study-group-november-2021

🙌 Sara Beery, Elijah Cole (Deactivated), Oisin Mac Aodha

😎 Jon Van Oast

Sara Beery (sara.beery@gmail.com)

2021-10-14 13:50:33

NSF call on Biodiversity on a Changing Planet: https://beta.nsf.gov/funding/opportunities/biodiversity-changing-planet-bocp

Beta site for NSF - National Science Foundation

Biodiversity on a Changing Planet (BoCP)

Original URL: https://beta.nsf.gov/funding/opportunities/biodiversity-changing-planet-bocp

😎 Jon Van Oast

🤑 Carly Batist, Casey Youngflesh, Amrita Gupta

👍 Vincent Landau, Justin Kay, Bistra Dilkina

Diego Marcos (diego.marcos.gonzalez@gmail.com)

2021-10-15 05:24:40

This call also opened recently (preproposals due on Nov 5th): https://www.biodiversa.org/1772

👍 Sara Beery

Benno Simmons (benno.simmons@gmail.com)

2021-10-18 12:22:58

Hi all! I’m a Lecturer in Ecological Data Science at the University of Exeter in the UK. I’m interested in lots of different things at the AI/ecology/conservation interface, but especially remote sensing of forests, ecological networks and camera traps. I’m putting together a small grant application for a 3 month project on supplementing the training datasets of species identification algorithms for camera traps with data augmentation methods, synthetic imagery and transfer learning. I’m aware of this (https://arxiv.org/abs/1904.05916) awesome paper by @Sara Beery. But are there any other examples? Don’t want to duplicate effort

arXiv.org

Synthetic Examples Improve Generalization for Rare Classes

The ability to detect and classify rare occurrences in images has important applications - for example, counting rare and endangered species when studying biodiversity, or detecting infrequent...

Original URL: https://arxiv.org/abs/1904.05916

❤️ Sara Beery

Sara Beery (sara.beery@gmail.com)

2021-10-18 12:25:47

I did a follow up with a student at TU Delft where we used image-to-image translation to improve on the results with the generic synthetic data: https://arxiv.org/abs/2106.12212

arXiv.org

Image-to-Image Translation of Synthetic Samples for Rare Classes

The natural world is long-tailed: rare classes are observed orders of magnitudes less frequently than common ones, leading to highly-imbalanced data where rare classes can have only handfuls of...

Original URL: https://arxiv.org/abs/2106.12212

👍 Benno Simmons

Ben Weinstein (benweinstein2010@gmail.com)

2021-10-18 12:39:02

talks with @Sara Beery inspired a tree crown paper that had a similar vibe. using LiDAR to create weak labels for RGB learning. https://www.mdpi.com/2072-4292/11/11/1309

MDPI

Individual Tree-Crown Detection in RGB Imagery Using Semi-Supervised Deep Learning Neural Networks

Remote sensing can transform the speed, scale, and cost of biodiversity and forestry surveys. Data acquisition currently outpaces the ability to identify individual organisms in high resolution imagery. We outline an approach for identifying tree-crowns in RGB imagery while using a semi-supervised deep learning detection network. Individual crown delineation has been a long-standing challenge in remote sensing and available algorithms produce mixed results. We show that deep learning models can leverage existing Light Detection and Ranging (LIDAR)-based unsupervised delineation to generate trees that are used for training an initial RGB crown detection model. Despite limitations in the original unsupervised detection approach, this noisy training data may contain information from which the neural network can learn initial tree features. We then refine the initial model using a small number of higher-quality hand-annotated RGB images. We validate our proposed approach while using an open-canopy site in the National Ecological Observation Network. Our results show that a model using 434,551 self-generated trees with the addition of 2848 hand-annotated trees yields accurate predictions in natural landscapes. Using an intersection-over-union threshold of 0.5, the full model had an average tree crown recall of 0.69, with a precision of 0.61 for the visually-annotated data. The model had an average tree detection rate of 0.82 for the field collected stems. The addition of a small number of hand-annotated trees improved the performance over the initial self-supervised model. This semi-supervised deep learning approach demonstrates that remote sensing can overcome a lack of labeled training data by generating noisy data for initial training using unsupervised methods and retraining the resulting models with high quality labeled data.

Original URL: https://www.mdpi.com/2072-4292/11/11/1309

👍 Benno Simmons

Vincent Landau (vincent.landau@gmail.com)

2021-10-18 17:57:32

@Tony Chang

Zhongqi Miao (zhongqi.miao@berkeley.edu)

2021-10-18 18:18:25

Hello everyone! I am a sixth-year PhD student from UC Berkeley. We have just published a paper on Nature Machine Intelligence about deployable wildlife AI recognition systems with efficient humans in the loop and imperfect models! Please check it out if you are interested! This project is moving towards practical deployment in Africa. We hope this work can become an important step to the AI solutions for real-world conservation problems with imperfect AI models. I am very happy to answer any questions you have! Here is the link to the journal webpage (with a preprint link incorporated in case you don't have subscriptions): https://www.nature.com/articles/s42256-021-00393-0 Thank you very much!

Nature Machine Intelligence

Iterative human and automated identification of...

Nature Machine Intelligence - Camera trapping is a widely adopted method for monitoring terrestrial mammals. However, a drawback is the amount of human annotation needed to keep pace with...

Original URL: https://www.nature.com/articles/s42256-021-00393-0

👏 Sara Beery, aruna, Suhail Alnahari, Subhransu Maji, Omiros Pantazis, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Talia Speaker, Emilio Luz-Ricca

:thumbsup_all: Frederic Fol Leymarie, Monty Ammar

👋 Carl Boettiger

Matt Weldy (matthewjweldy@gmail.com)

2021-10-19 14:38:40

Hello everyone! I just joined this channel, and I am excited to see and explore some of the resources here. I am a first year PhD student at Oregon State University, in the US. Most of my dissertation work will explore extensions and applications of machine learning tools to bioacoustics data. In particular, I am interested in improving our in-place tools for processing and classifying bird call types with a focus on noisy, overlapping, multilabel data. There are very few people in my department working on applied ML projects, so I am interested in meeting others working on similar problems.

👋 Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Jon Van Oast, Jason Holmberg (Wild Me)

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2021-10-19 15:01:13

Welcome, Matt! Are you by chance working with Prof. Rebecca Hutchinson? If not, you may be interested in her work at the intersection of ML + ecology 🙂

❤️ Sara Beery

Declan (declan.pizzino@consbio.org)

2021-10-19 15:25:57

Hello from Corvallis! ❤️ OSU

👋 Matt Weldy

Matt Weldy (matthewjweldy@gmail.com)

2021-10-19 15:38:45

I am not working with Rebecca. However, I took a few courses with her during my Master's. She is a great teacher and really knowledgeable about the intersection of ML and ecology. Especially about ML methods applied to distributions and interactions.

Zhongqi Miao (zhongqi.miao@berkeley.edu)

2021-10-19 15:40:33

Welcome, Matt! You might also be interested in Justin Kitzes from UPitt who is doing very similar projects on bird calls and multi-label classifications.

👍 Matt Weldy

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-10-19 17:19:10

^ And Tessa Rhinehart, who is a PhD student in that lab! Also might be worth talking to Stefan Kahl, who runs BirdNet (from Cornell Lab of O)

Sara Beery (sara.beery@gmail.com)

2021-10-19 20:14:39

And @gvanhorn who just released sound ID in the merlin app!

👍 Carly Batist

Matt Weldy (matthewjweldy@gmail.com)

2021-10-19 22:16:07

Thanks for the recommendations everyone. I'll reach out to a number of these researchers. Digging through some of the recent literature I get the feeling it might be time for a small symposium to gather people working through similar problems.

❤️ Sara Beery, Yves Bas, Carly Batist

Ross Gardiner (ross.gardiner@dynaikon.com)

2021-10-20 05:39:37

Hi,
Can anyone recommend a dataset containing humans in a wildlife context (ie people with appropriate backgrounds)?
The intention being to train an animal detector to distinguish human from animal observations.

Thijs (thijs@q42.nl)

2021-10-20 08:26:41

I've found it's hard to get datasets with humans. Because these are usually stripped.

👍 Ross Gardiner

Thijs (thijs@q42.nl)

2021-10-20 08:27:50

The humans that are in there are not respresentative, because it's usually people deploying or maintaining the cameras, very close up

Thijs (thijs@q42.nl)

2021-10-20 08:28:19

In what context do you want to use the model?

Thijs (thijs@q42.nl)

2021-10-20 08:28:43

I know the megadetector is pretty good at picking out humans.

👍 Sara Beery, Mitch Fennell

Sara Beery (sara.beery@gmail.com)

2021-10-20 09:33:46

The human data is usually stripped before publication for ethical reasons, because the humans in the images have not given consent (and frequently are not aware their photo is being taken). MegaDetector already does a pretty good job of what you're describing, try it out here: https://github.com/microsoft/CameraTraps/blob/master/detection/megadetector_colab.ipynb

<https://github.com/microsoft/CameraTraps/blob/master/detection/megadetector_colab.ipynb | megadetector_colab.ipynb>

``<code> { "cells": [ { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "s-Nso4g7VXxy" }, "source": [ "&lt;a href=\"https://colab.research.google.com/github/microsoft/CameraTraps/blob/master/detection/megadetector_colab.ipynb\"&gt;\n", " &lt;img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open in Colab\"/&gt;\n", "&lt;/a&gt;\n", "\n", "Link in case the above badge doesn't redirect you correctly: [Open in Colab](<https://colab.research.google.com/github/microsoft/CameraTraps/blob/master/detection/megadetector_colab.ipynb>)\n", "\n", "This notebook replaces a previous example by [@louis030195](<https://github.com/louis030195>). Improvements: updated environment setup, MegaDetector model version and support for mounting Google Drive folders so you can process your own images here." ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "vUXNQZtwEYiQ" }, "source": [ "# Running MegaDetector on camera trap images using Google Colab\n", "Put together by Alistair Stewart, Alice Springs, May 2020.\n", "@alsnothome\n", "\n", "For reference please read the [MegaDetector guide on GitHub](<https://github.com/microsoft/CameraTraps/blob/master/megadetector.md>) and check there for updates. Here we have roughly followed the steps for running under Linux.\n", "\n", "This notebook is designed to load camera trap image files already uploaded onto Google Drive. If you don't have images already loaded onto Google Drive or just want to see a demo of MegaDetector in action, we also provide code to download some sample images.\n", "\n", "The steps walk through copying of all of the required model and helper files to the Colab runtime and installing all the required packages. You can then connect to your Google Drive folder and process all of the images in a folder using the MegaDetector saved model. The output is saved in a JSON file - a text based database file whose format is described in this [section](<https://github.com/microsoft/CameraTraps/tree/master/api/batch_processing#batch-processing-api-output-format>) in the batch API user guide. The detections (as bounding boxes) can then be rendered on your images.\n", "\n", "The Google Colab instance will only stay open for a maximum 10-12 hrs and after that it will close and any unsaved data will be lost. We recommend saving the JSON output and annotated images into your Google Drive folder for persistent storage." ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "9aUlxnm7cnWy" }, "source": [ "## Set up the Colab instance to run on GPU processing\n", "\n", "\n", "Navigate to Edit→Notebook Settings and select \"GPU\" from the Hardware Accelerator drop-down " ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "LUyqKSAWRGNw" }, "source": [ "## Copy the model, install dependencies, set PYTHONPATH\n", "\n", "Note: from here on you'll start seeing a mix of code. Most are Linux system commands, rather than Python. The system commands are prefixed by a shebang</code>!<code>, which tells this notebook to execute them on the command line." ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "ddPlAKHFTn3m" }, "source": [ "### Install TensorFlow v1\n", "\n", "TensorFlow is already installed in Colab, but our scripts are not yet compatible with the newer version of TensorFlow. \n", "\n", "Please follow the next three steps in sequence and do not skip any steps :) If you were not able to follow these, you can reset the runtime by going to \"Runtime\" in the top menu and \"Factory reset runtime\".\n", "\n", "\n", "1. Uninstall the existing version of TensorFlow (this doesn't affect your other Colabs, don't worry)\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "pip uninstall tensorflow" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "2. Install the older TensorFlow version using</code>pip<code>, with GPU processing by specifying</code>-gpu<code>and version number</code>1.13.1<code>. We also install the other required Python packages that are not already in Colab -</code>humanfriendly<code>and</code>jsonpickle<code>." ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": {}, "colab_type": "code", "id": "EMEkgpy6T0pr" }, "outputs": [], "source": [ "pip install tensorflow-gpu==1.13.1 humanfriendly jsonpickle" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "3. Importantly, you now need to ****re-start the runtime**** of this Colab for it to start using the older version TensorFlow that we just installed.\n", "\n", "Click on the \"Runtime\" option on the top menu, then \"Restart runtime\". After that, you can proceed with the rest of this notebook.\n", "\n", "Let's check that we have the right version of TensorFlow (1.13.1):" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "import tensorflow as tf\n", "print(tf.__version__)" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "hXn_-PZqTWB4" }, "source": [ "### Download the MegaDetector model file\n", "\n", "Currently, v4.1 is avaialble by direct download. The link can be found in the GitHub MegaDetector readme: MegaDetector v4.1, 2020.04.27 frozen model (.pb)" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": {}, "colab_type": "code", "id": "s5uwmpmaTZMX" }, "outputs": [], "source": [ "!wget -O /content/megadetector_v4_1_0.pb <https://lilablobssc.blob.core.windows.net/models/camera_traps/megadetector/md_v4.1.0/md_v4.1.0.pb>" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "nmJ6lQX8S4im" }, "source": [ "### Clone the two required Microsoft git repos\n", "This will copy the latest version of the Microsoft AI for Earth \"utilities\" and \"Camera Traps\" repositories from GitHub. These make data handling and running the model easy. " ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": {}, "colab_type": "code", "id": "7qhltAaRSe1W" }, "outputs": [], "source": [ "!git clone <https://github.com/microsoft/CameraTraps\n>", "!git clone <https://github.com/microsoft/ai4eutils>" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "XQTdfBPZiXiV" }, "source": [ "We'll also copy the Python scripts that run the model and produce visualization of results to the working directory." ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": {}, "colab_type": "code", "id": "t4Ns5PjeiTro" }, "outputs": [], "source": [ "!cp /content/CameraTraps/detection/run_tf_detector_batch.py .\n", "!cp /content/CameraTraps/visualization/visualize_detector_output.py ." ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "2pzfM5Y-iby1" }, "source": [ "### Set</code>PYTHONPATH<code>to include</code>CameraTraps<code>and</code>ai4eutils<code>\n", "\n", "Add cloned git folders to the</code>PYTHONPATH` environment variable so that we can import their modules from any working directory.\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": {}, "colab_type": "code", "id": "d8vanlgAOlEj" }, "outputs": [], "source": [ "import os\n", "os.environ['PYTHONPATH'] += \":/content/ai4eutils\"\n", "os.environ['PYTHONPATH'] += \":/content/CameraTraps\"\n", …

👍 Ross Gardiner

Ross Gardiner (ross.gardiner@dynaikon.com)

2021-10-20 09:43:26

Thanks both.
I want to train a small SSD model for edge deployment on a video camera trap. We are interested in human rejection from our captured images.
Perhaps megadetector could be used to generate my training data... 🤔

👍 Sara Beery

Sara Beery (sara.beery@gmail.com)

2021-10-20 09:46:49

Yeah I use it for weak labeling all the time!

👍 Ross Gardiner

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2021-10-20 11:14:56

@Elizabeth Bondi has a BIRDSAI dataset with exactly this, for thermal imagery! https://sites.google.com/view/elizabethbondi/dataset

sites.google.com

Elizabeth Bondi - DATASET

Original URL: https://sites.google.com/view/elizabethbondi/dataset

👍 Elizabeth Bondi

Elizabeth Bondi (ebondi@g.harvard.edu)

2021-10-20 11:15:55

Thanks @Lily Xu - OLD ACCOUNT, DO NOT MESSAGE! Please let me know if you have any questions, @Ross Gardiner

Nasibah Azhari (nasibah.azhari@live.com)

2021-10-22 10:32:06

Hi☺️ I’m currently enrolled in an immersive data science program and I’m doing a presentation on spatial data science with a focus on its applications in conservation. I was wondering if anyone had any interesting resources/case studies/tips?

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-10-22 12:01:36

Check out WILDLABS, an online conservation tech community! And the Conservation Tech Directory for examples of resources/companies/organizations working in geospatial applications for the environment. Search ‘geospatial’ or ‘remote sensing’

conservationtech.directory

NEON Ecological Forecasting Challenge

Visit the post for more.

Original URL: https://conservationtech.directory/

Nasibah Azhari (nasibah.azhari@live.com)

2021-10-23 09:12:41

Thank you, @Carly Batist ! ☺️

👍 Carly Batist

Carl Boettiger (cboettig@berkeley.edu)

2021-10-22 12:19:08

👋 Hi all,
I'm an Assistant Professor in UC Berkeley's Dept of Environmental Science, Policy, and Management where I focus in dealing with uncertainty in conservation models and decision-making. Recently we have been developing Deep RL approaches to sequential decision-making problems in conservation and I'm interested in developing a library of such problems that could serve as an open benchmark. (see https://boettiger-lab.github.io/conservation-gym/). My group is also interested in ecological forecasting, folks looking for open problems might consider submitting iterative, probabilistic forecasts to the ongoing NEON forecast challenge: https://projects.ecoforecast.org/neon4cast-docs/. Lastly, our group has an ongoing collaboration with political scientists and ethicists to better understand ethical and political considerations that arise from the application of AI to conservation problems. I'd be thrilled to discuss any of these themes!

projects.ecoforecast.org

Documentation for the NEON Ecological Forecasting Challenge

Original URL: https://projects.ecoforecast.org/neon4cast-docs/

👋 Declan, Jason Holmberg (Wild Me), Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Ayan Mukhopadhyay, Sara Beery, Benno Simmons, Ritwik, Vincent Landau, Matt Weldy, Chris Yeh, David, Atriya Sen, Casey Youngflesh, Zhongqi Miao, Emilio Luz-Ricca, Carly Batist, Angjoo Kanazawa, Anthony Bao, Monty Ammar

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-10-26 15:43:52

The Conservation Tech Directory has gone past the 500 mark!🎉🎊 Now at 512, to be exact. Thanks to all who have contributed new entries & corrections on our Add/Update a Resource form (keep them coming 🙂).
AND my co-developer @Gracie Ermi has given us a logo (see below)!

conservationtech.directory

How artificial intelligence has become the latest wildlife conservation tool

Visit the post for more.

Original URL: https://conservationtech.directory/

😍 Sara Beery, Gracie Ermi, Jason Holmberg (Wild Me), Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Ankita Shukla, Olivier Gimenez

🎉 Gracie Ermi, Talia Speaker, Declan, Emily Charry Tissier, Jason Holmberg (Wild Me), Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Agnethe Seim Olsen, Carl Boettiger

Olga Mierzwa-Sulima (olga@appsilon.com)

2021-10-28 05:29:28

Hi folks 👋 I'm a Data4Good Lead with a Data science background at Appsilon. I run there an internal program where we support sustainability initiatives. One of my colleagues @Jędrzej Świeżewski is already here for some time. We are the team behind Mbaza project, AI tool that helps classify animals from camera traps deployed in National Parks in Gabon. Next year we will have a strong focus on biodiversity. I'm here to learn about challenges that we could help solving with our team skills (ML Vision and data viz) and network with people from biodiversity space. I'd especially would love to meet researchers who's research would benefit from cooperation with us. If you want to grab a virtual coffee with me, send ma a DM. I'd love to chat!

The Independent

We are working with conservation charity Space for Giants to protect wildlife at risk from poachers due to the conservation funding crisis caused by Covid-19. Help is desperately needed to support wildlife rangers, local communities and law enforcement personnel to prevent wildlife crime

Original URL: https://www.independent.co.uk/stop-the-illegal-wildlife-trade/ai-technology-wildlife-conservation-gabon-b1783812.html

👋 Sara Beery, Talia Speaker, Jason Holmberg (Wild Me), Ayan Mukhopadhyay, Anthony Bao, Carl Boettiger, Carly Batist, Jędrzej Świeżewski

Angjoo Kanazawa (kanazawa@berkeley.edu)

2021-10-31 20:20:26

Hi all👋! What an amazing community! I’ve been lurking a little bit!
My name is Angjoo and I’m an assistant professor at UC Berkeley in the department of Electrical Engineering and Computer Science, specializing in computer vision, graphics, and learning. My research is about perceiving the dynamic 3D world that underlies our images and videos. In particular I focus on non-rigid objects, such as humans or animals, and my thesis was on 3D reconstruction of Animals 🙂. This is such a challenging problem with a lot of depth that we will continue to pursue and I would love to hear from you about real problems and wishlists of what computer vision methods could provide! http://people.eecs.berkeley.edu/~kanazawa/

👋 Sara Beery, Justin Kay, Silvia Zuffi, Oisin Mac Aodha, Emily Charry Tissier, Benno Simmons, Emilio Luz-Ricca, Declan, Jason Holmberg (Wild Me), Armin Bazarjani

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2021-11-01 05:59:38

ようこそ! Great to see you here! 😄 And a great initiative too; I am likewise interested in getting to know more about the potential applications of vision(ary) work for conservation & Co.

🙂 Angjoo Kanazawa

Rowan Converse (rowanconverse@unm.edu)

2021-11-01 15:59:54

Hi all, I’m a PhD candidate in Geography at the University of New Mexico. My dissertation work is focused on detection and identification of waterfowl from UAS imagery. We have a sizable set of labeled UAS images of birds (mostly ducks) that we are interested in making public for research purposes-- any suggestions on a good platform to publish it? Thanks for any advice!

👋 Sara Beery, Oisin Mac Aodha, Declan, Jason Holmberg (Wild Me)

Sara Beery (sara.beery@gmail.com)

2021-11-01 16:01:11

@Ben Weinstein

Carl Boettiger (cboettig@berkeley.edu)

2021-11-01 16:11:12

@Rowan Converse What's the database size? For scientific research purposes, a DOI-granting repository like Zenodo.org (which is built on CERN's Data Center) is probably preferable.

👍 Sara Beery, Rowan Converse

Carl Boettiger (cboettig@berkeley.edu)

2021-11-01 16:16:29

I believe up to 50 GB per 'dataset' is free, and there is not a limit on how many free datasets (https://help.zenodo.org/). They encourage you to contact them for larger use cases. You've probably already thought through copyright for distribution and possible concerns about geolocation of rare or protected species.

Sara Beery (sara.beery@gmail.com)

2021-11-01 16:30:27

There's also LILA.science @Dan Morris

👀 Carl Boettiger

Dan Morris (agentmorris@gmail.com)

2021-11-01 16:57:02

As Sara suggests, we host some data like this on lila.science, e.g.:

https://lila.science/datasets/aerial-seabirds-west-africa/

If you're interested, drop us an email at info@lila.science. Thanks!

LILA BC

Aerial Seabirds West Africa - LILA BC

Written by

lilawp

Est. reading time

2 minutes

Original URL: https://lila.science/datasets/aerial-seabirds-west-africa/

👍 Rowan Converse

Carl Boettiger (cboettig@berkeley.edu)

2021-11-01 17:26:39

LILA looks really cool, I hadn't seen that. No doubt there are some significant performance advantages in using Azure-based hosting, especially if your data is in the TB+ range (like those NOAA seals). It's generous of Microsoft to host such a resource. All the same, I do think it would be nice to see modern metadata standards, such as the FAIR principles, for data intended for scientific archives.

Carl Boettiger (cboettig@berkeley.edu)

2021-11-01 17:31:11

For instance, for that seal image data, NOAA maintains a rich metadata record, https://www.fisheries.noaa.gov/inport/item/63322, which includes schema.org markup, allowing it to be included in linked-data indexes like https://datasetsearch.research.google.com or data.gov. DOI-granting repositories also integrate into publication services for purposes of citation metrics to the original data, track versions, and must meet a high standard of archival persistence for scientific research. These may be considerations if citability and discoverability are important, particularly if LILA grows to index a more datasets than can be easily browsed manually.

fisheries.noaa.gov

A Dataset for Machine Learning Algorithm Development | InPort

This dataset consists of imagery, imagery footprints, associated ice seal detections and homography files associated with the KAMERA Test Flights conducted i...

Original URL: https://www.fisheries.noaa.gov/inport/item/63322

Ben Weinstein (benweinstein2010@gmail.com)

2021-11-01 17:48:06

@Rowan Converse as part of the bird paper, we placed your portion on zenodo. See newmexico.zip https://zenodo.org/record/5033174#.YYBggNbMKlo I recommend zenodo for the rest.

👍 Rowan Converse, Sara Beery

Rowan Converse (rowanconverse@unm.edu)

2021-11-01 17:48:36

Thanks everyone for the replies, this is great!

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-11-02 08:36:29

Also, figshare waives dataset size limits if it’s set as public (good for large datasets)

👍 Rowan Converse

Dan Sheldon (sheldon@cs.umass.edu)

2021-11-03 16:33:48

Does anyone have experience/recommendations for archiving larger data sets — on the order of, say, 2-3 TB?

I checked with zenodo (a while back) and there was no mechanism that would work for both my institution and theirs to provide payment for the extra resources needed for a data set of this size, and I understand that figshare has a limit of 1TB.

Sara Beery (sara.beery@gmail.com)

2021-11-03 16:35:14

There are several datasets on LILA that are many TB. I would reach out to Dan Morris (email info@lila.science).

👍 Dan Sheldon

Dan Sheldon (sheldon@cs.umass.edu)

2021-11-03 16:38:14

Thanks!

Dan Sheldon (sheldon@cs.umass.edu)

2021-11-03 16:40:19

I can ask Dan this, but do you know if LILA grants DOIs?

Sara Beery (sara.beery@gmail.com)

2021-11-03 16:42:42

I'm not sure. If not I wonder if that's something that could be set up.

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-11-03 16:56:15

Oh really, I didn’t know figshare had a 1TB limit? (Though I’ve not actually tried) I know they have a ‘figshare+’ that allows up to 5TB but that requires a one-time data publishing charge (so you may run into the same problems you had with zenodo)

Dan Sheldon (sheldon@cs.umass.edu)

2021-11-03 16:59:00

Thanks! I’ll look into the details of figshare. My information came from a publisher’s page and may not be up-to-date. A charge could be OK, the issue with zenodo was more that they didn’t have an easy payment mechanism.

Dan Morris (agentmorris@gmail.com)

2021-11-03 17:04:44

LILA doesn't generate DOIs, plus if I told you we did, I would then immediately tell you not to use any DOI that I generated for you. 🙂 IMO you want to be in total control of any DOIs on which you take dependencies (e.g. by embedding in an archival publication).

[Now I will drift from facts about LILA into philosophy about DOIs...]

Regardless of where you host data, unless you are 100% sure your host will be there in 10 years, my personal recommendation is that you administer your own DOI.

👍 Sara Beery

Dan Sheldon (sheldon@cs.umass.edu)

2021-11-03 17:18:20

Interesting!

Jorrit van Gils (vangilsjorrit@gmail.com)

2021-11-02 10:55:53

Dear people from AI for Conservation, I'm master-student Forest Nature conservation in Wageningen (The Netherlands). In my thesis I compare two deep learning behaviour classification techniques on wildlife camera trap images of Red deer. One technique includes pose estimation. As I'm really passionate about AI combined with camera traps and satellites, I'm looking for PhD positions or other career opportunities that allow to further develop in artificial intelligence after I graduate January 2022. Looking forward to meet you!

👋 Sara Beery, Carl Boettiger, Jason Holmberg (Wild Me), Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Thijs

Thijs (thijs@q42.nl)

2021-11-03 12:18:16

@Jorrit van Gils maybe we should have a quick chat. Sound like we're working on similar things (in the Netherlands)!

Sara Beery (sara.beery@gmail.com)

2021-11-02 12:19:28

Come learn more about the CV4Ecology summer school! Do you have any questions about the course? the application process? the timeline? Not sure if you're a good fit? We're hosting a zoom infosession on 11/4 at 9:30am PT (GMT -7:00), sign up here: https://docs.google.com/forms/d/e/1FAIpQLSdSOA-C9bJc1bhgdyB1vk3xVkcD8u10dSeaFRLZMoQydvn3TA/viewform

👍 Mitch Fennell, Carly Batist, Jason Holmberg (Wild Me), Rowan Converse, Reshu Bashyal

Axel Rossberg (axel@rossberg.net)

2021-11-03 14:04:56

Hi, I am a theoretical ecologist working on biodiversity, food webs, and meta-community ecology, and just joined AI for Conservation.

👋 Sara Beery, Stefan Schneider, Declan, Oisin Mac Aodha, Benno Simmons, Carl Boettiger, Nicolas Betancourt, Monty Ammar

Reshu Bashyal (bashyalreshu@gmail.com)

2021-11-10 00:11:16

Hi AI for Conservation! I am Reshu Bashyal, a conservationist based in Nepal. I have no prior experience in AI but am very much fascinated with how AI is being integrated in conservation activities. I know, this is a great platform for AI aspirants. Looking forward to learning from the community.

👋 Sara Beery, Arjun Subramonian (they/them), Suzanne Stathatos, Talia Speaker, Benjamin Kellenberger, Lukas Picek, Omiros Pantazis, Ritwik, Nicolas Betancourt, Declan, aruna, Angjoo Kanazawa, Jon Van Oast, Monty Ammar

:thumbsup_all: Frederic Fol Leymarie

Lucia Gordon (luciagordon@college.harvard.edu)

2021-11-11 21:04:52

Hello everyone! I recently joined AI for Conservation and am excited to be part of the community. I am a senior in college interested in pursuing a PhD in AI for Conservation. As I haven’t found any compiled list of academic labs pursuing this kind of work, I thought that we could create one for the community. Please add any academic labs you know of working on AI for Conservation to this spreadsheet!

AI for Conservation Academic Labs

👍 Sara Beery, Oisin Mac Aodha, Carly Batist, Hemal Naik, Jorrit van Gils, Monty Ammar

🙌 Emilio Luz-Ricca, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, aruna, Gracie Ermi

🤩 Rosie Crawford

Suzanne Stathatos (suzanne.stathatos@gmail.com)

2021-11-12 10:07:27

https://www.compsust.net/

👍 Sara Beery

Ted Schmitt (teds@allenai.org)

2021-11-12 13:58:37

I’m not sure who to add but you should be aware of this work: https://www.wur.nl/en/show/Imagine-all-the-animals-living-life-in-sociable-groups.htm in the Netherlands

WUR

Imagine all the animals, living life in sociable groups

On October 21st, the program anIMAl Group sENsor (IMAGEN) was kicked off. In IMAGEN, researchers of Wageningen University & Research collaborate with various stakeholders to enable and stimulate a transition to sustainable livestock production. The program consists of three projects, each of which is explained in a short video.

Original URL: https://www.wur.nl/en/show/Imagine-all-the-animals-living-life-in-sociable-groups.htm

Katie Wetstone (she, her) (katie@drivendata.org)

2021-11-12 15:08:31

Hi everyone! 🐵 We are spreading the word about a free, open-source tool called Zamba that automatically detects and classifies animals in camera trap videos. If you use camera traps to capture videos, we’d love your feedback! 🐵 Try your videos with the species we cover or use our training functionality to make yourself a custom model just for your data.

We want to make the tool as useful as possible, and are hoping to gather user feedback. In particular, we’d love to have users test out the Zamba python package. We’ve just released v2 of this package with brand new models and more features!

For background, we at DrivenData developed the tool in partnership with experts from the Max Planck Institute for Evolutionary Anthropology (MPI-EVA). A few basics about Zamba:

🧠 Zamba uses artificial intelligence and computer vision to perform intensive camera trap video processing work, freeing up more time for humans to focus on interpreting the content and using the results.

:female_technologist: Zamba can be accessed through an easy command line interface or as a Python package - the code is all open-source on Github!

🐘 Pretrained models are available to predict 42 different species common to western Europe and central Africa, plus blank versus non-blank.

🌍 Zamba can be adapted to any set of species or ecosystem. Users can easily use their own labeled videos to generate a retrained model specific to their use case.

📸 Zamba is trained on over 27,000 hand-labeled camera trap videos.

A couple ways you can contribute:

Flag any bugs you find while using the Zamba python package - or submit an issue directly to the Github repo

Let us know which parts of the package documentation are confusing or could be improved

Train a new custom model and make it available to others through the Model Zoo Wiki

You can send us any feedback or thoughts by commenting on this post, filling an issue on the GitHub repository, or by emailing info@drivendata.org. Close collaboration with subject area expects has been critical to the development of Zamba, and we look forward to hearing your perspectives! 🎉

😎 Jon Van Oast, Jason Holmberg (Wild Me), Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Sachith Seneviratne, Ayan Mukhopadhyay

🎁 Peter Bull

👍 Sara Beery, Bistra Dilkina, Cameron Trotter, Ross Gardiner, Dan Morris

🎉 Emily Dorne, Jason Parham

Katie Wetstone (she, her) (katie@drivendata.org)

2021-11-12 15:11:49

Tagging some of our awesome team members! @Peter Bull @Emily Dorne @Greg Lipstein

Oc (boulaisoceane@gmail.com)

2021-11-15 14:00:12

has this been used for underwater ecosystems?

Peter Bull (peter@drivendata.org)

2021-11-15 14:08:39

Not yet—do you have more info on how the data is collected? I expect that since some of the parameters are tuned 1 minute long videos from terrestrial motion-triggered camera traps there is likely a fair amount of work to tune the models to the underwater setting.

Hemal Naik (hnaik@ab.mpg.de)

2021-11-16 10:01:02

Hi everyone, which conferences or journals accept datasets? (Both biology and computer science venues would be fine) I am working with MPI- Animal Behavior and we are trying to push out lot of interesting and unique datasets for CV and ML community. I know that NeurIPS announced dataset track recently. Does anyone here have experience of publishing datasets? Would love to connect and get some feedback on our work.

Oisin Mac Aodha (macaodha@caltech.edu)

2021-11-16 10:10:40

Hey Hemal.

We often have lots of dataset papers at the FGVC workshop at CVPR. e.g. from last year
https://sites.google.com/view/fgvc8

sites.google.com

FGVC8

Announcements Join our slack page for updates and live Q&A during the workshop. The workshop schedule is now available here. FGVC8 will be held on June 25th. FGVC8 accepted workshop papers are now available here. Our paper submission deadline has been extended to April 8th. The CMT site for

Original URL: https://sites.google.com/view/fgvc8

👍 Hemal Naik, Sara Beery

Beckett Sterner (bsterne1@asu.edu)

2021-11-16 11:14:28

The Biodiversity Data Journal could be an option: https://bdj.pensoft.net/journals.php?journal_name=bdj

bdj.pensoft.net

Biodiversity Data Journal

Biodiversity Data Journal (BDJ) is a community peer-reviewed, open access journal, designed to accelerate publishing, dissemination and sharing of biodiversity-related data of any kind. All structural elements of the articles &ndash; text, morphological descriptions, occurrences, data tables, etc. &ndash; are treated and stored as data. The journal publishes papers in biodiversity science containing taxonomic, floristic/faunistic, morphological, genomic, phylogenetic, ecological or environmental data on any taxon of any geological age from any part of the world. BDJ aims at integrating data and narrative in the article content to the maximum extent possible. Supplementary data files that underpin graphs, hypotheses and results should also be published with the article or deposited in trusted open access data repositories. The journal provides rich biodiversity data import and export facilities through the ARPHA Writing Tool and Darwin Core Archives.

Original URL: https://bdj.pensoft.net/journals.php?journal_name=bdj

Ayan Mukhopadhyay (ayanmukg@gmail.com)

2021-11-16 12:38:38

Hi @Hemal Naik, Nature Scientific Data (https://www.nature.com/sdata/) is a great place and so is the NeurIPS database track. We just published one at NeurIPS so I would be happy to answer any questions that you might have.

👍 Sara Beery

Hemal Naik (hnaik@ab.mpg.de)

2021-11-16 12:40:01

Hi @Ayan Mukhopadhyay thanks for the offer. Would definitely like to get in touch. Can you share link of your paper. I assume the paper would also have communication email. Looking forward.

Ayan Mukhopadhyay (ayanmukg@gmail.com)

2021-11-16 12:44:00

Sure, here you go https://ayanmukhopadhyay.github.io/files/neurips2021.pdf

Hemal Naik (hnaik@ab.mpg.de)

2021-11-16 12:49:04

Thanks I will get in touch.

Ted Schmitt (teds@allenai.org)

2021-11-19 19:58:37

Clare Barclay, Chief Executive Officer at Microsoft UK, said: “The untapped potential of AI and machine learning can help solve some of the world’s most complex environmental challenges. Our first-of-its-kind multispecies AI model Project SEEKER can help tackle the wildlife trafficking trade, while protecting animal ecosystems. The importance of collaboration and partnership with more organisations couldn’t be greater as we look to protect the environment and the world’s most endangered species.”
https://news.microsoft.com/en-gb/2021/11/18/first-of-its-kind-multispecies-ai-model-to[…]egal-wildlife-trafficking-is-ready-to-roll-out-to-airports/

Microsoft News Centre UK

First-of-its-kind multispecies AI model to detect illegal wildlife trafficking is ready to roll out

HRH The Duke of Cambridge visits Microsoft's UK headquarters to learn about Project SEEKER as part of his work with The Royal Foundation.

Written by

Est. reading time

7 minutes

Original URL: https://news.microsoft.com/en-gb/2021/11/18/first-of-its-kind-multispecies-ai-model-to-detect-illegal-wildlife-trafficking-is-ready-to-roll-out-to-airports/

👍 Sara Beery, Jason Holmberg (Wild Me), Megan Cromp, Dan Morris

😎 Jon Van Oast, Jason Holmberg (Wild Me)

Gracie Ermi (gracieermiifthen@gmail.com)

2021-11-22 14:35:24

Thanks to @Lucia Gordon’s great idea to compile a list of academic labs doing conservation tech research, @Carly Batist and I have added a new “academic lab” tag to https://conservationtech.directory and a number of new entries in that category (we’re up to 568 total entries now 🎉). Know of any university groups working on conservation tech that aren’t included? Let us know so that we can get them added! And please keep letting us know about any corrections that need to be made to anything in the directory. Thanks everyone!

conservationtech.directory

Climate Change AI Summer School 2022

Visit the post for more.

Original URL: https://conservationtech.directory

❤️ Lucia Gordon, Sara Beery, Justin Kay, Talia Speaker, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Declan, Carly Batist, Atriya Sen

🎉 Jon Van Oast, Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Carly Batist, Atriya Sen

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2021-11-22 15:08:58

Thank you Lucia for the idea and Gracie and Carly for the implementation!! What a wonderful resource (I wish I had this 4 years ago when I was applying to grad schools!)

😃 Lucia Gordon, Gracie Ermi, Carly Batist

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2021-11-22 15:12:35

Climate Change AI is excited to announce a call for applications for their upcoming summer school on Climate Change and Artificial Intelligence in 2022. This summer school (to be held virtually on weekdays between Aug 15th - 26th 2022 (time zone TBD), is designed to educate and prepare participants with a background in artificial intelligence and/or a background in a climate-change related field to tackle major climate problems using AI.

climatechange.ai

Climate Change AI

Tackling Climate Change with Machine Learning

Original URL: https://www.climatechange.ai/

Climate Change AI

Original URL: https://www.climatechange.ai/events/summer_school2022.html

Climate Change AI

Climate Change AI Summer School 2022

Original URL: https://www.climatechange.ai/events/summer_school2022#frequently-asked-questions

❤️ Sara Beery, Elijah Cole (Deactivated), Lucia Gordon, Chris Yeh, Sachith Seneviratne, Peter Bull, Stefan Schneider, Carly Batist, David, Gracie Ermi

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2021-11-22 15:12:58

Climate Change AI Summer School 2022
Virtual: weekdays between Aug 15th – 26th, 2022 (time zone TBD)

Website: https://www.climatechange.ai/events/summer_school2022.html
Application deadline: December 17th 2021
Contact: summerschool@climatechange.ai

The Climate Change AI summer school is designed to educate and prepare participants with a background in artificial intelligence and/or a background in a climate-change related field to tackle major climate problems using AI. The summer school aims to bring together a multidisciplinary group of participants and facilitate project-based team work to strengthen collaborations between different fields and foster networking in this space.

The first part of the summer school will consist of a mix of lectures and hands-on tutorials organized into two tracks, one focused on AI fundamentals and one focused on climate change. In both tracks, the program will provide an overview of machine learning applications in a broad range of climate change-related areas. This includes covering foundational machine learning methods and state-of-the-art tools, while underlining their advantages and limitations, and describing how they can be used in practice to address the climate crisis. The second part of the summer school will consist of a collaborative project at the intersection of climate change and machine learning. Participants will work together in multidisciplinary groups under the guidance of a mentor to develop AI-based solutions for climate change problems.

The cohort will be composed of applicants from complementary areas of study/work, to be selected on the basis of their background and experience as well as their motivation for joining the summer school.

Applications are due by: Dec 17, 2021 23:59 AOE (Anywhere on Earth, UTC-12). To apply, please submit your application through this form:
https://forms.gle/9eSETySjVvdzkfiM6
Admission notifications will be sent out during the week of Feb 21, 2022.

The summer school is free to attend. Applicants who are accepted will be asked to confirm their attendance for the entire duration of the summer school. This course will be instructed by members of CCAI and world-renowned experts in ML and Climate Change.

For further information please check out the summer school website or contact summerschool@climatechange.ai

accounts.google.com

Google Forms - create and analyze surveys, for free.

Create a new survey on your own or with others at the same time. Choose from a variety of survey types and analyze results in Google Forms. Free from Google.

Original URL: https://forms.gle/9eSETySjVvdzkfiM6

Krasi Georgiev (krasi@arribada.org)

2021-11-23 08:22:47

@Dan Morris hey Dan, re this issue https://github.com/microsoft/CameraTraps/issues/260

I am working on developing a smart camera as part of the arribada.org initiative.
The idea of the camera is to reduce human-animal conflict by providing an early alarming system for camps, villages etc.

Basically run a detection model on low cost/ low energy devices - rpi, arduino etc. and when it detects a lion or an elephant for example to send a signal to people in charge of the camp security.

We are working together with edgeimpulse.com on this project and for the idea was to avoid some manual tagging when building the ML model for the edge device.

<https://github.com/microsoft/CameraTraps/issues/260|#260 Is the batch processing API and the real-time API running the same classifiers ?>

Looking into classifying some thermal images to filter out empty images and roughly label when there is are different objects in the pictures. Will process around 3000 images per day and need to label 3-4 different categories - empty , animal, human Wondering if the live web API or the batch processing would be a better option? The readme says that the web API runs species classifiers so wondering if they are the same for the batch processing Also do you use some public discord or slack?

Comments

😎 Jon Van Oast

Jonathan Crall (erotemic@gmail.com)

2021-11-24 14:32:50

To all those interested in small-object detection problems, augmented reality, web 3.0, open source benchmark datasets, change detection, and re-examining the existing academic publishing paradigm:

I've been working on a personal dataset-collection project for almost a year now, and it's at the point where I'm ready to start releasing details.

Allow me to introduce: ShitSpotter - An open source algorithm and dataset for detecting poop in pictures. The github README gives details about the motivation, data collection, annotation process, and algorithm. The dataset will be published on Web 3.0 via IPFS.

https://github.com/Erotemic/shitspotter

While this is only tangentially related conservation, I think there could be ecological applications of either the dataset itself, or the tools I'm going to write to bootstrap the annotation process. Collecting fecal samples is important in some aspects of ecology, and this dataset domain might be transferred to something like detecting feces (and perhaps even guess the species of the feces) in wooded areas as one metric for population monitoring.

Erotemic/shitspotter

An open source algorithm and dataset for finding poop in pictures. A work in progress.

Stars

Language

Python

😎 Jon Van Oast, Elijah Cole (Deactivated), Sara Beery, Jason Holmberg (Wild Me), Chuck Stewart, Sanjana Baliga, Declan, Yves Bas

💩 Jon Van Oast, Sanjana Baliga, Carly Batist, Agnethe Seim Olsen, Petar Gyurov, Jason Parham

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2021-11-25 07:42:28

Hello everyone,

Are you interested in working on habitat suitability mapping with Earth observation and machine/deep learning, or happen to know an excellent MSc. candidate you could recommend for it?
Thanks to our recently awarded Swiss National Foundation project “Learning unbiased habitat suitability at scale with AI”, we are looking for two motivated PhD candidates to join the ECEO lab at EPFL in Sion, Switzerland!

For more information see the official advertisement here: https://www.epfl.ch/about/working/enac-two-phd-positions-large-scale-habitat-suitability-mapping-with-machine-learning/
I am looking forward to hearing from you!

😍 Sara Beery, Nico Lang, Yihang She, Jorrit van Gils, Oc, Monty Ammar

Monty Ammar (montyx23@gmail.com)

2022-01-10 12:29:56

Hey Benjamin, I actually came across this post today and was interested in applying. I’m still doing my masters which ends in September though, is it still alright to apply with the possibility of starting late if accepted?

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2022-01-10 12:37:40

Hello Monty! At this point we have a selection of applicants to be interviewed. Also, the chances of us postponing the project to the second half of the year are unfortunately slim, as we are bound by the grantees. I’d recommend to keep an eye out for announcements, though!

👍 Monty Ammar

Monty Ammar (montyx23@gmail.com)

2022-01-12 00:54:42

Thanks Benjamin!

Cameron Trotter (cater@bas.ac.uk)

2021-11-25 10:05:29

Hi all, I’m looking for a (labelled) dataset of camera trap images which have been captured in the UK/Northern Europe - ideally from countryside/forested locations. Does anyone have any links to sets I could look at? Sets that contain images from this geographical area alongside others would be fine too provided it would be easy enough to subset and only use UK/NE images. Thanks in advance 🙂

👍 Sara Beery

Alan Ma (alanma393141@gmail.com)

2021-11-27 16:56:08

Howdy all,

I am currently working on a roadkill prevention project that involves using a trap camera to help collect wildlife activity data. It's purpose is to warn drivers of animal presence detected and simultaneously collect image data for monitoring wildlife activity along road environments in the pursuit of preserving wildlife/human wellbeing. In particular this project is expanding the computer vision side of things through greater species identification power. Via multiple infrared sensors and a radar sensor, the project hopes to detect wildlife presence through changes in the environment heat signatures and depth. Once the sensors have been triggered a NoIR camera snaps a picture of it's field of vision. After the picture has been collected, hopefully capturing wildlife in the process, a offline pre-trained ML model runs to identify the said wildlife. The identified wildlife species as well as metadata are incorporated to be mapped for the purpose of identifying regions of dense wildlife activity as well as possible migration patterns.

This project is looking for suggestions and collaboration due to my limited scope on the implementation of artificial intelligence, I was hoping to receive assistance from the community on improving the machine learning capabilities of this project. More specifically, the project is looking to expand on a larger wildlife species prediction capability ** (1) and identify migration/animal behavior patterns (2). (** If you know or are willing to share a database of wildlife images to help improve this project's machine learning model it would be greatly appreciated! One of the major challenges in helping improve the current wildlife identification model is a lack of available images in model training.)

If there are individuals willing to, I would love to discuss more deeply the methods used or other ideas on how to improve on the project's limitations.

Thanks in advance for your time - Alan

👍 Sara Beery, Jon Van Oast, Jason Holmberg (Wild Me), Kai Waddington

Subhransu Maji (smaji@cs.umass.edu)

2021-11-29 06:31:28

Hi all!
I’ve been lurking for a (long) while. I’m an associate professor in computer science at university of Massachusetts, Amherst specializing in computer vision, graphics, and ml (https://people.cs.umass.edu/~smaji/). I’ve organized the fine-grained visual recognition workshops (https://sites.google.com/view/fgvc8) and lately been looking into analyzing bird migration via RADAR data (e.g., https://www.nature.com/articles/s41558-019-0648-9). On the cv+ml side we have been interested in semi- / self-supervised learning for fine-grained classification and have organized a couple of kaggle challenges at the last two FGVC workshops.

We have been thinking about doing the same for part segmentation. Question to the community: Are there any fine-grained segmentation (part labels) datasets of animals out there? We wanted to explore few-shot segmentation tasks, but couldn’t find any datasets of birds or animals with a large number of part labels. Most of these contain keypoints or a just few parts (3-4). For comparison, there are datasets with dozens of labeled parts for human faces or cars.

sites.google.com

FGVC8

Original URL: https://sites.google.com/view/fgvc8

Nature Climate Change

Phenology of nocturnal avian migration has shifted at the continental scale

Nature Climate Change - Climate change affects the timing of bird migration, which can lead to mismatch with resource availability. Migration occurred earlier in spring and autumn in the United...

Original URL: https://www.nature.com/articles/s41558-019-0648-9

👋 Oisin Mac Aodha, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Sara Beery

Daniel Davila (daniel.davila@kitware.com)

2021-11-29 10:16:12

It may be a stretch, but perhaps the recent pose datasets could be of interest? It is not segmentation-ready, but it may be useful to bootstrap some segmentation ground truth.

https://github.com/AlexTheBad/AP-10K

AlexTheBad/AP-10K

NeurIPS 2021 Datasets and Benchmarks Track

Stars

Language

Python

👍 Subhransu Maji

Subhransu Maji (smaji@cs.umass.edu)

2021-11-29 10:20:12

Thanks, this looks great!

Diego Marcos (diego.marcos.gonzalez@gmail.com)

2021-12-21 10:02:15

Hey @Subhransu Maji! Sorry for the delay here. The CUB dataset does have quite extensive part/attribute annotations for 200 species of North American bird species: http://www.vision.caltech.edu/visipedia/CUB-200.html

Sara Beery (sara.beery@gmail.com)

2021-12-01 11:30:01

DEADLINE EXTENSION

We've received multiple requests to submit late applications to the CV4Ecology Summer School, so to be fair to all applicants the deadline is being extended to Friday, December 3rd at midnight Pacific Time. We'll be emailing the mailing list and updating the webpage, please share!

👍 Jorrit van Gils, David, Mitch Fennell, Jason Holmberg (Wild Me), Talia Speaker, Casey Youngflesh, Carly Batist, Benjamin Kellenberger, Riccardo de Lutio, Elijah Cole (Deactivated)

Monty Ammar (montyx23@gmail.com)

2021-12-21 18:21:41

Hey Sara, will this Summer school be running on a yearly basis?

David Will (david.will@islandconservation.org)

2021-12-03 15:44:37

Exciting volunteer opportunity on Robinson Crusoe Island, Chile.

We are pleased to announce the “Work for Humankind” partnership between Lenovo, Island Conservation and the Robinson Crusoe Island community that calls on volunteers from around the world to take part in a once-in-a-lifetime opportunity: to experience first-hand how to make a long-lasting difference with a remote island community, while working from one of the most remote offices in the world enabled by tech.

We are looking for volunteers with a range of skills, backgrounds, and specialties to travel to Robinson Crusoe to help prevent the extinction of endangered species and support the local community as it works toward achieving sustainability, using newly established internet connectivity and computing infrastructure hosted in a community technology hub established by Lenovo.

On the conservation side we are looking for volunteers with experience in machine learning, artificial intelligence, and data management to advance the first community-led, data driven island restoration project to protect the threatened Pink footed shearwater. This includes supporting the local team in training and implementing camera trap classification models and aggregating detection data and field observations into quantitative analysis of project progress.

Those interested in becoming one of the lucky volunteers to support this project on Robinson Crusoe Island while being able to continue working their current day jobs remotely can find out more and apply at www.LenovoWFH.com by December 30, 2021.

https://www.islandconservation.org/work-for-humankind-lenovo-invites-you-to-work-from-one-of-earths-most-remote-locations-with-smarter-technology/

Island Conservation

Work for Humankind: Lenovo invites you to work from one of Earth's most remote locations with smarter technology

Global News Release Work for Humankind: Lenovo invites you to work from one of Earth's most remote locations with smarter technology Research conducted by Lenovo™ reveals more than two-thirds of Gen Z and Millennial respondents feel it’s important to give back and contribute positively to local communities during a ‘Work from Anywhere’ experience Work for Humankind is a bold initiative in partnership with Island Conservation and the Robinson Crusoe Island community that calls on volunteers from around the world to take part in a once-in-a-lifetime opportunity: to experience first-hand how to make a long-lasting difference with a remote island community,

Written by

Claudio Uribe

Est. reading time

9 minutes

Original URL: https://www.islandconservation.org/work-for-humankind-lenovo-invites-you-to-work-from-one-of-earths-most-remote-locations-with-smarter-technology/

😍 Sara Beery, Elijah Cole (Deactivated), Jorrit van Gils, Oc, Arthur Wandzel

Justin Kay (justinkay92@gmail.com)

2021-12-03 16:47:45

Do you think Lenovo would be open to paying participants for their conservation work? The work Island Conservation is doing is amazing and the conservation goals here are obviously very important - but if Lenovo wants to make a larger impact, it seems to me a great contribution would be to directly invest in the conservationist labor. Especially considering Lenovo has $60B+ in revenue and seems to be getting a good amount of marketing benefit out of this (their name is mentioned 26 times in this article). I also can’t help but feel the study on Gen Z work preferences is not recognizing the anxiety a lot of people face now due to lack of job security in the gig economy - being paid is also a great way to communicate that your work has value! Again, I’m all for the conservation goals, this is more so a comment on big tech getting benefit from free labor… maybe they can be convinced to invest further 🙂

🙌 Sean Carter

David Jarrett (david.jarrett@durham.ac.uk)

2021-12-04 06:43:45

Hi all, I'm @David Jarrett I'm new to the group, (it looks really interesting and useful!, thanks to @Dan Morris for the suggestion. I'm using Audiomoths to monitor breeding productivity in shorebirds (particularly Eurasian Curlew) in Scotland, aiming to develop binary classifiers using e.g. CNNs in Python over the next year or so to extract alarm calls / chick warning calls from large datasets. Really interested to talk to people who've done similar, share code etc.

A question to start - if I want to train a classifier on the call in this image attached, is there a material benefit to my labelling each individual call separately for training data (so resulting in 11 labelled pieces of data in example) as against labelling the group of calls (so having 1 piece of data here). In practise these calls most often occur in repeated sets like this, but can also occur in 1/2/3s etc.

Cheers,
David

Beckett Sterner (bsterne1@asu.edu)

2021-12-04 18:07:59

👆@Caleb Powell

Yves Bas (yves.bas@gmail.com)

2021-12-05 04:26:52

If you have little overlap in time AND frequency with other vocalisations, you can use basic automatic segmentation as a basis for a quick labelling, like in Tadarida software: https://openresearchsoftware.metajnl.com/articles/10.5334/jors.154/
You select groups of pre-segmented calls at once

openresearchsoftware.metajnl.com

Journal of Open Research Software

Article: Tadarida: A Toolbox for Animal Detection on Acoustic Recordings

Original URL: https://openresearchsoftware.metajnl.com/articles/10.5334/jors.154/

Caleb Powell (cpowel21@asu.edu)

2021-12-06 13:29:31

Sounds like a fun project! Assuming all the audio you'd like to process is of variable lengths, you might find it necessary to split the audio up into discrete pieces. With this in mind, knowing the time of each call would help you decide how best to split up the data.

If your training data is scarce, splitting it up can be useful for augmentations such as layering it over clips of environmental noise. The one catch here is you will have to take additional caution when splitting training and testing data.

Also, might be worth looking into attention modules (i.e., transformers). I've had great success in the past using them for audio processing.
Here is a source that helped me: https://arxiv.org/abs/1803.02353

arXiv.org

Multi-level Attention Model for Weakly Supervised Audio Classification

In this paper, we propose a multi-level attention model to solve the weakly labelled audio classification problem. The objective of audio classification is to predict the presence or absence of...

Original URL: https://arxiv.org/abs/1803.02353

Daniel Davila (daniel.davila@kitware.com)

2021-12-06 10:41:16

Hey folks, I wanted to share one of the projects I've been working on here at Kitware. We've built an open-source system (both software and hardware) that implements a drone payload for computer vision applications, called ADAPT. We noticed there are many groups out there spending hard-earned research dollars on re-engineering the same payload, over and over, that others have largely built already in their own silo. Our goal is to provide an MVP which reduces this redundant NRE required to field an AI-enabled drone application. In order to fly your own AI mission with high-quality georegistered data collection or analytics, you simply need to purchase the parts that we've curated and integrated, follow the build instructions, and bring your own AI model. This leaves more resources to municipalities, universities, and other stakeholders for the data-driven aspects of the mission.

As many of you know, we are an open source scientific computing company so this is not a product but rather a result of our research efforts funded through NOAA, which we make free to the community. We are currently seeking feedback on the payload, as well as stakeholders who might want to fly an upcoming mission and let us know how it goes. We’ve already proven out the system with our friends at the University of Alaska-Fairbanks, who built and flew an ice reconnaissance / data collection mission for just $10k (vs $100k+ from scratch). Please find more details about our project, including parts list and build instructions here: https://kitware.github.io/adapt/

Thanks yall! - Dan

kitware.github.io

ADAPT

Documentation for the ADAPT payload

Original URL: https://kitware.github.io/adapt/

😎 Jon Van Oast, Emilio Luz-Ricca, Suhail Alnahari

👏 Malte Pedersen, Juan Arrechea, Howard L Frederick

Ben Weinstein (benweinstein2010@gmail.com)

2021-12-06 11:55:04

@Heather Lynch

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2021-12-06 17:59:16

Hello! Can anyone recommend papers to read for state-of-the-art ML processing for camera trap videos? Looking for existing approaches, open source libraries, segmentation experiments, etc. Thanks in advance!

🎉 Sara Beery, Jon Van Oast

Peter Bull (peter@drivendata.org)

2021-12-06 18:08:06

Hey @Jason Holmberg (Wild Me), hope you’re doing great! There’s not a ton of work at exactly the intersection of camera traps + video, but I’d be happy to share what we learned in our latest round of reviewing things and implementing zamba.

Peter Bull (peter@drivendata.org)

2021-12-06 18:08:48

Shoot me an email and we can find a time

👍 Jason Holmberg (Wild Me)

Dan Morris (agentmorris@gmail.com)

2021-12-12 13:43:03

FYI we have a bit of experience with using MegaDetector on videos... what we do is wildly inelegant (sample frames, run MD, do something sensible to aggregate the results from frames back up to videos), but has been useful to a number of users. Timelapse (which is how most of our users consume MD results) has good support for this, so it didn't require any new UI tooling. So, this is at least an "existing approach". Whether this is "state-of-the-art" is left as an exercise for the reader. 🙂

https://github.com/microsoft/CameraTraps/blob/master/detection/process_video.py

<https://github.com/microsoft/CameraTraps/blob/master/detection/process_video.py | process_video.py>

<pre><code>###### # # process_video.py # # Split a video into frames, run the frames through run_tf_detector_batch.py, and # optionally stitch together results into a new video with detection boxes. # ###### #%% Constants, imports, environment import os import tempfile import shutil import argparse from detection import run_tf_detector_batch from visualization import visualize_detector_output from ct_utils import args_to_object from detection.video_utils import video_to_frames from detection.video_utils import frames_to_video from detection.video_utils import find_videos #%% Main function class ProcessVideoOptions: model_file = '' input_video_file = '' output_json_file = None output_video_file = None render_output_video = False delete_output_frames = True reuse_results_if_available = False confidence_threshold = 0.8 n_cores = 1 debug_max_frames = -1 def process_video(options): if options.output_json_file is None: options.output_json_file = options.input_video_file + '.json' if options.render_output_video and (options.output_video_file is None): options.output_video_file = options.input_video_file + '.detections.mp4' tempdir = os.path.join(tempfile.gettempdir(), 'process_camera_trap_video') frame_output_folder = os.path.join( tempdir, os.path.basename(options.input_video_file) + '_frames') os.makedirs(frame_output_folder, exist_ok=True) frame_filenames, Fs = video_to_frames( options.input_video_file, frame_output_folder) image_file_names = frame_filenames if options.debug_max_frames &gt; 0: image_file_names = image_file_names[0:options.debug_max_frames] if options.reuse_results_if_available and \ os.path.isfile(options.output_json_file): print('Loading results from {}'.format(options.output_json_file)) results = None else: results = run_tf_detector_batch.load_and_run_detector_batch( options.model_file, image_file_names, confidence_threshold=options.confidence_threshold, n_cores=options.n_cores) run_tf_detector_batch.write_results_to_file( results, options.output_json_file, relative_path_base=frame_output_folder) if options.render_output_video: # Render detections to images rendering_output_dir = os.path.join( tempdir, os.path.basename(options.input_video_file) + '_detections') detected_frame_files = visualize_detector_output.visualize_detector_output( detector_output_path=options.output_json_file, out_dir=rendering_output_dir, images_dir=frame_output_folder) # Combine into a video frames_to_video(detected_frame_files, Fs, options.output_video_file) if options.delete_output_frames: shutil.rmtree(tempdir) return results #%% Interactive driver if False: #%% Process a folder of videos import re model_file = r'c:\temp\models\md_v4.1.0.pb' input_dir = r'C:\temp\bellevue_test\videos' video_files = find_videos(input_dir) video_files = [s for s in video_files if 'detections' not in s] # input_video_file = video_files[1] for input_video_file in video_files: print('Processing file {}'.format(input_video_file)) options = ProcessVideoOptions() options.reuse_results_if_available = True options.model_file = model_file options.input_video_file = input_video_file options.output_video_file = re.sub('.avi','.detections.mp4',input_video_file,flags=re.I) options.output_json_file = re.sub('.avi','.detections.json',input_video_file,flags=re.I) options.render_output_video = True options.n_cores = 1 options.confidence_threshold = 0.55 options.delete_output_frames = True process_video(options) #%% Load video and split into frames model_file = r'c:\temp\models\md_v4.0.0.pb' input_video_file = r'C:\temp\LIFT0003.MP4' options = ProcessVideoOptions() options.model_file = model_file options.input_video_file = input_video_file options.debug_max_frames = 10 process_video(options) # python process_video.py "c:\temp\models\md_v4.0.0.pb" "c:\temp\LIFT0003.MP4" --debug_max_frames=10 --render_output_video=True #%% Command-line driver def main(): parser = argparse.ArgumentParser(description=('Run the MegaDetector each frame in a video, optionally producing a new video with detections annotated')) parser.add_argument('model_file', type=str, help='MegaDetector model file') parser.add_argument('input_video_file', type=str, help='video file to process') parser.add_argument('--output_json_file', type=str, default=None, help='.json output file, defaults to [video file].json') parser.add_argument('--output_video_file', type=str, default=None, help='video output file, defaults to [video file].mp4') parser.add_argument('--render_output_video', type=bool, default=False, help='enable/disable video output rendering (default False)') parser.add_argument('--delete_output_frames', type=bool, default=False, help='enable/disable temporary file detection (default True)') parser.add_argument('--confidence_threshold', type=float, default=0.8, help="dont render boxes with confidence below this threshold") parser.add_argument('--n_cores', type=int, default=1, help='number of cores to use for detection (CPU only)') parser.add_argument('--debug_max_frames', type=int, default=-1, help='trim to N frames for debugging') args = parser.parse_args() options = ProcessVideoOptions() args_to_object(args,options) process_video(options) if __name__ == '__main__': main() </code></pre>

👍 Jason Holmberg (Wild Me)

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2021-12-07 11:13:28

Data4Wildlife challenge: https://www.bright-tide.co.uk/data4wildlifechallenge

Historically, the top three factors driving the global biodiversity extinction crisis have been habitat loss, human-wildlife conflict, and poaching. Over about the past decade, a new factor has propelled into the spotlight: online crime. Online wildlife crime causes species declines. The algorithmic amplification of wildlife crime via social media platforms, mobile applications and high-speed broadband access is a grave concern for nongovernmental organizations working to combat online crime. Join us for the #DATA4WILDLIFE challenge and help slow species declines!
The #DATA4WILDLIFE challenge will take place on 29/30 January 2022. Applications are now open and will close at 22.00 on 12 January 2021. Teams will be announced on 17 January 2022 and drop in mentorship sessions will be held during that week to prepare teams for the two day challenge on 29/30 January 2022.

Bright Tide

Data4Wildlife Challenge | Bright Tide

Original URL: https://www.bright-tide.co.uk/data4wildlifechallenge

🙌 Sara Beery, Jorrit van Gils

👀 Cameron Trotter

Sara Beery (sara.beery@gmail.com)

2021-12-07 14:15:49

https://www.nytimes.com/2021/09/30/opinion/animal-extinction.html?smid=tw-share

The New York Times

By Henry M. Paulson Jr.

We’re Living Through One of the Most Explosive Extinction Episodes Ever

The growing damage to the world’s biodiversity presents dire risks to human societies.

Original URL: https://www.nytimes.com/2021/09/30/opinion/animal-extinction.html?smid=tw-share

👀 Burak Ekim, Omiros Pantazis

🙁 Benjamin Kellenberger, Armin Bazarjani

Juan Arrechea (juan.arrechea.conservation@gmail.com)

2021-12-08 12:33:20

Hi all, I'm @Juan Arrecheanew to this group and already amazed with all the awesome stuff posted here! I am CS student at UT Dallas looking to learn more about all things related to computer science in conservation. I work at the Heard Natural Science Museum & Wildlife Sanctuary in sanctuary management/animal care. If anyone is near by please come visit! Huge thanks to @Gracie Ermi for sending me an invite.

👋 Sara Beery, Justin Kay, Jason Holmberg (Wild Me), Alex Borowicz, Benjamin Kellenberger, Gracie Ermi, Declan, Jon Van Oast

Mohit Dubey (mohit.dubey96@gmail.com)

2021-12-08 18:33:24

Hey everyone,

Mohit Dubey (mohit.dubey96@gmail.com)

2021-12-08 18:34:02

I'm @Mohit Dubey and I am a post-bac researcher at Los Alamos National Labs where I do robust machine learning for data from the Mars Rover ChemCam. Applying for PhD's in Climate/ML and looking forward to connecting to you all!

👋 Sara Beery, Declan, Jason Holmberg (Wild Me), Daniel Davila, Jon Van Oast, Ankita Shukla

Sara Beery (sara.beery@gmail.com)

2021-12-10 12:04:39

Rainforest XPRIZE is looking for a Tech Lead! https://www.linkedin.com/jobs/view/technical-lead-xprize-rainforest-at-xprize-2828856128

👍 Jon Van Oast

🎉 Jon Van Oast

Carl Boettiger (cboettig@berkeley.edu)

2021-12-14 16:14:49

Hi all -- a question for those here in any non-academic career trajectory: undergrads in my courses (i.e. environmental data science classes at Berkeley) will often ask for advice about careers: what background/experience/degrees they should pursue. What advice would you give? I feel I've a good grasp on describing the good & bad of academia and what's expected, and to some extent a grasp on federal agencies like NOAA/EPA, but increasingly out of my depth in private sector. What skills & experiences do you look for? What do you wish someone had told you as an undergraduate?

🙌 Sara Beery, Britney Muller

❤️ Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Britney Muller

Carl Boettiger (cboettig@berkeley.edu)

2021-12-14 18:39:08

Also realize that many of you in academic positions have a lot more experience with industry careers and with answering this kind of question for students than I do, so y'all fire away here as well please!

Daniel Davila (daniel.davila@kitware.com)

2021-12-15 09:54:19

Hey again Carl! Ive been working for industry (mostly R&D groups) for while now, and at various points responsible for interviewing and hiring. Here are some general thoughts, hope they help.

I do not speak for any company in particular, but when evaluating non-researcher MLEs (researchers typically have a mandate very similar to their counterparts in academia), I am typically looking for people with a track record of solving interesting, application-minded problems. A solid grasp of the fundamentals is necessary, but what I usually want to see is evidence that a candidate can apply all that useful theory they learn in schools to actually solve an open ended customer problem. This can be shown with project work, contributing to open source tools, making an impact at a past internships, etc... Degrees I typically look for are the traditional STEM ones for this field (EE, CS, CompE, etc..). Other majors, including math, applied math, physics, other engineering disciplines (e.g. BME), and others arent disqualifiers but sometimes these candidates are weaker on the core tools needed for the job (like programming) and need to demonstrate that experience through prior project work as I mentioned. Non-academic project work also usually implies they have experience with programming languages in the wild, working on a team, and delivering under customer imposed deadlines. Also, it helps not to be an entitled jerk. 🙂 Cant tell you how many students from top programs Ive met who throw their resume in my face and expect a job offer on pedigree alone. Insta-no

👍 Carl Boettiger

Brian Cohen (bcohen@tnc.org)

2021-12-15 13:59:43

If the students are interested in conservation/environmental fields, getting involved now as a volunteer is a great way to build experience specific to how agencies and organizations in that world operate, and even more importantly a great way to build a professional network that is just as important as the resume. The volunteer work doesn't have to be specific to tech, you just want to get involved in whatever the work is (habitat restoration, river cleanups, species protection, clean water, affordable housing, etc.) and then help the orgs figure out how tech can make that work more efficient, scalable, etc.

👍 Carl Boettiger, Sara Beery

Arthur Wandzel (arthur.wandzel@gmail.com)

2022-01-16 17:53:55

Maybe recommend Designing your Life as a general resource for designing a nonstandard career path?

Hannah Yin (hannah.yin@rice.edu)

2022-02-21 16:03:18

As someone who recently graduated from college with some (limited) industry experience, I'd say networking is key. Don't be afraid to cold email someone for an informational interview or chat over tea/coffee/video/phone if they're part of a startup/company working on sustainability and conservation.

👍 Carl Boettiger

Kakani Katija (kakani@mbari.org)

2021-12-16 17:25:13

Hi everyone!! I'm Kakani Katija, a bioengineer at MBARI (Monterey Bay Aquarium Research Institute) where I lead the Bioinspiration Lab (we do lots of underwater imaging, robotics, bioinspired design). We (along with LOADS of collaborators) recently launched FathomNet, an underwater image database that contains labeled data for ocean life that will hopefully grow with community contributions. Check it out (and the blog and github) and let us know where we can do better. If you're interested in the ocean space, I'm also happy to chat!

bioinspirationlab

Home | MBARI's Bioinspiration Lab | Moss Landing, CA

The Bioinspiration Lab develops new deep sea exploration tools, utilizing advanced imaging and underwater robotics, to observe marine organisms in their natural environment.

Original URL: https://www.bioinspirationlab.org

Medium

FathomNet – Medium

Powering Ocean Vision AI since 2021.

Original URL: https://medium.com/fathomnet

🎉 Daniel Grzenda, Sara Beery, Jason Holmberg (Wild Me)

🦑 Daniel Grzenda, Sara Beery, Frederic Fol Leymarie, Jason Holmberg (Wild Me), Ben Weinstein

Ted Schmitt (teds@allenai.org)

2021-12-16 18:51:22

Please share this fantastic internship opportunity with anyone you think may be interested
https://boards.greenhouse.io/thealleninstitute/jobs/3738739

boards.greenhouse.io

Applied Sciences and Technology - Artificial Intelligence Internship, Skylight Team

Seattle

Original URL: https://boards.greenhouse.io/thealleninstitute/jobs/3738739

😍 Sara Beery, Jason Holmberg (Wild Me), Carly Batist

👍 Jon Van Oast, Charlotte, Jason Holmberg (Wild Me), Monty Ammar

Ben Weinstein (benweinstein2010@gmail.com)

2021-12-17 17:12:13

I just got this message from the Toronto Zoo, anyone know of a GUI that might help them. @Petar Gyurov

I'm part of the Adopt-A-Pond program with the Toronto Zoo, and we set out camera traps over the summer to capture photos of animals who may be predating wild turtle populations. 

We have over 400,000 photos to look through so we're trying to find something to help us out with the analysis.

Ben Weinstein (benweinstein2010@gmail.com)

2021-12-17 17:12:48

if any students here want to take on a small project.

👍 Fuzail Dawood

Elijah Cole (Deactivated) (ecole@caltech.edu)

2021-12-17 19:17:58

@Benjamin Kellenberger Maybe AIDE is a good fit?

Caleb Robinson (calebrob6@gmail.com)

2021-12-17 20:01:16

I think the Microsoft AI4E camera trap classifier lives on with Dan Morris -- he might be able to help them find which images have any animals

Caleb Robinson (calebrob6@gmail.com)

2021-12-17 20:33:31

@Dan Morris

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2021-12-18 01:42:53

Good opportunity to ﬁnally combine MegaDetector (MD) with AIDE? It’s been on my list ever since. Pre-processing with MD and import into AIDE is another option that is almost immediately available.
I’d be happy to help out if you ﬁnd my software to be a potential candidate!

petargyurov/megadetector-gui

Tools for detecting wildlife in aerial images using active learning

Stars

144

Language

Python

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-12-19 02:20:02

Megadetector or Zamba are good for this. @Sara Beery

Petar Gyurov (pgyurov93@gmail.com)

2021-12-20 05:50:09

@Ben Weinstein you could point them to MegaDetector GUI . It depends if MegaDetector fits their needs in the first place. I haven't made updates to that project in some time but it's in a relatively stable state (I think there are some bugs surrounding metadata export). Feel free to point them to me if they want to discuss things. Cheers.

A desktop application that makes using MegaDetector's model easier

Stars

Language

JavaScript

👍 Benjamin Kellenberger, Ben Weinstein

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2021-12-20 05:51:29

Cool project @Petar Gyurov! Didn’t know about it before.

👍 Petar Gyurov, Sara Beery

Dan Morris (agentmorris@gmail.com)

2022-01-10 12:09:52

@Ben Weinstein MegaDetector may be helpful here, but I agree with Petar that it’s helpful to determine whether MegaDetector fits their needs in terms of accuracy and image distribution, and if it does, we can help them find a workflow to use the results. We try to “fail fast” if MegaDetector isn’t going to work for someone, so feel free to have them email cameratraps@lila.science and we’ll help them figure it out. If you’re interested in following along, fine to to have them cc you as well. Thanks!

Carly Batist (cbatist@gradcenter.cuny.edu)

2021-12-19 03:35:49

WILDLABS’ annual state of conservation tech survey is still accepting submissions!
“The aim of the project is to advance efforts by capturing recent progress, key constraints, and critical opportunities for growth in leveraging technology for conservation. We are seeking input from individuals who have experience developing, using, or otherwise engaging with technology for conservation purposes.”
https://colostate.az1.qualtrics.com/jfe/form/SV_aW6C50iOOzOPsB8

colostate.az1.qualtrics.com

WILDLABS Conservation Tech Survey 2021

Contribute your thoughts to an evaluation of the State of Conservation Technology by researchers at WILDLABS & Colorado State University

Original URL: https://colostate.az1.qualtrics.com/jfe/form/SV_aW6C50iOOzOPsB8

👍 Fuzail Dawood, Sara Beery

Monty Ammar (montyx23@gmail.com)

2021-12-21 18:43:32

Hello everyone, I’m Monty Ammar.
Im currently doing a Conservation Biology MSc at DICE University of Kent, and I have just found the most incredible community to be a part of!

My background is Animal Biology & Wildlife Conservation, I am doing a dissertation on understanding the effects of small scale gold mining in Amazonia on forest regeneration using ML. I am also writing a special topics module paper reviewing ‘The use of Deep Learning in Conservation Biology: consolidating progress and identifying research gaps’. Any pointers to already published review papers investigating the same topic would be appreciated!

I am so excited to be a part of this community and totally overwhelmed with the amount of useful resources being posted here. 😁

😍 Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Jason Holmberg (Wild Me), Juan Arrechea

🎉 Mark Roth

👍 Oisin Mac Aodha

Charlotte (chalange@uos.de)

2021-12-25 11:24:54

Hello Monty, I stumbled about this very recent review (October 21) called "Seeing biodiversity: perspectives in machine learning for wildlife conservation" when I got interested in the same topic. They already did an amazing job and I think it might be of help to you. You can find it here:
https://arxiv.org/abs/2110.12951

Merry Christmas!

arXiv.org

Seeing biodiversity: perspectives in machine learning for wildlife...

Data acquisition in animal ecology is rapidly accelerating due to inexpensive and accessible sensors such as smartphones, drones, satellites, audio recorders and bio-logging devices. These new...

Original URL: https://arxiv.org/abs/2110.12951

👍 Monty Ammar

❤️ Sara Beery

Monty Ammar (montyx23@gmail.com)

2021-12-26 19:43:56

Merry Christmas! Thanks for this, it looks like a fantastic review. Will definitely be useful for me!

Dan Morris (agentmorris@gmail.com)

2022-01-10 12:19:37

Though it’s specific to AI for camera traps, you may find the list of papers (most with one-paragraph summaries) here to be useful:

https://agentmorris.github.io/camera-trap-ml-survey/#camera-trap-ml-papers

Camera Trap ML Survey

Overview

Everything I know about machine learning and camera traps.

Original URL: https://agentmorris.github.io/camera-trap-ml-survey/#camera-trap-ml-papers

👍 Monty Ammar, Sara Beery

Monty Ammar (montyx23@gmail.com)

2022-01-10 12:27:11

This is vastly useful, thanks Dan 👌

Sara Beery (sara.beery@gmail.com)

2022-01-05 19:48:50

Biden-Harris Administration Invites Public Comment on Development of New Conservation and Stewardship Tool

Sierra Sun Times

Biden-Harris Administration Invites Public Comment on Development of New Conservation and Stewardship Tool

Original URL: https://goldrushcam.com/sierrasuntimes/index.php/news/local-news/35790-biden-harris-administration-invites-public-comment-on-development-of-new-conservation-and-stewardship-tool

💜 Arjun Subramonian (they/them), Omiros Pantazis, Kakani Katija, Jason Holmberg (Wild Me), David Russell, Yuval Boss, Armin Bazarjani, Dhruv Sheth, Carly Batist

Sara Beery (sara.beery@gmail.com)

2022-01-13 19:04:59

The CompSust Doctoral Consortium is an awesome place to present academic work in AI for Conservation and grow your community, applications due January 28! http://www.compsust.net/compsust-2022/

👍 Ankita Shukla, Bistra Dilkina, Dhruv Sheth, Justin Kay, Jason Holmberg (Wild Me), Mark Roth, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Carly Batist, Chris Yeh

🎉 Jon Van Oast, Dhruv Sheth, Jason Holmberg (Wild Me), Suzanne Stathatos, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Chris Yeh

Jason Parham (bluemellophone@gmail.com)

2022-01-14 02:31:27

rubbing hands
giphy logo
Posted using /giphy

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2022-01-14 12:19:05

^also open to the broader community beyond PhD students, including undergrad, master's, and postdocs as well as folks outside academia!

❤️ Sara Beery, Monty Ammar

Sara Beery (sara.beery@gmail.com)

2022-01-14 16:50:36

@Jake Wall and I brainstormed this joint local/international conservation technology internship at the Mara Elephant Project while we were in the field last August, and I'm so excited to see it come to fruition thanks to Ai2! If you have a student who might be interested, please pass the information along! https://boards.greenhouse.io/thealleninstitute/jobs/3820329

boards.greenhouse.io

Applied Science and Technology - Conservation Technology Internship in collaboration with Mara Elephant Project

Seattle

Original URL: https://boards.greenhouse.io/thealleninstitute/jobs/3820329

❤️ Jon Van Oast, Bistra Dilkina, Ted Schmitt, Justin Kay, Suzanne Stathatos, Arjun Subramonian (they/them), Lucia Gordon, Mitch Fennell, Megan Cromp, Dhruv Sheth, Phuc Le, Talia Speaker, Avi Sundaresan, Armin Bazarjani, Catherine Villeneuve

💯 Carly Batist, Jason Parham

Sara Beery (sara.beery@gmail.com)

2022-01-14 16:51:07

@Bistra Dilkina you were looking for undergraduate opportunities

❤️ Bistra Dilkina

Gracie Ermi (gracieermiifthen@gmail.com)

2022-01-14 16:53:28

Whoa!! What a cool opportunity!

❤️ Sara Beery

Ted Schmitt (teds@allenai.org)

2022-01-14 17:03:43

Kudos as well to @Jes Lefcourt for making this happen!

💯 Sara Beery, Gracie Ermi, Jake Wall

Sara Beery (sara.beery@gmail.com)

2022-01-14 17:05:34

Yes absolutely!!

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-01-15 00:12:21

What are the start dates? And is there an application deadline? Is it paid/are expenses like flight/lodging covered?

Sara Beery (sara.beery@gmail.com)

2022-01-15 08:05:59

@Ted Schmitt can probably best answer re: the logistical details from the Ai2 side

👍 Carly Batist

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-01-16 00:07:50

Ok, asked about it on twitter too & didn’t get a reply. Seems like some important details…

Sara Beery (sara.beery@gmail.com)

2022-01-16 05:47:55

I agree! I just don't know the answers :)

👍 Carly Batist

Sara Beery (sara.beery@gmail.com)

2022-01-16 05:48:33

Though I'm 99.9% sure that it's paid and travel is covered

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-01-16 07:38:47

Yeah I figured it would be given the funders behind it. I’ll wait for Ted’s response. Thanks Sara!

Jes Lefcourt (jeslefcourt@gmail.com)

2022-01-18 08:53:05

Sorry, I was double-confirming the details. The start date is flexible, and all expenses are paid.

❤️ Sara Beery, Carly Batist

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-01-18 12:07:13

Ok great, thanks for the clarification Jes!

Charlotte (chalange@uos.de)

2022-01-19 05:25:33

i know the start date is flexible, but is there still an application deadline?

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-01-16 00:44:08

“TorchGeo is a PyTorch domain library, similar to torchvision, that provides datasets, transforms, samplers, and pre-trained models specific to geospatial data.
The goal of this library is to make it simple:

for machine learning experts to use geospatial data in their workflows, and

for remote sensing experts to use their data in machine learning workflows.”

https://github.com/microsoft/torchgeo

microsoft/torchgeo

TorchGeo: datasets, transforms, and models for geospatial data

Stars

540

Language

Python

👍 Sara Beery, Oisin Mac Aodha, Cameron Trotter, Ritwik

Caleb Robinson (calebrob6@gmail.com)

2022-01-17 14:09:08

Hey @Carly Batist, I work on torchgeo 🙂, please don't hesitate to reach out if you have any questions, comments, or something is broken!

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-01-17 14:13:33

Ah so sorry I didn’t tag you in the original post😬, didn’t realize you were already here! Saw it on Twitter and thought it would super useful

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-01-17 10:10:10

💥New year, new update to the Conservation Tech Directory! 577 entries now🥳🤩. @Gracie Ermi
https://conservationtech.directory/

Reminder - now with academic lab as resource type. If you’ve got a #ConservationTech lab, please add it! (link below)
Google form to add new resources. And updated static doc/PDF on figshare.

conservationtech.directory

Applied Science and Technology - Conservation Technology Internship in collaboration with Mara Elephant Project

Visit the post for more.

Original URL: https://conservationtech.directory/

😍 Sara Beery

🎉 Megan Cromp, Omiros Pantazis, Lucia Gordon, Agnethe Seim Olsen, Talia Speaker, Gracie Ermi, Howard L Frederick

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2022-01-19 01:00:53

A wonderful internship opportunity at the Allen Institute to work on their Mara Elephant Listening Project

https://boards.greenhouse.io/thealleninstitute/jobs/3820329

boards.greenhouse.io

Seattle

Original URL: https://boards.greenhouse.io/thealleninstitute/jobs/3820329

👋 Jason Holmberg (Wild Me), Lloyd Hughes, Monty Ammar

Catherine Villeneuve (catherine.villeneuve.9@ulaval.ca)

2022-01-20 19:07:35

Hi all! I’m new to this group. I’m a master’s student in Computer Science in Canada. I’m currently working on ML-based Arctic foxes movement models, to further our understanding of how predation influences prey distribution in a dynamic Arctic. I also work with snow geese, snowy owls, and lemmings. Is there anyone else here working on the Arctic? Or whose native language is French? In any case, I’m very happy to be part of this community. Looking forward to connecting to you all!

🐻‍❄️ Jason Holmberg (Wild Me), Dhruv Sheth, Mark Roth, Anthony Bao

❤️ Lucia Gordon, Sara Beery, Catherine

👍 Rhea Urquhart

Alexandre Tytgat (alextytgat@gmail.com)

2022-01-22 10:01:50

Hey everyone! I heard about this community from yesterday talk organized by Climate change AI (which was great!), and it looks amazing:)
A bit about myself: I graduated last September from a data science master, but I started my studies with a BSc in physics (a subject I'm still passionate about). Now I'm turning my sight to what to do next, and one of my interest is the application of AI to help efforts in domains such as climate change mitigation/adaptation, biodiversity protection, and extreme poverty reduction. Hence why I joined this community 😉 Also, I'm from Belgium and speak French (as I saw @Catherine Villeneuve asking if anyone was a native in the language:))
I look forward to chat and share great resources on the subject of AI for conservation with all of you!

👋 Oisin Mac Aodha, Catherine Villeneuve, Mitch Fennell, Thijs, Mark Roth, Sara Beery, Juan Arrechea, Jason Holmberg (Wild Me), Dhruv Sheth, Monty Ammar, Anthony Bao

❤️ Catherine Villeneuve, Jason Holmberg (Wild Me), Catherine

Ben Weinstein (benweinstein2010@gmail.com)

2022-01-25 23:20:50

I am getting conflicting information on whether how/whether we can download google satellite images and use them in publications/training models. Maybe @Sara Beery and can get me an answer here. I see two different licenses with opposite interpretations. I saw this (https://www.mdpi.com/2072-4292/14/3/476) today which uses our lab's data + some google earth they scraped at 11cm.

MDPI

Transformer for Tree Counting in Aerial Images

The number of trees and their spatial distribution are key information for forest management. In recent years, deep learning-based approaches have been proposed and shown promising results in lowering the expensive labor cost of a forest inventory. In this paper, we propose a new efficient deep learning model called density transformer or DENT for automatic tree counting from aerial images. The architecture of DENT contains a multi-receptive field convolutional neural network to extract visual feature representation from local patches and their wide context, a transformer encoder to transfer contextual information across correlated positions, a density map generator to generate spatial distribution map of trees, and a fast tree counter to estimate the number of trees in each input image. We compare DENT with a variety of state-of-art methods, including one-stage and two-stage, anchor-based and anchor-free deep neural detectors, and different types of fully convolutional regressors for density estimation. The methods are evaluated on a new large dataset we built and an existing cross-site dataset. DENT achieves top accuracy on both datasets, significantly outperforming most of the other methods. We have released our new dataset, called Yosemite Tree Dataset, containing a 10 km2 rectangular study area with around 100k trees annotated, as a benchmark for public access.

Original URL: https://www.mdpi.com/2072-4292/14/3/476

Sara Beery (sara.beery@gmail.com)

2022-01-25 23:28:26

Google seems to have really complex policies, I can try to ask.

👍 Ben Weinstein

aruna (arunas@mit.edu)

2022-01-26 08:50:38

Aside: Another imagery option is to request the ESA for data from the Pleiades satellites: https://www.intelligence-airbusds.com/imagery/constellation/pleiades/. The satellites offer high resolution imagery (50cm/px iirc and a 26 day revisit cycle). The data is free for researchers if accessing areas within a certain size limit.

intelligence-airbusds.com

Pléiades | Very High-Resolution (50cm) Satellites

Pléiades offers daily revisit capacity to any location on Earth and has outstanding acquisition capabilities to meet civil and military requirements

Original URL: https://www.intelligence-airbusds.com/imagery/constellation/pleiades/

Ben Weinstein (benweinstein2010@gmail.com)

2022-01-26 12:00:46

thanks @aruna!, this is actually on my to do list, I know it went up last year, we have been purchasing from MAXAR. Do you have any experience with these data? For example, PLANET claims 50cm data, but it turns out that is a bit of a stretch, its 65cm data sharpened to 50cm through interpolation. I had a long discussion with their technical team.

👍 Sara Beery

aruna (arunas@mit.edu)

2022-01-26 21:42:36

Hi Ben! No, unfortunately I don't have any experience with the actual data itself. Definitely sounds like you have spent more time with the team. have you had a chance to look at the actual data?

Sara Beery (sara.beery@gmail.com)

2022-01-26 09:45:27

https://latamt.ieeer9.org/index.php/transactions/announcement/view/41
CALL FOR PAPERS
IEEE Latin America Transactions
Special Issue on AI for Sustainability
We are at the cusp of two massive historic trends intimately intermingled. On the one hand, our technological revolutions have increased our standards of living quality, including a longer lifespan and larger wealth. On the other hand, we are amid sustainability threats, including climate change and loss of biodiversity. The way our generation responds to the latter challenge using our considerable accumulated level of understanding will have a significant impact in generations to come.
Artificial Intelligence (AI) for sustainability has emerged as a powerful tool with substantial groundbreaking advances. It has opened countless avenues to improve the understanding of the underlying problems while supporting effective sound solutions. However, to take full advantage of the available science and technology, we need to increase their visibility to stimulate its widespread adoption. AI refers to the development of machine capabilities such as learning, reasoning, knowledge representation, planning, perception, problem-solving, and pattern recognition. Due to the potential for broad impact of these capabilities, it is imperative to support sustainability efforts to extend their adoption and promote awareness about their potential. In particular, we call for participation through articles addressing issues along the following lines: a) Applications of AI tackling climate change, addressing biodiversity loss, and considering human vulnerability; b) research describing the construction, implementation, evaluation, and how-to-do capabilities supporting sustainability; and c) making sure AI leaves a small impact as possible on sustainability.
The purpose of this Special Issue is to provide a forum for the researchers and practitioners related to the rapidly developing field of AI for sustainability to share their novel and original research on the topic. Therefore, we encourage the submission of the practitioners’ latest unpublished work on AI for sustainability. Of our particular interest are the developments from research groups in Latin America and elsewhere. The areas of interest include both theory and applications of AI on the following topics (they are not limited to):

Wild animal and plants identification and monitoring.

Electric systems. Usage forecast, greenhouse gases leaking, models with small datasets

Efficient transportation models, smart shipping routing, optimal bike-sharing distribution

Buildings and cities. Cooling and heating control, lighting adaptation, vehicular traffic analysis

Food waste, cement and ammonia reduction,

Farms and forests. Gases, agriculture, pipelines leaking, carbon, deforestation remote sensing

Carbon dioxide removal. Identification and monitoring of placement sites.

Weather prediction. Improved weather prediction models, long-term and reliable climate monitoring, ocean reflectivity, and warming monitoring.

Social impacts. Adaptation, interaction with the environment.

Carbon offset pricing, sustainability risk assessment, biodiversity, and climate change impacts.

Call for papers: December 13, 2021
Submissions Deadline: March 21, 2022
Notification of Acceptance: May 23, 2022
Final Manuscript Due: July 4, 2022
Publication Date: September 2022

👍 Oisin Mac Aodha, Jason Parham, Dhruv Sheth, Anthony Bao

:thumbsup_all: Frederic Fol Leymarie, Dhruv Sheth

Oisin Mac Aodha (macaodha@caltech.edu)

2022-01-26 09:56:56

Apologies for the double post, but I see that not everyone is in the "upcoming_events" channel.

We are pleased to announce the upcoming 9th Workshop on Fine-Grained Visual Categorization (FGVC9) which will be held in conjunction with CVPR 2022 this June.

Paper submission deadline is 25th March 2022

Website with more information is here: https://sites.google.com/view/fgvc9

Historically, we have had a great turn out from people in the intersection of conservation and technology. If you are working in that space and have applications in image understanding, please consider submitting your work.

👍 Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Armin Bazarjani, gvanhorn

🐪 Omiros Pantazis

🐫 Omiros Pantazis

Oisin Mac Aodha (macaodha@caltech.edu)

2022-01-26 09:57:08

More info:

FGVC9 - The Ninth Workshop on Fine-Grained Visual Categorization
19th June 2022 @ CVPR 2022
Website:  https://sites.google.com/view/fgvc9
Twitter: @fgvcworkshop
Email: fgvcworkshop@googlegroups.com

This workshop brings together researchers to explore visual recognition across the continuum between basic level categorization (object recognition) and identification of individuals (face recognition, biometrics). Participants are encouraged to submit short papers and to take part in a set of competitions organized in conjunction with the workshop - details below. We will also have an exciting lineup of invited speakers from computer vision through to domain experts.


PAPER SUBMISSION
We invite submission of 4 page (excluding references) extended abstracts on topics related to fine-grained recognition. Reviewing of abstract submissions will be double-blind. The purpose of this workshop is not specifically as a venue for publication so much as a place to gather together those in the community working on or interested in FGVC. The workshop proceedings will not appear in the official CVPR 2022 workshop proceedings. Submissions of work which has been previously published, including papers accepted to the main CVPR 2022 conference are allowed.


PAPER SUBMISSION DATES
** Deadline for Submission - 25th March 2022
** Notification of Acceptance - 25th April 2022
** Camera Ready - 6th May 2022
** Submission will be via CMT

Scope
The purpose of this workshop is to bring together researchers to explore visual recognition across the continuum between basic level categorization and identification of individuals within a category population. Topics of interest include:

Fine-grained categorization
Novel datasets and data collection strategies for fine-grained categorization
Low/few shot learning
Self-supervised learning
Semi-supervised learning
Transfer-learning
Attribute and part based approaches
Taxonomic prediction
Long-tailed learning

Human-in-the-loop
Fine-grained categorization with humans in the loop
Embedding human experts' knowledge into computational models
Machine teaching
Interpretable fine-grained models

Multi-modal learning
Using audio and video data
Using meta data e.g. geographical priors
Learning shape

Fine-grained applications
Product recognition
Animal biometrics and camera traps
Museum collections
Agricultural
Medical
Fashion


COMPETITIONS
We will also be hosting several fine-grained computer vision challenges with tasks ranging from classification of attributes in art images through to classifying diseases in plants. The competitions will be hosted on Kaggle and will be announced in late February 2022.


ORGANIZERS
Sara Beery - Caltech
Serge Belongie - Cornell
Elijah Cole - Caltech
Xiangteng He - Peking University
Christine Kaeser-Chen - DeepMind
Oisin Mac Aodha - University of Edinburgh
Subhransu Maji - University of Massachusetts, Amherst
Abby Stylianou - Saint Louis University
Jong-Chyi Su - University of Massachusetts, Amherst
Grant Van Horn - Cornell
Kimberly Wilber - Google

Daniel Grzenda (grzenda@uchicago.edu)

2022-01-27 15:31:20

Anyone have a good reference or tool for converting an animal name to their scientific name? I'm working with data collected across 3 different languages (French, Spanish, English) identifying an animal by their common name and would like to standardize it to the scientific name (programmatically if possible)

Oisin Mac Aodha (macaodha@caltech.edu)

2022-01-27 15:35:40

Have you seen the iNaturalist API:
https://api.inaturalist.org/v1/docs/#!/Taxa/get_taxa

e.g. https://api.inaturalist.org/v1/taxa?q=european%20robin
returns a json file with the species name

🙏 Daniel Grzenda

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2022-01-27 15:36:02

We use the ITIS API

https://www.itis.gov/ws_description.html

itis.gov

ITIS - Web Service Description

Hosted by the USGS Core Science Analytics and Synthesis. Page designed through the cooperative efforts of interagency ITIS Teams. Point of Contact: <a href="mailto:itiswebmaster@itis.gov">itiswebmaster@itis.gov</a>.

Original URL: https://www.itis.gov/ws_description.html

🎉 Jon Van Oast, Sara Beery, Daniel Grzenda

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2022-01-27 15:36:12

Not sure if it multilingual

Beckett Sterner (bsterne1@asu.edu)

2022-01-27 19:34:28

Just keep in mind that common and scientific names frequently change meanings over time as the taxonomy is revised. Same name != same concept just like when you have to align multiple ontologies

👆 Sara Beery

Dan Morris (agentmorris@gmail.com)

2022-02-07 13:00:20

We had to solve this problem to a slightly weaker standard than others may have: we wanted consistent mapping to a common taxonomy for lots and lots of records from all over the world, but because we weren't publishing the results, it didn't have to be "right" in every single case, just "right enough" and consistent, and we had to have a way of quickly making manual corrections. E.g. we had to have a way to permanently indicate that "lion" in our Idaho data meant something different from "lion" in our Snapshot Safari data... as you can imagine, there are 1000 special cases like that. We also needed a way to quickly verify the results by looking at images, since I don't actually know the scientific names for anything, and if you're not careful, you will map "wolf" to a fish (yes, there is a fish called "wolf"). So, with a huge dose of "YMMV", we ended up with the code here:

https://github.com/microsoft/CameraTraps/tree/main/taxonomy_mapping

...mostly here:

https://github.com/microsoft/CameraTraps/blob/main/taxonomy_mapping/species_lookup.py

...which basically downloads both the iNat and GBIF taxonomy files, maps lots of common-name queries into both taxonomies (preferring iNat), and makes a big spreadsheet that makes it easy (or at least minimally painful) to double-check all the mappings and manually fill in things that failed to map to anything.

I'm not sure I would advocate anyone actually running that code, but I would definitely advocate for the overall approach, which worked well and got the manual work of mapping naming schemes for zillions of images down to maybe an hour.

<https://github.com/microsoft/CameraTraps/blob/main/taxonomy_mapping/species_lookup.py | species_lookup.py>

Kevin Webb (ktwebb86@gmail.com)

2022-01-28 10:51:27

Hi everyone, it’s great to be here! As a quick introduction, my name is Kevin Webb, and I invest in startups with positive impacts on biodiversity, so I’m excited to learn from you, share roles as they come up, and be a resource if my background can be useful. Some of the areas I’m most excited about here are AI for species and individual recognition, human-animal interaction, and the use of AI to inform conservation and climate-related decisions (like corridor placement). On the side, I also enjoy making things, and you can see past projects/writing at ktfoundry.com.

🐆 Jason Holmberg (Wild Me), Catherine Villeneuve

🎉 Jason Holmberg (Wild Me), Marcus Lapeyrolerie

Catherine Villeneuve (catherine.villeneuve.9@ulaval.ca)

2022-01-28 17:44:05

Could be interesting for some of you, a special issue from Frontiers on ML/DL for ecological monitoring : https://www.frontiersin.org/research-topics/32067/advances-in-machine-learning-and-deep-learning-for-monitoring-terrestrial-ecosystems. The deadline for the abstract is March 27, and the manuscript deadline is May 26.

Frontiers

Advances in Machine Learning and Deep Learning for Monitoring Terrestrial Ecosystems

The terrestrial biosphere provides critical functions and ecosystem services for global biogeochemical cycles and supports global socioeconomic activities. Rapid global changes in climate and human activities pose fundamental threats to these ecosystems, such as biodiversity loss, ecosystem degradation, and habitat destruction. Due to the inherent complexity of biological and ecological processes, improved ecological modeling methods are needed to better understand and predict the response of terrestrial ecosystems to future climate change.The growing availability of ecological data, facilitated by recent advances in Earth Observation data, sensor networks, and citizen science, provides great opportunities for understanding and predicting terrestrial ecosystem processes. Meanwhile, data analysis has been revolutionized by the introduction of machine learning (ML) and deep learning (DL) methods. The recent use of ML in ecological monitoring has increased tremendously, but its full potential is far from fully exploited, while DL applications remain limited.We propose this Research Topic to leverage emerging opportunities with ML and DL approaches for the analysis of ecosystem structure and function and ecological processes. Contributors to the topic will be challenged to summarize the current state-of-the-art and explore solutions to the roadblocks in ML and DL methods. We encourage accelerated advancements in identifying ecological structure and function using a wide range ...

Original URL: https://www.frontiersin.org/research-topics/32067/advances-in-machine-learning-and-deep-learning-for-monitoring-terrestrial-ecosystems

😍 Sara Beery, Dhruv Sheth, Anthony Bao, Monty Ammar

Chris Yeh (chrisyeh96@gmail.com)

2022-01-28 18:03:51

Calling all researchers working on areas related to computational sustainability! We invite you to submit your work to the 5th annual CompSust Doctoral Consortium 2022. We are also looking for students interested in giving tutorials.

When and Where: March 11-12, 2022; Virtual
Submission Deadline: extended to February 8, 2022 11:59 PM Pacific Time
Submission requirements: 2-page research abstract

More Info: http://www.compsust.net/compsust-2022/

We invite students and researchers advancing research in computational techniques with applications to sustainability-related topics, which include

Clean energy

Wildlife conservation

Food security

Public health

Public Transportation

Climate action

Disaster adaptation

... and many more!

❤️ Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Enoch Luk, Sara Beery, Anthony Bao, Ayan Mukhopadhyay, Armin Bazarjani, Chris Yeh

Carl Boettiger (cboettig@berkeley.edu)

2022-01-31 10:48:18

This upcoming (virtual) conference could be of interest to anyone working on or interested in applying ML to the digitization of biodiversity museum collections: https://www.idigbio.org/content/digital-data-2022-enhancing-advancing-quality-digitized-data

iDigBio

Digital Data 2022: Enhancing & advancing the quality of digitized data

QUICK LINKS Registration Abstract Submission Themes Speakers New for 2022 CLICK HERE TO REGISTER Sixth Annual

Original URL: https://www.idigbio.org/content/digital-data-2022-enhancing-advancing-quality-digitized-data

😍 Sara Beery, Nico Franz, Kevin Webb, Libby Ellwood, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Akronix

Beckett Sterner (bsterne1@asu.edu)

2022-02-02 14:18:11

Subject: NASA Student Internship - Seeking Applicants

Hello Biological Diversity and Ecological Forecasting Program Communities,
As you may know, we frequently host student interns supporting our programs at NASA Headquarters. This coming summer, we will again fill a post and are currently seeking applications. These are paid 10-week positions where students work remotely/virtually. The link to the position can be found here. I have pasted the position description below, Please, share this opportunity with students you think would be a good fit.

Sincerely,
Keith Gaddis

Position Description
This position will examine strategies for advancing the biological diversity and ecological forecasting programs within NASA’s Earth Science Division. Interns will advance the use of remote sensing for detecting, understanding, and forecasting patterns of life on Earth. This position will examine linkages and synergistic relationships between these programs and other activities within and outside of NASA.

Applicants with a background in conservation biology, ecology, evolution, computer programming, statistics, or communications/journalism are sought. Interns will advance outward communication of program activities, build program infrastructure, develop and implement evaluation metrics for science projects, and support science review. The mentor will develop projects specific to the intern’s background and interests on these themes.

😍 Sara Beery, Nico Franz, Catherine Villeneuve, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Jason Holmberg (Wild Me), Dhruv Sheth, Lucia Gordon

Björn Lütjens (bjoern.luetjens@gmail.com)

2022-02-03 18:05:39

Hello Everybody,

Does anybody know good ground-truth/validation datasets on wildlife, wildifire, tree crown segmentation, and/or belowground carbon quantification? We've been working on an awesome list that contains links and short decriptions to the best forest datasets out there. Thanks to contributions from this Slack there's already a lot of datasets but we're still missing links on those topics. I'd love to hear your finds, thank you so much! :):)

https://github.com/blutjens/awesome-forests

blutjens/awesome-forests

🌳 A curated list of ground-truth forest datasets for the machine learning and forestry community.

Stars

Last updated

4 days ago

❤️ Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Emmanuel Dufourq, David

🌳 Emmanuel Dufourq

Sara Beery (sara.beery@gmail.com)

2022-02-03 18:07:35

There are lots of great wildlife datasets on https://lila.science/

plus the iWildCam competition datasets https://github.com/visipedia/iwildcam_comp

And the iNaturalist datasets: https://github.com/visipedia/inat_comp

visipedia/iwildcam_comp

Stars

114

Last updated

a day ago

visipedia/inat_comp

Stars

537

Language

Python

🎉 Björn Lütjens

🐝 Björn Lütjens

Ben Weinstein (benweinstein2010@gmail.com)

2022-02-03 18:38:53

@Björn Lütjens does this relate to https://arxiv.org/abs/2112.00570, I was literally just emailing Alex and ccing @Sara Beery. We are getting some data out the door and maybe into torchgeo (@Caleb Robinson) spoke yesterday.

arXiv.org

Toward Foundation Models for Earth Monitoring: Proposal for a...

Recent progress in self-supervision shows that pre-training large neural networks on vast amounts of unsupervised data can lead to impressive increases in generalisation for downstream tasks. Such...

Original URL: https://arxiv.org/abs/2112.00570

👍 Sara Beery, Björn Lütjens

David (dwddao@gmail.com)

2022-02-04 21:53:56

Some of the datasets included in the climate change benchmark are also listed in @Björn Lütjens’ repository but I believe the benchmark plans to be more than that (a full fledged environment to train and validate self-supervised algorithms). Alex mentioned it would be super cool to integrate your data @Ben Weinstein !! 📡

👍 Björn Lütjens, Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2022-02-04 21:56:33

yup. We are actively talking about it now.

🎉 Björn Lütjens

Sara Beery (sara.beery@gmail.com)

2022-02-05 11:44:02

@Chris Yeh Some of the SustainBench dataset (s) might also be relevant?

Björn Lütjens (bjoern.luetjens@gmail.com)

2022-02-05 13:13:56

Amazing thanks!!! I added pointers to lila.science, iWildCam, iNaturalist, and SustainBench. LILA has two cool datasets on forest health and canopy height. I couldn't find directly relevant datasets on SustainBench, but provided a pointer. Thanks!!

Lmk, in case you hear of any good wildfire or belowground carbon datasets 🙂

❤️ Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Sara Beery (sara.beery@gmail.com)

2022-02-05 13:21:03

There is a wildlife/wildfire intersection dataset on Wildlife Insights, from Australia, looking at recovery after brushfires. But I don't think it's ML ready. https://www.worldwildlife.org/stories/an-eye-on-recovery

World Wildlife Fund

An Eye on Recovery

WWF is helping to support Australia’s first large-scale collaborative camera trap project.

Original URL: https://www.worldwildlife.org/stories/an-eye-on-recovery

Björn Lütjens (bjoern.luetjens@gmail.com)

2022-02-05 13:27:43

very cool! I couldn't find a link to the dataset, they might still be processing it. But I'll keep an eye out for this data, thx!

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2022-02-07 12:14:50

I think @Carly Batist and @Gracie Ermi also have datasets listed in their Conservation Tech directory! https://conservationtech.directory/

conservationtech.directory

Fish sounds: towards the evaluation of marine acoustic...

Visit the post for more.

Original URL: https://conservationtech.directory/

👀 David, Björn Lütjens

Chris Yeh (chrisyeh96@gmail.com)

2022-02-07 18:32:17

@Chris Yeh Some of the SustainBench dataset (s) might also be relevant?

Re: @Sara Beery - none of the SustainBench datasets deal with wildlife, wildfire, trees, or carbon. We decided against wildlife because WILDS / iWildcam + lila + iNat already cover wildlife quite well, and we didn't have any in-house (Stanford SustainLab) expertise on the others. (We searched for trees / carbon datasets, but we didn't find any that seemed ready-to-go for inclusion. Maybe in a future version tho!)

👍 Björn Lütjens

🎉 Björn Lütjens

Björn Lütjens (bjoern.luetjens@gmail.com)

2022-02-14 15:49:44

Thank you so much for the thoughts @Lily Xu - OLD ACCOUNT, DO NOT MESSAGE and @Chris Yeh . I'll add a link to conservationtech.directory

😊 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-02-15 07:07:57

@Björn Lütjens We’ll add this forest dataset repo to the directory as well in our next update!

Björn Lütjens (bjoern.luetjens@gmail.com)

2022-02-17 19:38:54

Amazing, thank you!

Silvia Zuffi (silvia@mi.imati.cnr.it)

2022-02-07 10:16:43

Hi all, we have published recently a report on some work we have been doing on fish sounds, I would be interested in some feedback and suggestions! https://arxiv.org/abs/2201.05013v2

arXiv.org

The marine ecosystem is changing at an alarming rate, exhibiting biodiversity loss and the migration of tropical species to temperate basins. Monitoring the underwater environments and their...

Original URL: https://arxiv.org/abs/2201.05013v2

😍 Sara Beery, Justin Kay, Kewal Shah, Jason Holmberg (Wild Me), Jason Parham, Dhruv Sheth, Silvia Zuffi, Emilio Luz-Ricca, Björn Lütjens

👍 Ritwik

🐟 Björn Lütjens

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-02-09 03:39:55

Hi all! @Gracie Ermi & I are working on a new update to the Conservation Tech Directory -- we’ll now be including ‘r-package’ and ‘python-package’ tags and adding conservation tech-related packages as resources (for analyzing data from camera traps, passive acoustics, tracking/telemetry, eDNA, etc.). I’m an R person myself so I know a bunch of R packages but have only got a few Python ones I’ve come across.
Wanted to access the hive mind and see if anyone knew of others to include in our next update that aren’t already on this list? Thanks in advance!

😍 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Declan, Justin Kay, Sara Beery, Talia Speaker, Jason Holmberg (Wild Me), Catherine Villeneuve, Ando Shah, Hannah Yin

👍 Benjamin Kellenberger, Björn Lütjens

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2022-02-09 09:02:11

y'all are doing such a valuable thing thank youuuuu

🤩 Carly Batist

💯 Talia Speaker

👍 Benjamin Kellenberger

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2022-02-09 15:21:59

That’s an amazing list on your Web page indeed! I second @Lily Xu - OLD ACCOUNT, DO NOT MESSAGE’s comment; thanks a lot for your work!

Regarding contents in the screenshot: I just noticed TensorFlow, which begs the question on how generic you allow entries to be. In that case you could open the box of Pandora with Python and add PyTorch, SciPy, Scikit-Image, sklearn, etc.

Otherwise I’ll just speak for a colleague and mention a must-have: DeepLabCut: https://github.com/DeepLabCut/DeepLabCut

DeepLabCut/DeepLabCut

Official implementation of DeepLabCut: Markerless pose estimation of user-defined features with deep learning for all animals incl. humans

Website

<http://deeplabcut.org>

Stars

2731

❤️ Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Carly Batist, Sara Beery

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-02-09 23:44:11

We actually just took TensorFlow (& some others like seewave) out because of being too ‘general’, we were debating where to draw the line and it gets hard! But we agree, thanks for your input!

👍 Benjamin Kellenberger, Sara Beery

Alayna Van Dervort (av@thebigwild.com)

2022-02-09 14:05:22

HI all, Is anyone working on Mountain Lion Individual Identification? We @thebigwild would be very keen to speak! av@thebigwild.com

👍 Ed Miller, Aarnav Sawant

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2022-02-09 15:23:49

Mountain Lion doesn’t seem to be on the list, but the go-to experts in animal re-ID for me are the folks at WildMe: https://www.wildme.org/#/
@Jason Parham @Jason Holmberg (Wild Me)

Jason Parham (bluemellophone@gmail.com)

2022-02-09 15:26:18

We currently don’t have a Mountain Lion re-ID project, but we in discussion about starting a North America carnivore platform

:bearid: Ed Miller, Sara Beery

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2022-02-09 15:27:14

We have never seen the baseline dataset of known IDs that would be the starting point of a future effort

Jason Parham (bluemellophone@gmail.com)

2022-02-09 15:27:43

Agreed, @Alayna Van Dervort do you have a current Mountain Lion photo ID dataset?

Kevin Webb (ktwebb86@gmail.com)

2022-02-09 16:56:21

I love this idea. Are you in touch with Beth Pratt (NWF, P-22 advocate) or Liz Hadly at Stanford?

Jason Parham (bluemellophone@gmail.com)

2022-02-09 16:58:42

@Jason Holmberg (Wild Me)

Alayna Van Dervort (av@thebigwild.com)

2022-02-09 18:06:04

Hi @Jason Holmberg (Wild Me) I will email you directly !

Jason Parham (bluemellophone@gmail.com)

2022-02-09 15:01:43

@Carly Batist Can you add a docker-image tag? We have Python packages at Wild Me, but we also have fully-configured and public Docker images with all of our code, plugins, required dependencies, and CUDA pre-installed.

😎 Jon Van Oast, Sara Beery, Carl Boettiger

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-02-10 02:44:56

Hi Jason! We can absolutely add that term to your description so that if someone searches ‘docker’ WildMe will still show up, but we try to keep the tags to more generalized topic areas so that we don’t end up with dozens of them. Hope this is an ok solution!

Jason Parham (bluemellophone@gmail.com)

2022-02-10 13:40:11

that works!

👍 Carly Batist

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-02-10 02:40:45

https://www.prnewswire.com/news-releases/terrapulse-launches-worlds-first-10-meter-resolution-global-forest-monitoring-platform-301477047.html

prnewswire.com

TerraPulse Launches World's First 10-Meter Resolution Global Forest Monitoring Platform

/PRNewswire/ -- TerraPulse, Inc., the green technology startup that introduced the world's first high-resolution satellite map of global tree canopy cover in...

Original URL: https://www.prnewswire.com/news-releases/terrapulse-launches-worlds-first-10-meter-resolution-global-forest-monitoring-platform-301477047.html

🌳 Sara Beery, Monty Ammar, Kevin Webb

Akronix (akronix5@gmail.com)

2022-02-11 02:47:55

https://kili-technology.com/blog/kili-s-community-challenge-plastic-in-river-dataset?u[…]470itulP62G32JVyIDxxeObqvLelL1EN9mXnsvRkf3Rl17wYcUfNxHujeH4

kili-website

Kili's Community Challenge - Plastic in River dataset

Plastic pollution in oceans threatens the whole ecosystem! Rivers are the very first source of plastic in Oceans and AI can play a huge role in the struggle against it. Help us to create a major benchmark dataset for training and testing river plastic detection models!

Original URL: https://kili-technology.com/blog/kili-s-community-challenge-plastic-in-river-dataset?utm_campaign=Sendingblue%20email&utm_source=sendinblue&utm_medium=email&_hsenc=p2ANqtz-9QOpTyhp6MI470itulP62G32JVyIDxxeObqvLelL1EN9mXnsvRkf3Rl17wYcUfNxHujeH4

❤️ Kewal Shah

🙌 Emmanuel Dufourq, Björn Lütjens

Devis Tuia (devis.tuia@epfl.ch)

2022-02-16 09:02:09

Hey folks, I have a couple of PhD positions open in my lab at EPFL, if interested (or if you know interesting candidates) send me a message!

👍 Oisin Mac Aodha, Nico Lang, Omiros Pantazis, Yihang She, Sara Beery, Ben Koger, Monty Ammar, Björn Lütjens, Alexandre Tytgat

🏔️ Omiros Pantazis, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Sara Beery, Elijah Cole (Deactivated), Björn Lütjens

Devis Tuia (devis.tuia@epfl.ch)

2022-02-16 09:03:12

https://www.epfl.ch/about/working/phd-position-mapping-wildlife-environment-interactions-in-the-swiss-alps-with-ai-wildai/ this is one

EPFL

PhD position Mapping wildlife-environment interactions in the Swiss Alps with AI (WildAI)

Your mission As a PhD student, you will investigate the interactions between wild animals and their environment in the Swiss National Park. You will collect videos of animals in the wild, analyze their behavior with advanced computer vision techniques and relate them with fine grained environmental reconstructions obtained from drone data. Technical challenges related to (...)

Original URL: https://www.epfl.ch/about/working/phd-position-mapping-wildlife-environment-interactions-in-the-swiss-alps-with-ai-wildai/

Devis Tuia (devis.tuia@epfl.ch)

2022-02-17 05:06:47

https://www.epfl.ch/about/working/phd-position-target-shift-in-species-distribution-modeling/ and this is the other!

EPFL

PhD position – Target shift in Species distribution modeling

Your mission As a PhD student, you will investigate the potential of deep learning models for describing the suitability of habitats to host animal and vegetal species. In particular, you will develop models based on remote sensing data to model habitats where species can be found and prosper. To do so, you will leverage large-scale (...)

Original URL: https://www.epfl.ch/about/working/phd-position-target-shift-in-species-distribution-modeling/

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2022-02-18 11:18:55

Hi all, thanks very much to @Björn Lütjens for inviting me to this group. I've just joined ETH Zürich/Restor as a postdoc where I'm working on applying computer vision and ML to monitor forest restoration sites. I just got back from Antarctica as a Winterover for the IceCube Neutrino Observatory, but before that I was in Liverpool at LJMU working on real-time animal detection in thermal imagery. Otherwise I've also been working on and off with Frontier Development Lab putting ML into space for disaster response. My background is a hodgepodge of computer vision, remote sensing, space science, edge hardware and ecology so I'm always up for collaboration and conversation that mixes those domains. Nice to meet you all!

👋 Oisin Mac Aodha, Benjamin Kellenberger, Ben Weinstein, Stephanie O'Donnell, Kangyu Zheng, Akronix, Yihang She, Elijah Cole (Deactivated), Ankita Shukla, Omiros Pantazis, Ando Shah, Catherine Villeneuve, Dhruv Sheth, Monty Ammar, Emilio Luz-Ricca, David, Dan Morris, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Sara Beery, Sicily Fiennes, Alexandre Tytgat

🐝 Björn Lütjens, Dhruv Sheth, David, Lee Wall

Lee Wall (lmw17@my.fsu.edu)

2022-02-22 19:10:52

Hey everyone! I just graduated with my Bachelor's in applied math in December, and I've been growing more and more interested in both AI and the environmental field. Happened upon this group while scouring the Internet for projects, groups, etc. making organized efforts to connect people across fields to solve the challenges facing our planet-- and it seems that's exactly what this is! Really glad to have found y'all. Feel free to reach out if you wanna chat or anything!

👍 Omiros Pantazis, Mitch Fennell, Catherine Villeneuve, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Akronix, Benno Simmons, Lucia Gordon, Devis Tuia, Alexandre Tytgat, Lee Wall

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2022-02-23 02:40:44

Welcome! 😄

Sicily Fiennes (sicilyfiennes@gmail.com)

2022-02-23 15:33:01

Hi everyone! I am a PhD student and AI for Earth grantee working on using machine learning to detect species of birds in the wildlife trade. Any other PhD students using machine learning or running code on Microsoft Azure here? We’ve started a separate Slack channel, Ie. for researchers at a similar coding level and career stage. Please PM me or comment if you’d like to join!

🌳 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Jason Holmberg (Wild Me), Catherine Villeneuve

👋 Jon Van Oast, Benjamin Kellenberger, Devis Tuia, Alexandre Tytgat, Zac Winzurk

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2022-02-23 18:21:53

I've got a project on Azure right now and would love to be added to this Slack channel! Thanks Sicily 🙂

👋 Sicily Fiennes

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2022-02-24 02:24:48

I have had my fair share with Azure too, also through my software, and would like to contribute to, and learn from, this channel, too! Aside that, we’re also working on (aerial) bird detection.

👍 Sicily Fiennes, Sara Beery

Sicily Fiennes (sicilyfiennes@gmail.com)

2022-02-24 07:17:16

Hi both! Could you please share your emails via PM

Caleb Robinson (calebrob6@gmail.com)

2022-02-25 13:51:39

Hi Sicily 👋 could you add me as well? I'm not a PhD student but do use Azure!

👍 Benjamin Kellenberger, Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Alayna Van Dervort (av@thebigwild.com)

2022-05-31 18:11:53

Hello, Yes please I would love to be added, av@thebigwild.com

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2022-02-28 18:28:43

Hi everyone! We are excited to host Prof. @Carl Boettiger at the Harvard CRCS Social Impact Seminar series next week.

Carl Boettiger (UC Berkeley) will be talking about the intersection of AI and conservation for ecological monitoring, political ecology, and so on. See his full abstract/bio.

Monday, March 7, 11am ET
Register for the talk

Carl is also happy to chat with folks afterwards! I've set up this sign up for a chat slot with Carl; please feel free to sign up, and invite others in your group/department. This can be 1-on-1, several members in your group, etc.

😍 Sara Beery, Akronix, Lee Wall, Ando Shah, Yuerou Tang

🙌 Marcus Lapeyrolerie, Akronix, Millie Chapman, Jessica Couture

👍 Benjamin Kellenberger, Casey Youngflesh

Oisin Mac Aodha (macaodha@caltech.edu)

2022-03-01 10:19:01

Hi everyone!

I have an open postdoc position on Machine Learning for Large Scale Biodiversity Mapping at the University of Edinburgh that might be of interest to some people here. I'll add more details in this thread. Don't hesitate to reach out if you have any questions.

I've also got a tweet here in case you would be willing to spread the word:
https://twitter.com/oisinmacaodha/status/1498678226310287371

🙌 Sara Beery, Elijah Cole (Deactivated), Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Ben Weinstein, Pietro Perona, Ando Shah, Omiros Pantazis, Shane Lubold, Subhransu Maji

Oisin Mac Aodha (macaodha@caltech.edu)

2022-03-01 10:19:27

Summary:
Postdoc in Machine Learning for Large Scale Biodiversity Mapping at the University of Edinburgh
Duration: One year, starting from April 2022 (can be negotiated)
More information and how to apply: https://edin.ac/3hutyUG
Additional queries: Contact Oisin Mac Aodha - https://homepages.inf.ed.ac.uk/omacaod/

Details:
The School of Informatics, University of Edinburgh invites applications for a research associate in the area of machine learning for large scale biodiversity mapping. The project is part of an exciting collaboration between the School of Informatics and the citizen science platform iNaturalist, and features input from the International Union for Conservation of Nature (IUCN).

Species range maps, describing where species do and do not occur, are critical to prioritizing scarce conservation resources in order to mitigate some of the worst potential impacts of climate change on biodiversity. Citizen science platforms such as iNaturalist generate millions of casual observations each year which contain information about where thousands of different species have been observed. However, constructing accurate species range maps is a time consuming and laborious process that is further hampered by data quality and coverage issues.

This project aims to develop novel machine learning-powered methods for producing plausible species range maps from citizen science data. The developed methods will enable the joint modelling of thousands of different species' distributions. These methods will make use of, and advance, the state of the art in deep learning-based density estimation in the context of noisy geospatial data.

For additional information please check our previous related project: https://edin.ac/3vvFyh2.

Dan Morris (agentmorris@gmail.com)

2022-03-03 18:57:16

Around a zillion months ago, on a thread on this channel, @Carl Boettiger suggested that we add some structured metadata to LILA to improve standardization and discoverability, and I said (paraphrasing) "great idea, I'm on it!", and then I did nothing about it for six months, but I finally got a chance to add schema.org markup to LILA pages, such that they are now searchable through, e.g., Google Dataset Search:

https://datasetsearch.research.google.com/search?src=0&query=lila.science&docid=L2cvMTFwenlwODAwYg%3D%3D

If anyone sees anything amiss with the metadata, let me know!

🙌 Caleb Robinson, Talia Speaker, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Mitch Fennell, Carl Boettiger, Jason Holmberg (Wild Me), Sara Beery

🎉 Jon Van Oast, Subhransu Maji, Carl Boettiger, Riccardo de Lutio, Jason Holmberg (Wild Me), Sara Beery

👍 Jon Van Oast, Eelke, Benjamin Kellenberger, Oisin Mac Aodha, Jessica Couture

Elijah Cole (Deactivated) (ecole@caltech.edu)

2022-03-03 22:59:28

https://www.nytimes.com/interactive/2022/03/03/climate/biodiversity-map.html

The New York Times

By Catrin Einhorn and Nadja Popovich

This Map Shows Where Biodiversity Is Most at Risk in America

New maps provide a valuable tool to officials and scientists scrambling to protect threatened species.

Original URL: https://www.nytimes.com/interactive/2022/03/03/climate/biodiversity-map.html

👍 Benjamin Kellenberger, Riccardo de Lutio, Oisin Mac Aodha, Omiros Pantazis, Catherine Villeneuve, Kewal Shah, Casey Youngflesh, Suzanne Stathatos, Jessica Couture, Ted Schmitt, Sara Beery, Mark Roth

Monty Ammar (montyx23@gmail.com)

2022-03-06 05:44:40

Hi all, long shot, but I was wondering if anyone has come across any databases containing labelled gold-mines/ artisanal gold mines from Landsat 7 or 8/9/ Sentinel-2 satellite images?

Thanks 😊

Sara Beery (sara.beery@gmail.com)

2022-03-06 05:59:15

@Laure Delisle what type of mines were you looking at?

👍 Monty Ammar

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-03-06 06:52:24

You might ask the folks from the ConX artisinal mining challenge? I think some of the solutions were working on remote sensing of mines. https://www.artisanalminingchallenge.com/previous-round

👍 Sara Beery, Monty Ammar

Monty Ammar (montyx23@gmail.com)

2022-03-06 16:54:29

@Carly Batist Thank you, i’m already in contact with them 🙂. While waiting to hear more information, I thought to come here and see if there are any platforms containing them that I’m missing out on.

👍 Carly Batist

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-03-07 03:25:20

https://techlaw.uottawa.ca/aisociety/alextrebek-postdoc-environment-2022

CDTS / CLTS uOttawa

Alex Trebek Postdoctoral Fellowship, AI and Environment

Type: Postdoctoral FellowshipStart date: Flexible (tentatively September 1, 2022)Duration: One-year fellowshipSalary range: $45,000, plus benefitsReports to: Dr. Kelly Bronson, Canada Research Chair in Science and Society (in collaboration with Dr. Colleen M. Flood and Dr. Florian Martin-Bariteau)

Original URL: https://techlaw.uottawa.ca/aisociety/alextrebek-postdoc-environment-2022

👍 Sara Beery, Justin Kay, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Zac Winzurk

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-03-07 03:27:12

Another postdoc opportunity -

Kewal Shah (kewalshah077@gmail.com)

2022-03-08 10:05:16

Call for contributions is now open for the ACI Conference!

https://twitter.com/animal_computer/status/1501142891347030018?s=20&t=buS9fhpefOOMBr2WyuqK2g

Devis Tuia (devis.tuia@epfl.ch)

2022-03-08 10:15:36

Gentle reminder, one week left to apply for PhD positions in AI for conservation in beautiful Switzerland! https://www.epfl.ch/labs/eceo/eceo/open-positions/

EPFL

Open positions

ECEO

Original URL: https://www.epfl.ch/labs/eceo/eceo/open-positions/

🙌 Stephanie O'Donnell, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Sara Beery, Dhruv Sheth

Dan Situnayake (daniel@edgeimpulse.com)

2022-03-09 13:00:51

Hi everyone! I work at Edge Impulse; we create tools to help developers build machine learning models for deployment to embedded devices. We're not a conservation tech company, but we do a ton of conservation-related work:

We fund and mentor the WILDLABS Fellowship: On the edge

We organize community conservation tech efforts, like ElephantEdge

We work very closely with some amazing conservation tech organizations, like Arribada and CXL

We are regular speakers at community events like WILDLABS tech tutors

We donate 1% of our revenue to environmental nonprofits

I'm posting here because we're looking for a machine learning engineer to join our team. You'd spend your time creating new technologies around machine learning on edge devices—including those that have the potential to directly benefit conservation research. For example, one big project we're working on right now is training embedding models for camera traps. If this type of thing sounds appealing, drop me an email at dan@edgeimpulse.com. Thank you!!

jobs.lever.co

Edge Impulse - Machine Learning Engineer

We are growing. This is an opportunity for a talented generalist machine learning engineer to build the best MLOps tools in the world—specifically for embedded ML. As part of our industry-leading team, you will design and build tools that enable the world’s embedded engineers to capture their domain knowledge in effective machine learning <a href="http://applications.Do">applications.Do</a> you love the idea of combining open-ended independent research with direct applied ML within a supportive and experienced team of professional engineers? Come join us!

Location

Amsterdam (Remote)

Team

Original URL: https://jobs.lever.co/edge-impulse/f1fadc92-ed4b-4619-b763-4e8a3bfc5687

❤️ Stephanie O'Donnell, Talia Speaker, Dhruv Sheth, Benjamin Akera, Yihang She

👍 gvanhorn, Oisin Mac Aodha, Justin Kay, Sam Kelly, Daniel Davila, Benjamin Kellenberger, Dhruv Sheth, Carly Batist, Cameron Trotter

😍 Sara Beery, Dhruv Sheth

Angjoo Kanazawa (kanazawa@berkeley.edu)

2022-03-09 23:29:36

Hi all! We’re organizing the 2nd CV4Animals workshop at CVPR 2022!
We’re inviting abstracts to present at the workshop (either by poster or oral), deadline April 29th, please see the website and attached PDF for more details!
https://www.cv4animals.com/home

cv4animals.com

About

Original URL: https://www.cv4animals.com/home

👍 gvanhorn, Jason Parham, Kakani Katija, Benjamin Kellenberger, Devis Tuia, Oisin Mac Aodha, Daniel Davila, Dhruv Sheth, Jason Holmberg (Wild Me)

:zebra_face: Jason Parham, Elijah Cole (Deactivated), Dhruv Sheth, Jason Holmberg (Wild Me)

🙌 Stephanie O'Donnell

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-03-10 03:27:44

💥Conservation Tech Directory update!💥

New feature - we’ve included conservation tech stats packages (python & R/rstudio tags). And now at 618 resources!🤩. We’ve also created a new resource type (Tool) & re-classified resources to this. Includes datasets, toolkits, software packages, etc. https://conservationtech.directory/
PDF version: https://doi.org/10.6084/m9.figshare.15442200. @Gracie Ermi

conservationtech.directory

https://tejaswid.github.io/publication/2019-08-12-Automatic-Segmentation-Trees

Visit the post for more.

Original URL: https://conservationtech.directory/

😍 Dhruv Sheth, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Monty Ammar, Jason Holmberg (Wild Me), Sara Beery

🙌 Dhruv Sheth, Declan, Jason Holmberg (Wild Me), Emilio Luz-Ricca

👍 Benjamin Kellenberger, Jason Holmberg (Wild Me)

🎉 Jon Van Oast, Jason Holmberg (Wild Me), Yuerou Tang, Hannah Yin

💪 Monty Ammar

Chris Yeh (chrisyeh96@gmail.com)

2022-03-10 20:24:34

If you are interested in applications of machine learning (ML) for sustainability applications, we welcome you to attend the talks and presentations for the Computational Sustainability Doctoral Consortium Conference (CompSust DC 2022), which is held virtually tomorrow (Friday) and Saturday.

The conference program and schedule can be found here:
https://www.compsust.net/compsust-2022/program.php

Zoom link: https://vanderbilt.zoom.us/j/4153638703?pwd=QjZ2UFlwdDRZanB2ODlZUDhEVFBwdz09
Meeting ID: 415 363 8703
Passcode: 2248

❤️ Ayan Mukhopadhyay, Sara Beery, Justin Kay, Dhruv Sheth, Subhransu Maji, Monty Ammar, Yihang She, Oisin Mac Aodha, Stephanie O'Donnell, Akronix, Caleb Robinson, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Akronix (akronix5@gmail.com)

2022-03-11 10:05:07

How can we attend to the talks?

Ayan Mukhopadhyay (ayanmukg@gmail.com)

2022-03-11 10:39:17

@Akronix Just use the zoom link to log in

Peter van Lunteren (peter@addaxdatascience.com)

2022-03-12 09:02:13

Unfortunately I can’t make it on time. Will it be recorded?

Chris Yeh (chrisyeh96@gmail.com)

2022-03-12 10:57:57

Yes, the talks are recorded. We will upload recordings to YouTube

Peter van Lunteren (peter@addaxdatascience.com)

2022-03-13 06:08:35

Could you send me the link once available on YouTube? Thanks!

Chris Yeh (chrisyeh96@gmail.com)

2022-03-12 11:56:43

CompSust networking session on Gather.town for the next 2 hours! (12-1:15p ET + 45-min "lunch break")
https://app.gather.town/events/g92MikIFkdovAgODSdwI

Chris Yeh (chrisyeh96@gmail.com)

2022-03-12 15:44:09

Talk by Victor Anton (founder & CEO of Wildlife.ai) in 15 minutes!
https://vanderbilt.zoom.us/j/4153638703?pwd=QjZ2UFlwdDRZanB2ODlZUDhEVFBwdz09

Riccardo de Lutio (riccardo.delutio@geod.baug.ethz.ch)

2022-03-15 10:23:25

Hi everyone! Would anyone have recommendations for pretrained models for tree detection, or tree crown segmentation for urban trees? I’m aware of the DeepForest prebuilt model by @Ben Weinstein which looks amazing (but not specifically designed for urban trees) or the TreeTect models. We could of course also retrain a model but since we only have the coordinates of the trees and we don’t necessarily want to relabel bounding boxes or masks it’s maybe not the best way to go about it. Any other ideas? Thanks!

Devis Tuia (devis.tuia@epfl.ch)

2022-03-15 10:34:35

For urban tree detection + species classification, there was the registree project led by Caltech and ETH (http://www.vision.caltech.edu/registree/)

vision.caltech.edu

RegisTree - Home

A system that automatically generates inventories of street trees using aerial and street-view imagery. A trees geo-location, species and trunk diameter are included in the inventory.

Original URL: http://www.vision.caltech.edu/registree/

Devis Tuia (devis.tuia@epfl.ch)

2022-03-15 10:35:00

(but you should know it, actually 😉, now that i think about it )

Devis Tuia (devis.tuia@epfl.ch)

2022-03-15 10:40:16

For tree crown segmentation, look at works by Antonio Ferraz or Martin Weinmann. They have been doing a lot in that direction

Riccardo de Lutio (riccardo.delutio@geod.baug.ethz.ch)

2022-03-15 11:12:38

Thanks yes that’s definitely one option too! 🙂

Ben Weinstein (benweinstein2010@gmail.com)

2022-03-15 11:50:21

You should think of DeepForest as a baseline model that you build from, I've heard a number of people add in urban tree detections. Just spend 2 or 3 days annotating and finetune and you should be in good shape.

Ben Weinstein (benweinstein2010@gmail.com)

2022-03-15 11:51:15

i'm happy to host a urban tree model branch if gets up and running.

Riccardo de Lutio (riccardo.delutio@geod.baug.ethz.ch)

2022-03-15 12:01:59

Great thanks!

Dhruv Sheth (dhruvsheth.linkit@gmail.com)

2022-03-16 05:11:10

Sundara Tejaswi Digumarti

An Approach for Semantic Segmentation of Tree-like Vegetation

Postdoctoral Researcher at the Oxford Robotics Institute, UK. Works on Computer Vision and Robotics

Original URL: https://tejaswid.github.io/publication/2019-08-12-Automatic-Segmentation-Trees

👍 Riccardo de Lutio

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2022-03-15 10:30:18

https://coomeslab.org/dr-thomas-swinfield/ - Tom swinfield has been working on this for the last few years, but it's in tropic forests rather than urban landscapes. He /his research group might have some pointers

Forest Ecology and Conservation Group

coomeslab (https://coomeslab.org/author/coomeslab/)

Dr Thomas Swinfield

Research Associate I work on tropical forest restoration based jointly between the University of Cambridge – Wolfson College – and the Royal Society for the Protection of Birds. I use a variety of remote sensing techniques from structure from motion, LiDAR and multi-temporal satellite observations to assess forest condition and how it responds to external pressures. Much of my recent work has involved developing metrics that describe 3D habitat structure from LiDAR data and testing these for a variety of taxa with researchers from the Durrell Institute for Conservation and Ecology (DICE) and the Queen Mary University of London. I am interested in using experimental approaches to assess the efficacy of management interventions to improve forest restoration and currently working with partners within the centre to develop the use of unmanned aerial vehicles and computer vision as a low-cost tool for forest quality assessments.  Read More…

Original URL: https://coomeslab.org/dr-thomas-swinfield/

🙏 Riccardo de Lutio

Riccardo de Lutio (riccardo.delutio@geod.baug.ethz.ch)

2022-03-15 10:31:28

Thanks, I’ll look into that!

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2022-03-15 10:32:23

I also wonder if some of urban tree mapping citsci platforms have worked on something

Alexander Pfyffer (alexander.pfyffer@gmail.com)

2022-03-17 14:23:32

Hi everyone! 👋 I’m a social entrepreneur / software engineer looking for people interested in working in regenerative farming in developing countries. The idea is to connect the millions of smallholder famers to the international CO2 emission compensation market. The goal would be to use smartphone pictures / satellite imagery as a method to assess how much CO2 they are capturing (of course using AI 🙌). I’m planning to go to Ghana for the next few months to test my assumptions. Do you know people / research projects that are working on something similar?

💯 Chittesh T, Margaux Masson-Forsythe

Lauren E. Gillespie (gillespl@stanford.edu)

2022-03-18 18:15:58

Hi Alexander, you should check out David Dao! He’s also doing a lot of work in carbon credit estimation and secure payment for smallholder farms on the carbon markets in the Global South, specifically Central and South America

Margaux Masson-Forsythe (margaux.masson21@gmail.com)

2022-03-21 16:18:38

https://earthshot.eco/

earthshot.eco

Earthshot Labs

We are an open-source community, public benefit technology company, and regenerative asset manager using data science for planetary-scale climate mitigation.

Original URL: https://earthshot.eco/

Ando Shah (ando@berkeley.edu)

2022-03-25 19:39:24

Also check out Regen Network
https://www.regen.network/

regen.network

Regen Network

Buy and sell ecosystem service credits at the open marketplace for climate solutions.

Original URL: https://www.regen.network/

Ștefan Istrate (stefan.istrate@gmail.com)

2022-03-22 05:21:46

Happy to announce iWildCam 2022, our 5th annual camera trap challenge, focused on helping ecologists monitor biodiversity. Join the Kaggle competition and help us count individual animals across sequences of images.

https://www.kaggle.com/c/iwildcam2022-fgvc9

[Competition co-organized with @Sara Beery & @John Beuving, with valuable support from @Dan Morris.]

kaggle.com

iWildCam 2022 - FGVC9

Count the number of animals in a sequence of images

Original URL: https://www.kaggle.com/c/iwildcam2022-fgvc9

🙌 Stephanie O'Donnell, Akronix, Justin Kay, Dan Morris

Vincent Miele CNRS (vincent.miele@univ-lyon1.fr)

2022-03-22 09:12:32

Hi Stefan,The competitions are very exciting. However, as a non participant, I always have this question in mind about the IWILDCAM competition: besides knowing which team wins the competition, how can we actually know how this team proceed (algorithms, software, and so on)? In the open science era, it would be great to know... In any case, thanks so much for organizing this kind of event.

Sara Beery (sara.beery@gmail.com)

2022-03-22 11:50:40

Vincent, in the discussion board for each competition the winning teams post their solutions! You can go to any of the past competitions and read about the winning solutions. We also have recorded videos from the last two workshops of a discussion of the competition-winning results for all the FGVC competitions

Sara Beery (sara.beery@gmail.com)

2022-03-22 11:53:54

We're hoping to write up a larger "what we've learned" paper covering the past competitions and the themes of what does and doesn't work. Kaggle can't enforce teams publishing their code unfortunately (unlike LifeCLEF!) but some competition winners have published their code independently (like last year's winner here: https://github.com/alcunha/iwildcam2021ufam)

alcunha/iwildcam2021ufam

1st Place Solution to iWildcam 2021: Count the number of animals of each species present in a sequence of images

Stars

Language

Python

👍 Ștefan Istrate

Ștefan Istrate (stefan.istrate@gmail.com)

2022-03-22 11:55:58

Competitions like this don't end with a deadline and a winner. Oftentimes participants (are encouraged to) share their solutions afterwards, in forums, papers or workshops. I recommend to everyone interested in learning more to revisit such competitions after a while, and I am sure new materials will become available.

For example, for iWildCam 2021 there was a FGVC workshop afterwards: https://www.youtube.com/watch?v=yKP_tX_gTk0, and participants shared full solutions (as Sara mentioned), or just descriptions of their solutions on the Kaggle forums:

1st place: https://www.kaggle.com/c/iwildcam2021-fgvc8/discussion/245460
2nd place: https://www.kaggle.com/c/iwildcam2021-fgvc8/discussion/245559
3rd place: https://www.kaggle.com/c/iwildcam2021-fgvc8/discussion/244950

YouTube

FGVC Workshop (https://www.youtube.com/channel/UCp-X0QRcgfwBlkCmrF0Noqg)

FGVC8: iWildCam2021

Original URL: https://www.youtube.com/watch?v=yKP_tX_gTk0

Vincent Miele CNRS (vincent.miele@univ-lyon1.fr)

2022-03-23 05:26:32

Thanks for these precise answers. I missed these information, sorry about that. And, again, thanks for organizing these competitions 👍

Beckett Sterner (bsterne1@asu.edu)

2022-03-22 10:49:40

On March 15, 2022, the U.S. National Science Foundation (NSF) held a Town Hall bringing together a multidisciplinary group of researchers to discuss this topic and to identify challenges ripe for this approach. Knowinnovation is a company specializing in accelerating scientific innovation - we are facilitating this program on behalf of the NSF.

The ideas from that Town Hall have been distilled into four topics which will serve as focal points for four separate workshops. Each workshop will consider how all the STEM disciplines (including biology, chemistry, computer sciences, engineering, geosciences, mathematics, physics, social, behavioral, and economic sciences) could be used to tackle a specific problem. All workshops will incorporate cross-cutting themes of diversity, equity, and inclusion and STEM education, training, and workforce development.

Participation in the workshop is by application only. Applications close on Tuesday March 29, 2022.

Apply using this link: https://app.smartsheet.com/b/form/f45ea9e3f86845818d05d272d1f9e604

Additional ‘incubator’ events will provide further engagement for postdocs attending the workshops. Whether or not you are able to participate, we strongly encourage and request you share this information with postdocs in your networks.
Workshop Topics
Workshop 1: Stewarding an Integrated Biodiversity-Climate System (April 14, 2022 11:00 AM - 5:00 PM EDT)
We are learning the Rules of Life that govern the essential role of biodiversity in controlling function, maintenance, and adaptation of every ecosystem on Earth. We are also learning that biodiversity and climate are inextricably linked and that everything affecting one affects the other. How might these lessons help us to predict, preserve, and harness the benefits of biodiversity for human society and the natural world?

Workshop 2: Achieving a Sustainable Future (April 19, 2022 11:00 AM - 5:00 PM EDT)
We are learning the Rules of Life that govern the complexity of interconnected living systems at multiple scales, e.g., from natural and synthetic cells to organisms, populations, communities, ecosystems, and the biosphere. As we learn more about the ways that living systems use and re-use natural resources, how might these lessons help us devise strategies to improve sustainability?

Workshop 3: Harnessing Microbiomes for Societal Benefit (April 21, 2022 11:00 AM - 5:00 PM EDT)
We are learning the Rules of Life that govern the individual and collective metabolism, physiology, signaling, and interaction of different microbiomes, as well as their composition and responses to evolving environments. As we learn more about the roles of microbiomes in all living systems, how might these lessons help us to improve human society and the biosphere?

Workshop 4: Leveraging AI and Data for Predicting Mechanisms (April 26, 2022 11:00 AM - 5:00 PM EDT)
We are learning the Rules of Life that govern the prediction of an organism’s observable characteristics from interactions of its genome with the environment. At the same time, novel research on artificial intelligence and data analytics is providing essential tools for integrating Rules of Life data. How might these lessons help us to improve our ability to use AI and Data Science?

Postdoc 'incubators'
All postdoc participants will also be invited to attend a series of incubators to complement the workshops:

Kick-off Incubator (all selected postdocs): April 12, 1:00 PM to 4:00 PM EDT

Writing Incubator A: April 22, 11:00 AM to 4:00 PM EDT (for attendees of workshops 1 and 2)

Writing Incubator B: May 2, 11:00 AM to 4:00 PM EDT (for attendees of workshops 3 and 4)

Wrap-up Incubator (all selected postdocs): May 17, 11:00 AM to 4:00 PM EDT

😍 Sara Beery

Kevin Webb (ktwebb86@gmail.com)

2022-03-24 14:18:28

If anyone is doing any work around fisheries in the US or Canada, take a look at this forgivable loan program from Multiplier. Deadline April 15. https://multiplier.org/2022/03/announcing-the-technology-adoption-fund-for-sustainable-fisheries-and-the-inaugural-round-of-loan-funding/

Multiplier

Announcing the Technology Adoption Fund for Sustainable Fisheries and the Inaugural Round of Loan Funding

Announcing the Technology Adoption Fund for Sustainable Fisheries and the Inaugural Round of Loan FundingMultiplier, in collaboration with The Walton Family Foundation and Builders Initiative, is pleased to announce the Technology Adoption Fund for Sustainable Fisheries. This is a new pooled loa

Original URL: https://multiplier.org/2022/03/announcing-the-technology-adoption-fund-for-sustainable-fisheries-and-the-inaugural-round-of-loan-funding/

🐟 Justin Kay

Fernando Pérez (fernando.perez@berkeley.edu)

2022-03-25 16:20:04

👋 everyone - I’m Fernando Pérez from UC Berkeley, and wanted to introduce myself as well as let you know of a new program we are just launching that I hope will be a great point of connection with this community: the Eric and Wendy Schmidt Center for Data Science and Environment at Berkeley, or ‘DS4E’ for short (our website is here).

The short version is that:

Our mission is impact-oriented, not academic research. We want to combine the expertise of environmental scientists, engagement with stakeholders close to the issue, and expertise in software engineering and data science, towards concrete, implementable, deployed tools that can contribute to real-world problems.

Our specific topics to focus our effort into are not yet pre-defined, and instead we will organize later this year (once we have some staff hired!) events to consult with the community how to best allocate our resources and work together for maximal impact.

The above is only a launch press release, as we are currently in early stages of remodelling space, getting job ads posted, etc. But we wanted to connect with you all in the hope that you will be interested in opportunities to collaborate with us, apply for positions, participate in future events, etc.

I wanted to particularly thank @Sara Beery for kindly welcoming us to this space - we hope to meet many more of you over time!

From our launch team, both @Carl Boettiger and I are on this Slack and will be happy to answer any questions you might have. Thanks for reading!

Berkeley News

New UC Berkeley center will apply data science to solving environmental challenges

Open science and collaboration with communities is at the heart of the Eric and Wendy Schmidt Center for Data Science and Environment

Original URL: https://news.berkeley.edu/2022/03/23/new-uc-berkeley-center-will-apply-data-science-to-solving-environmental-challenges/

🔥 Daniel Davila, Sara Beery, Casey Youngflesh, Declan, Jason Holmberg (Wild Me), Mitch Fennell, Dhruv Sheth, Akronix, Yihang She, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Jinsu Elhance

👋 Daniel Davila, Kewal Shah, Sara Beery, Stephanie O'Donnell, Oisin Mac Aodha, Jason Holmberg (Wild Me), Emilio Luz-Ricca, Dhruv Sheth, Carly Batist, Elijah Cole (Deactivated), Devis Tuia, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Alex Brace, Jinsu Elhance, Angjoo Kanazawa, Ben Best

😍 Dhruv Sheth, Carly Batist, Lee Wall, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Jinsu Elhance

:thumbsup_all: Frederic Fol Leymarie, Ed Miller, Jinsu Elhance

🦜 Ben Best

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-03-26 06:48:31

So cool!! I’ve added it in to our next update of the Conservation Tech Directory!

Akronix (akronix5@gmail.com)

2022-03-26 16:05:18

Very exciting and interesting project! Looking forward to see the vacancies for new members of the team!

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2022-03-27 15:09:09

Wonderful news and congrats on the launch! An inspiring mission for all of us passionate about impact-oriented work

Danielle Montocchio (montocds@mcmaster.ca)

2022-03-28 18:06:30

Hi everyone ! I am a PhD candidate at McMaster University, Ontario, Canada. As a part of my thesis, I study fish communities in Great Lakes coastal wetlands in Georgian Bay, Ontario. I have been attempting to refine and apply a remote underwater camera system to survey fish abundance, and species richness. I now have hours of footage to go through manually, and was hoping to find an automated method that would be less time-consuming. Ideally, the software could flag when there is a fish-occurrence that could be later processed by researchers.

The major issue I've encountered however, is the requirement of having significant coding language experience to use anything that has been done in the literature, rather than a user-friendly software with a front-end interface. I am not sure if anyone has a way of processing 15-minute action camera video clips in .mov file format (1050p 60 fps), but I thought I would reach out, just in case. Currently, I have 4 and half hours of footage processed manually, where 80 to 95% of the footage can be empty (just water and/or plants). Usually if a 15-minute clip is completely empty, I delete it, otherwise the video is unedited. I have annotations of fish occurrences, species identification and time-stamps for the processed videos saved in an Excel spreadsheet currently. In total, I have 540 hours of footage, with similar environments, but always slightly different backgrounds of vegetation. I have attached some sample images of occurrences if that helps at all. Again, I am very coding inexperienced so any direction/advice would be greatly appreciated!

👋 Sara Beery, Arjun Subramonian (they/them), Daniel Davila, Subhransu Maji

🎉 Jon Van Oast

Sara Beery (sara.beery@gmail.com)

2022-03-28 18:07:40

@Kakani Katija sounds similar to some of the FathomNet data?

Daniel Davila (daniel.davila@kitware.com)

2022-03-28 18:27:55

We maintain an open source DIY-AI tool for NOAA, called VIAME. Originally targeting fisheries but we've since expanded to a number of subsea applications. It supports a number of workflows such as detection, tracking, counting, etc... while also providing a means of labeling the data. The idea is for a scientists or decision maker to be able to train models without knowing anything about ml, or even code. Just bring the SME perspective and some time to label haha. It's been deployed at many of noaa's sites already and some of our collaborators at various uni and commercial companies have found it useful. It's completely free and open source, we dont make products or hold any IP at kitware. Please feel free to reach out if you'd like me to connect you to our program manager or engineering team.

😍 Sara Beery

🙌 Tarun Kumar Verma, Ando Shah

🎉 Jon Van Oast

Daniel Davila (daniel.davila@kitware.com)

2022-03-28 18:30:18

https://github.com/Kitware/dive

Kitware/dive

Media annotation and analysis tools for web and desktop. Get started at <https://viame.kitware.com>

Website

<https://kitware.github.io/dive>

Stars

Kakani Katija (kakani@mbari.org)

2022-03-28 19:12:25

Yes, there's also the Tator annotation tool by CVision AI that has AI tooling as well: tator.io. Web-based and easy to get started.

Ben Weinstein (benweinstein2010@gmail.com)

2022-03-28 22:32:57

I made a tiny GUI a few years ago 'motion detection only' for finding periods of activity. Nothing like what @Daniel Davila is doing in terms of sophistication. http://benweinstein.weebly.com/deepmeerkat.html

Dr. Ben Weinstein

DeepMeerkat

Machine learning for video-based biodiversity surveys.

Original URL: http://benweinstein.weebly.com/deepmeerkat.html

Ando Shah (ando@berkeley.edu)

2022-03-29 00:41:25

In case you want a full blown underwater camera trap (definitely overkill if you have access to continuous power), I worked on one for manta ray research in 2013 that was more for deeper reefs (~30m) :
https://ando.xyz/work/manta-id

Ando Shah

MantaID : LIDAR based underwater camera trap — Ando Shah

The world's first underwater camera trap, designed to automatically capture the lives of manta rays and sharks

Original URL: https://ando.xyz/work/manta-id

🤩 Sara Beery

Malte Pedersen (mape@create.aau.dk)

2022-03-29 01:39:09

Hi Danielle, I am not sure if it is applicable for your study, but we published a bounding box annotated underwater dataset a few years ago in murky waters. Maybe it can be used as training data for your system 🙂
https://www.kaggle.com/datasets/aalborguniversity/brackish-dataset

kaggle.com

The Brackish Dataset

Bounding box annotated underwater image dataset

Original URL: https://www.kaggle.com/datasets/aalborguniversity/brackish-dataset

Danielle Montocchio (montocds@mcmaster.ca)

2022-03-29 14:57:21

Wow thank you so much everyone for your input and help! I have some more options to explore and see what works 🙂 Didn't think what I was doing was too novel and glad to see that it isn't

Kakani Katija (kakani@mbari.org)

2022-03-31 10:38:58

@Danielle Montocchio If you're interested, we're having a workshop Thursday/Friday morning (Pacific) (starts in ~20 minutes) that walks people through resources like Tator and FathomNet. Registration link is here: www.tinyurl.com/fathomnet

Danielle Montocchio (montocds@mcmaster.ca)

2022-03-28 18:09:52

Sorry! Here are the images

Sara Beery (sara.beery@gmail.com)

2022-04-07 12:26:34

Some exciting improvements for iWildCam 2022!!!

https://twitter.com/sarameghanbeery/status/1512102482641444868

twitter

Sara Beery (https://twitter.com/sarameghanbeery/status/1512102482641444868)

This year for #iWildCam 2022 we're maintaining our goal of counting individual animals across sequences, but we've made a few changes from 2021 to help focus on the counting task and improve the interpretability of the leaderboard! <https://www.kaggle.com/competitions/iwildcam2022-fgvc9>

Original URL: https://twitter.com/sarameghanbeery/status/1512102482641444868

🎉 Jason Holmberg (Wild Me), Subhransu Maji, Jason Parham, Carly Batist, Mitch Fennell, Dan Morris, Jan Kees

❤️ Jon Van Oast, Justin Kay

💕 Jon Van Oast

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2022-04-07 12:52:20

I hadn't realized this was the focus. This is awesome!

😁 Sara Beery

💯 Jon Van Oast

Oisin Mac Aodha (macaodha@caltech.edu)

2022-04-07 13:09:06

Cool!

Dan Morris (agentmorris@gmail.com)

2022-04-07 18:17:07

People ask us relatively often: "can I just count MD boxes above some threshold and use that to estimate the # of animals in an image?" We basically always say "no we don't recommend that", but I actually have no idea what the error distribution looks like, we just don't want to assume that it does anything remotely useful for counting. I know this competition is focused on sequence-level counting, but surely some enterprising competitor will make a baseline entry that just takes the per-image MD results that are on the competition page and the GT counts for the training data, and see what the numbers look like? And maybe that enterprising competitor will post a summary here? :)

😁 Sara Beery, Mitch Fennell

Sara Beery (sara.beery@gmail.com)

2022-04-07 18:18:09

We might even put that on the leaderboard as a baseline (we did last year, but it had the complexity of species ID tied in) 🙂

🎉 Jon Van Oast

Dan Morris (agentmorris@gmail.com)

2022-04-07 18:19:02

Yes, do that! I recommend the following advanced algorithm: { count = max(# of boxes) }

👍 Sara Beery, Jason Holmberg (Wild Me), Justin Kay

❤️ Jon Van Oast

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-04-11 07:58:24

🚨Conservation Tech Directory update!🚨

Lots more entries, but biggest update is that we now have a listserv you can subscribe to if you want direct notifications in your inbox of future directory updates! So we don’t have to keep spam’ing you on social media😅
Through the site: https://conservationtech.directory/
Directly via sign-up form: http://eepurl.com/hWwV0X

@Gracie Ermi

conservationtech.directory

The Latest Computer Vision Model Updates

Visit the post for more.

Original URL: https://conservationtech.directory/

❤️ Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Catherine Villeneuve, Gracie Ermi, Sara Beery, Kakani Katija, Emily Charry Tissier, Jason Holmberg (Wild Me), Kai Waddington, Jinsu Elhance

🎉 Jon Van Oast, Ritwik, Akronix, Emily Charry Tissier, Jason Holmberg (Wild Me), Jinsu Elhance

Yves Bas (yves.bas@gmail.com)

2022-04-13 02:30:25

Now 55000 taxa in Inaturalist computer vision model: https://www.inaturalist.org/blog/63931-the-latest-computer-vision-model-updates

iNaturalist

We’ve released a new computer vision model for iNaturalist. This is our first model update since July 2021. The iNaturalist website, mobile apps, and API are all now using this new model. Here’s what’s new and different with this change: * It includes 55,000 taxa (up from 38,000) * Hybrid taxa are excluded * It’s more likely to suggest the correct taxon To see if a particular species is included in this model, you can look at the “About” section of the taxon page. Here’s more details about what changed and why: It’s a lot bigger Our previous model included 38,000 taxa and 21 million training photos. When we announced that we were kicking off a new vision training job in October, we planned to train a new model on 47,000 taxa and 25 million training images. However, in November, after training had started, we realized that we could make improvements to how we were picking our training data that would improve the quality of the vision models. After making those changes, we re-started...

Original URL: https://www.inaturalist.org/blog/63931-the-latest-computer-vision-model-updates

🤩 Sara Beery, Nico Lang, Jason Holmberg (Wild Me), Emilio Luz-Ricca, Daniel Grzenda, Howard L Frederick

🤯 Oisin Mac Aodha, Justin Kay, Jason Holmberg (Wild Me), Stefan Schneider, Yuerou Tang

🎉 Sara Beery, Kakani Katija, Elijah Cole (Deactivated), Alex Brace, Dan Morris, Jason Holmberg (Wild Me), Jon Van Oast, Ben Best

👍 Jan Kees, Carly Batist

gvanhorn (grv22@cornell.edu)

2022-04-18 16:55:17

Sound ID 2.0 is now out for iOS devices! Android will be available next month.
While the UI looks the same for 2.0, everything "under the hood" has been rewritten and optimized:

efficiency is 10 to 100x better, depending on iPhone version

You should be able to start Sound ID and leave it running while on a walk, hike, bike ride, etc. Get out and explore!

precision and recall are up across the board for all species

we now cover 560 US/Canada species, and 250 Western Paleartic species

Lots of additional species on the horizon!

Hope you all enjoy!
https://merlin.allaboutbirds.org/

Merlin Bird ID - Free, instant bird identification help and guide for thousands of birds

Merlin Bird ID - Home

Bird ID Wizard—Step-by-step Answer three simple questions about a bird you are trying to identify and Merlin will give you a list of possible matches. Merlin offers quick identification help for all levels of bird watchers to help you learn about the birds across the Americas, Europe, Asia, Afric

Original URL: https://merlin.allaboutbirds.org/

🎉 Sara Beery, Justin Kay, Declan, Oisin Mac Aodha, Tiffany Deng, Suzanne Stathatos, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Mitch Fennell, Kakani Katija, Ed Miller, Jon Van Oast, Nico Lang, Ali Johnston, Benjamin Hoffman, Sunnie S. Y. Kim, Stefan Schneider, Dan Morris, Elijah Cole (Deactivated), Carly Batist, Ritwik, Jason Holmberg (Wild Me), Noah Giebink, Sergei Nozdrenkov

🐦 Sara Beery, Oisin Mac Aodha, Casey Youngflesh, Stefan Schneider, Carl Boettiger

🙏 Jes Lefcourt

🎤 Subhransu Maji, Stefan Schneider, Noah Giebink

Declan (declan.pizzino@consbio.org)

2022-04-18 16:59:01

Amazing. My wife and I have loved using this app while on walks and hikes. Thanks to everyone who's worked hard on this, both app and model-wise. Cool cool stuff

👍 gvanhorn

Oisin Mac Aodha (macaodha@caltech.edu)

2022-04-18 16:59:13

Very cool!

Jes Lefcourt (jeslefcourt@gmail.com)

2022-04-18 17:02:09

Totally! I was using it extensively just this weekend!

👍 gvanhorn

Ben Weinstein (benweinstein2010@gmail.com)

2022-04-18 17:32:15

amazing. What was the main innovations that increased accuracy? Just more data?

gvanhorn (grv22@cornell.edu)

2022-04-18 17:34:02

More data, smarter loss functions, smarter and more aggressive augmentation functions. We have a pretty good grip on annotating audio now and its allowed us to move way past weakly supervised methods.

Ben Weinstein (benweinstein2010@gmail.com)

2022-04-18 17:34:58

(we are still at weakly supervised methods for trees). Good to hear.

gvanhorn (grv22@cornell.edu)

2022-04-18 17:35:24

We also have a 1 week active learning cycle in place, so the annotators have a much better grip on where effort is helping, and where effort needs to be spent

Declan (declan.pizzino@consbio.org)

2022-04-18 17:35:54

the models' ability to ID even with lots of background noise is pretty impressive

Ben Weinstein (benweinstein2010@gmail.com)

2022-04-18 17:36:02

how do you find annotators? If I can be useful for connecting anyone in ecuador/columbia, let me know.

gvanhorn (grv22@cornell.edu)

2022-04-18 17:38:46

Awesome, I'll keep that in mind! Currently we just tap into folk we know from prior experience. We've built out tool ecosystem around maximizing the efforts of a few experts instead of tapping into the big citizen science community. At some point we'll probably try to embrace some of that energy.

gvanhorn (grv22@cornell.edu)

2022-04-18 17:40:00

@Declan Thank you! Yeah, that can be attributed to our more aggressive background augmentation strategies and to the detailed annotation protocol.

:the_horns: Declan

Ed Miller (ed@hypraptive.com)

2022-04-18 22:41:03

I'm looking forward to the Android update. The app really helps my birding. When I can't tell where a call or song is coming from, the Merlin ID helps me know where to look!

👍 gvanhorn, Jon Van Oast

Jon Van Oast (jon@wildme.org)

2022-04-18 23:43:14

this app is great! ......... would love to have a "mic trap" mode (is that a thing? like camera-trap but for audio?) where i could set this thing up outside and see what birds it picks up over the timeline of the day. 😄

👍 Jason Holmberg (Wild Me)

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-04-20 00:23:14

Grab some Audiomoths 🙂

Ben Weinstein (benweinstein2010@gmail.com)

2022-04-23 12:30:43

Just extensively tested this morning. The new version is really quite impressive. Huge congratulations. I love the highlight spectrogram and compare to call feature. I feel like this is a new high water mark for ecology machine learning and integration with existing tech.

💯 Jon Van Oast, Declan, Noah Giebink

Ben Weinstein (benweinstein2010@gmail.com)

2022-04-23 12:32:19

Correctly caught a single western sandpiper in a flock of sanderlings from 40 feet amongst the heavy surf.

gvanhorn (grv22@cornell.edu)

2022-04-25 09:57:43

Thanks Ben!

Priya Donti (priyald17@gmail.com)

2022-04-22 09:41:17

The Global Partnership on AI (GPAI) is launching a call for proposals to write an action-oriented roadmap for the responsible use of AI for biodiversity preservation, with proposals due May 10. See this post for more info: https://www.linkedin.com/feed/update/urn%3Ali%3Aactivity%3A6922846974228504576/

For context, GPAI is a government-recognized multi-stakeholder initiative (hosted through the OECD) that brings together leading experts from science, industry, civil society, international organizations, and government. Their goal is to provide governments and international organizations with actionable advice on the responsible use of AI.

I figure those in this group would be perfectly suited to write such a report, so I hope some of you will consider submitting to the RFP 🙂 I’m not directly affiliated with GPAI, but I and several colleagues wrote GPAI’s report on AI and climate climate last year, which we presented at the UN Climate Change Conference (COP26) - so happy to answer any questions from that perspective, or connect you to those in GPAI with more info!

🎉 Jon Van Oast, Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Kakani Katija

🤩 Sara Beery, Lama Saouma

Lama Saouma (lama.saouma@gmail.com)

2022-04-22 17:21:36

Thanks for spreading the word Priya! Happy to answer any questions (I am the contact person in the RFP)

🙌 Priya Donti

Sara Beery (sara.beery@gmail.com)

2022-04-22 11:54:35

New season of WILDLABS virtual meetups centered around movement ecology!

https://twitter.com/WILDLABSNET/status/1517065606415142913

twitter

WILDLABS Community (https://twitter.com/WILDLABSNET/status/1517065606415142913)

Virtual Meetups are back! Come along for this exciting season as we bring together leading experts to explore the future of movement ecology and uncover where we need to invest to help #tech4wildlife catapult this field forward. <https://www.wildlabs.net/virtual-meetups-season-4> <https://pbs.twimg.com/media/FQ1hQ3rVIAAZUbD.jpg>

Original URL: https://twitter.com/WILDLABSNET/status/1517065606415142913

🤩 Olivier Gimenez, Oisin Mac Aodha, Jason Holmberg (Wild Me), Kakani Katija, Carly Batist

🎉 Jon Van Oast, Jason Holmberg (Wild Me), Carl Boettiger, Dan Morris

❤️ Talia Speaker, Catherine Villeneuve, Emilio Luz-Ricca, Stephanie O'Donnell, Léonard Boussioux

Björn Lütjens (bjoern.luetjens@gmail.com)

2022-04-26 08:26:15

Does anybody know PhDs, Postdocs, or Profs in Conservation, AI, and Polar Regions? We are organizing an NSF workshop and still have room for a speaker or two :))

🐻‍❄️ Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Daniel Grzenda

Holger Klinck (hk829@cornell.edu)

2022-04-26 08:29:09

Yes, Heather Lynch at Stoney Brook. She uses AI and sat images to monitor penguin colonies.

👍 Devis Tuia, Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, aruna

Emily Charry Tissier (hello@whaleseeker.com)

2022-04-26 09:07:39

@Justine Boulent would be fantastic!

👍 Sara Beery

🐳 Justine Boulent

Akronix (akronix5@gmail.com)

2022-04-28 03:02:52

does anyone know if the WILDLABS session of yesterday was recorded: "Virtual Meetup: Data Collection in Movement Ecology"?

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2022-04-28 05:27:06

It was!

👍 Ștefan Istrate, Oisin Mac Aodha, Sara Beery, Justine Boulent, Akronix, Carly Batist

🤩 Jon Van Oast

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2022-04-28 05:27:26

We're processing it now - we'll post the full meetup and the individual talks to our youtube channel and wildlabs

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2022-04-28 05:27:36

possibly today, definitely by early next week

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2022-04-28 05:29:01

Also - the next session about data analysis is now open for registrations, it'll be of particular interest to this group. @Sara Beery is one of fantastic speakers

🙌 Akronix, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Dan Morris, Carly Batist

🎉 Jon Van Oast, Talia Speaker

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2022-04-28 05:29:40

https://www.eventbrite.co.uk/e/virtual-meetup-data-analysis-in-movement-ecology-tickets-328359842127

Eventbrite

Virtual Meetup: Data Analysis in Movement Ecology

On Wednesday, May 11 at 4:00 PM BST / 11:00 AM EDT, our expert panel will discuss developments on the horizon in movement data analysis.

Original URL: https://www.eventbrite.co.uk/e/virtual-meetup-data-analysis-in-movement-ecology-tickets-328359842127

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2022-05-02 15:52:53

"Leopards and hyenas and belugas. Oh my!"

Dan Morris (agentmorris@gmail.com)

2022-05-02 16:03:05

I learned about this competition from the beluga data set:

https://www.drivendata.org/competitions/96/beluga-whales/

DrivenData truly has the market cornered on conservation puns. See:

https://www.drivendata.org/competitions/59/camera-trap-serengeti/

DrivenData

Where's Whale-do?

Help the Bureau of Ocean Energy Management (BOEM), NOAA Fisheries, and Wild Me accurately identify endangered Cook Inlet beluga whales from photographic imagery. Scalable photo-identification of individuals is critical to population assessment, management, and protection for these endangered whales.

Original URL: https://www.drivendata.org/competitions/96/beluga-whales/

DrivenData

Hakuna Ma-data: Identify Wildlife on the Serengeti with AI for Earth

Can you predict which animals are present in camera trap images? Leverage millions of images of animals on the Serengeti to build a classifier that distinguishes between gazelles, lions, and more!

Original URL: https://www.drivendata.org/competitions/59/camera-trap-serengeti/

😅 Carly Batist

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2022-05-02 16:04:57

I think that was physical pain I just felt with "Hakuna ma-data".

🤩 Jon Van Oast

Jason Parham (bluemellophone@gmail.com)

2022-05-02 16:06:04

@Dan Morris “Whale Known Belugas”, “Belugas Got Back”, and “Fluke Box Hero” were also floated internally before the competition launched. One of the hardest things we had to do to get the competition set up was deciding on the best pun.

🙂 Dan Morris

😍 Declan

😂 Carly Batist

Dan Morris (agentmorris@gmail.com)

2022-05-02 16:06:57

I'll make you a deal... if you do another competition next year, and you call it Fluke Box Hero, I will personally record the parody theme song.

👀 Jason Parham

🎸 Greg Lipstein, Mitch Fennell, Carly Batist

Jason Parham (bluemellophone@gmail.com)

2022-05-02 16:07:12

Shoutout to @Greg Lipstein and the rest of the DataDriven team for making that decision really hard.

🎉 Jon Van Oast

Jason Parham (bluemellophone@gmail.com)

2022-05-02 16:07:41

done
giphy logo
Posted using /giphy

Nicholas Osner (nicholasosner@gmail.com)

2022-05-05 08:46:51

Hi everybody, I am Nicholas Osner from WildEye Conservation. Thanks @Dan Morris for inviting me. I look forward to being more involved in the community.

I would like to make you all aware of the project I have been working on called TrapTagger. It's a web application that uses AI to process camera trap data, and was developed in close collaboration with WildCRU - part of the University of Oxford's Zoology Department. It's open source and completely free to use.

At present, our species classifier is only capable of identifying Southern African species, with a full sub-Saharan model in the works. However, we do use MegaDetector to perform the job of blank-image removal, and we also have a heavily-optimised annotation workflow, so you can easily use our site as a nice GUI wrapper around MegaDetector for your particular biome. Moreover, we are open to training more biome-specific models if you have sufficient annotated data to do so (which you can easily generate through TrapTagger).

You can find more information on our website, including some more detailed reports in the documentation section. Additionally, you can find our repo here.

I would also like to draw your attention to another project we have in the works called the Elephant Survey System where we aim to both reduce the cost, and improve the accuracy of elephant surveys using a combination of AI and a light-aircraft-mounted camera rig, which you can read about here. If you would like to make use of the associated aerial elephant data set for your own purposes, we have available it here.

If you have any questions, please don't hesitate to get in touch with me.

WildEyeConservation/TrapTagger

AI-Powered Camera-Trap Image Processing

Language

Python

Last updated

a month ago

🙌 Stephanie O'Donnell, Ștefan Istrate, Sara Beery, gvanhorn, Oisin Mac Aodha, Brandon Davis, Carly Batist, Omiros Pantazis, Justin Kay, Jason Holmberg (Wild Me), Talia Speaker, Dan Morris, Akronix, Emilio Luz-Ricca, Antón Álvarez, Sinan Robillard, Juan Arrechea

😎 Jon Van Oast

Ștefan Istrate (stefan.istrate@gmail.com)

2022-05-05 09:04:52

Welcome and congrats for the well designed TrapTagger!

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-05-05 09:19:55

Awesome!! Will put this on our To-Add list for the next Cons Tech Directory update

Nicholas Osner (nicholasosner@gmail.com)

2022-05-06 04:35:02

Thanks, it's much appreciated!

Ben Weinstein (benweinstein2010@gmail.com)

2022-05-10 12:37:40

Just bringing this conversation to the wider community. I am writing a tree species classification paper (120 species, full continent scale). The question from an ecology collaborator on reading the draft was that our accuracy of about 65% is lower than many machine learning papers in ecology. I'm copying my response below and welcome discussion from others.

"One of my ongoing concerns for publication and presentations is the overall low accuracy rates.  I have not done a complete survey of the literature by any means, but it seems to me that machine learning prediction rates are around 70%.   Just wondering what your thoughts on if and how to deal with that."

I think your point is true, but indicative of a problem rather than a benefit. There is a pernicious bias in the ecological machine learning literature to 'solving' problems. It is a bit of paradox because every introduction wants to set an applied problem up as being crucial, but every result section wants to show high accuracy values. The result is that there is a real epidemic of lax evaluation criteria. Either through a lack of data, or through weak rigor in setting up the problem that actually reflects the full breadth of the difficulty during prediction. The above figure is a perfect example of that. It is completely natural to do a random train/test split for each class and assess the accuracy per species. If you only had NEON data, you would be deceived into thinking that your accuracy when it comes time to make prediction on the full site scale is more than 150% higher than it actually is. The funny thing is that these problems seem to manifest mostly in the applied papers and analysis. Pure computer vision research that focuses solely on a specific theory doesn't have the pretense of solving any applied problem, and often has low evaluation scores because most problems that are worth solving 1) need lots of data, 2) are genuinely hard. For example, the famous mask-rcnn paper, which has nearly 18,000 citations in 3 years. The first analysis has an average precision of less than 0.4 (https://arxiv.org/abs/1703.06870). As I put together and outline and figures for the paper, my goal is to articulate that tree species prediction, when framed from the perspective of wanting to make predictions on the scale of tens of millions of trees, is vastly harder than the literature would have you believe. The ugly consequence of this dynamic is that applied ecologists become numb and distrusting to results in these papers, since every paper purports to have outstanding accuracy scores. For example in the tree crown delineation task, literature has been reporting greater than 85% accuracy for nearly 15 years, which really begs the question, why would people keep writing papers about such an easy and already solved problem? The only real conclusion is that the analysis as constructed doesn't reflect the actual needs of the applied workflow, but rather makes assumptions either due to a lack of data, or the desire to have higher values. We can show that our model is 2.5x better than a standard off the shelf approach, I think is the crux of it.

arXiv.org

Mask R-CNN

We present a conceptually simple, flexible, and general framework for object instance segmentation. Our approach efficiently detects objects in an image while simultaneously generating a...

Original URL: https://arxiv.org/abs/1703.06870

👏 Dan Morris, Ando Shah, Declan, Kirsten Crane, Riccardo de Lutio, Omiros Pantazis, Justine Boulent, Sara Beery, Paige Ngo, Kasirat, Emilio Luz-Ricca, Elijah Cole (Deactivated)

👍 Justin Kay, Sara Beery, Rita Pucci

Dan Morris (agentmorris@gmail.com)

2022-05-10 14:25:04

Mic drop! Really well-written. I think there are really two issues here: (a) artificial inflation of accuracy numbers via (IMO usually unintentional) questionable evaluations (which you mention), and (b) an assumed coupling of "accurate" and "useful". Addressing (b) likely requires us to start incorporating more human subjects studies, and also some less-quantitative, more anecdotal case studies about whether/how/why AI systems are actually getting used/abandoned. Exciting that some of those have started to roll out in the last couple of years.

🙌 Mitch Fennell

Daniel Davila (daniel.davila@kitware.com)

2022-05-10 14:34:19

"machine learning prediction rates are around 70%" - is this for a specific, well studied task or benchmark? It should be noted that you cant generalize performance like that across datasets. A reasonable score on e.g. coco isnt necessarily a reasonable score on crowdhuman. And when you talk about the constraints of most real world problems, namely that there is usually a significant lack of relevant data, you can usually take SOTA and chop it in half for out of the box performance (without significant engineering around the ML model itself to boost performance)

Daniel Davila (daniel.davila@kitware.com)

2022-05-10 14:35:35

That being said, what's right here may be beside the point of whether or not you are queuing up an uphill battle with reviewers. If the magic number for a "good" paper is 70%, or 85%, or whatever, that's as good a reason as any to get rejected these days

🎉 Ben Weinstein, Sara Beery

Pietro Perona (perona@caltech.edu)

2022-05-14 14:09:24

Nice discussion. Sometimes there is need for a study that identifies specific challenges that algorithms face and labels datasets in ways that allow you to benchmark against those challenges. You will find two examples here and here. If you guys are interested in doing this for tree species identification I will be happy to have a discussion with you to help bring clarity to this area.
One more suggestion: iNaturalist is a natural benchmark and you should be able to use the iNaturalist API to benchmark your algorithm against iNaturalist on the iNaturalist dataset. @gvanhorn will be able to tell you if this is feasible.

ieeexplore.ieee.org

Pedestrian Detection: An Evaluation of the State of the Art

Pedestrian detection is a key problem in computer vision, with several applications that have the potential to positively impact quality of life. In recent years, the number of approaches to detecting pedestrians in monocular images has grown steadily. However, multiple data sets and widely varying evaluation protocols are used, making direct comparisons difficult. To address these shortcomings, we perform an extensive evaluation of the state of the art in a unified framework. We make three primary contributions: 1) We put together a large, well-annotated, and realistic monocular pedestrian detection data set and study the statistics of the size, position, and occlusion patterns of pedestrians in urban scenes, 2) we propose a refined per-frame evaluation methodology that allows us to carry out probing and informative comparisons, including measuring performance in relation to scale and occlusion, and 3) we evaluate the performance of sixteen pretrained state-of-the-art detectors across six data sets. Our study allows us to assess the state of the art and provides a framework for gauging future efforts. Our experiments show that despite significant progress, performance still has much room for improvement. In particular, detection is disappointing at low resolutions and for partially occluded pedestrians.

Original URL: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=5975165

Pietro Perona (perona@caltech.edu)

2022-05-14 14:13:04

Thread Reply:

Ritwik (rittyun@yahoo.com)

2022-05-20 09:15:41

sorry for coming to this so late.. thanks for raising this topic and for the excellent well rounded argument.. the origins of this sentiment seem to be the trickle down effect of the publication bias in journals for so called "striking" work expected by editors.
As for evaluation, i think it's crucial to test the models on "out of distribution" data. we all develop these models in the future hope of them being applied in real life usage, while the standard train-test splits can be far away from what the model will actually see.. it can be hard to get relevant out of distribution data but it gives a more honest expectation of how the model is expected to react to real life.

💯 Carl Boettiger

Sara Beery (sara.beery@gmail.com)

2022-05-11 10:00:08

Starting in one hour!

https://twitter.com/WILDLABSNET/status/1523690388581920769

twitter

WILDLABS Community (https://twitter.com/WILDLABSNET/status/1523690388581920769)

<a href="https://twitter.com/VikramIyerUW">@VikramIyerUW</a> Join us and over 330 participants on Wednesday, May 11 at 4:00 PM BST / 11:00 AM EDT for our next Virtual Meetup exploring the latest advancements and emerging developments in data analysis for movement ecology! Register here for your free ticket: <https://www.eventbrite.co.uk/e/virtual-meetup-data-analysis-in-movement-ecology-tickets-328359842127>

Original URL: https://twitter.com/WILDLABSNET/status/1523690388581920769

🙌 Omiros Pantazis, Stephanie O'Donnell, Oisin Mac Aodha, Yihang She, Thijs, Declan, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Mitch Fennell, Talia Speaker

Thijs (thijs@q42.nl)

2022-05-11 10:44:36

Too bad I can't join, is there a recording later on?

Sara Beery (sara.beery@gmail.com)

2022-05-11 10:47:06

Yes! They always post recordings after, I'll share the link when it's up!

Thijs (thijs@q42.nl)

2022-05-11 10:47:55

Awesome, have fun!

Sara Beery (sara.beery@gmail.com)

2022-05-11 11:42:03

Welcome to everyone joining from the WILDLABS movement ecology meetup! Feel free to introduce yourselves 🤩

👋 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Kenady Wilson, Stephanie O'Donnell, Jason Holmberg (Wild Me), Paige Ngo, Sophia Abraham, Ted Schmitt, Sinan Robillard, Jason Parham, Fagner Cunha, Carly Batist

🎉 Jon Van Oast, Stephanie O'Donnell, Jason Holmberg (Wild Me), Talia Speaker, Sophia Abraham, Carly Batist

👍 Oorjit Mahajan, Stephanie O'Donnell, Jason Holmberg (Wild Me), Sophia Abraham

Alex Borowicz (alex.borowicz@stonybrook.edu)

2022-05-12 12:27:46

Is there anyone in the NYC area who'd be interested in joining forces to put together a workshop proposal for the Student Conference on Conservation Science in October? Deets on the conference here, but in short it's a really awesome and supportive little conference that's all about building up undergrad and grad students in conservation. And it's at the American Museum of Natural History which makes it super fun.

I'm envisioning a pretty basic workshop aimed at getting participants with absolutely no background in AI familiar with what's out there that's easy to use/get started with to help with research they're already doing/planning, how to start thinking about data from a computer's perspective, some basic demonstrations

❤️ Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Sara Beery, Monty Ammar, Carly Batist

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-05-24 20:19:21

Hi! I'm a PhD student at CUNY and love that conference! It would be a great workshop opportunity for sure

Kasirat (kasirat_turfi@hotmail.com)

2022-05-13 04:02:46

Is anyone here working with forest/tree canopy image data? I am using Mask-RCNN to delineate the canopies, the dataset is very small. Any tips or tricks would be very appreciated. If anyone has applied Mask-RCNN on their tree canopy image data, how did it perform? Generally with a pretrained model (on COCO) as a starter, it is expected that the results would be quite acceptable, but my model performance is nowhere near "acceptable". Specially in segmenting the canopies, detection is okay, but getting many overlapping boxes generated for the same canopy.

Devis Tuia (devis.tuia@epfl.ch)

2022-05-13 05:12:47

Especially if the forest is broadleaf it is very difficult to make single trees segmentation from images. We were working on that in the past (segmenting single trees from images or Lidar scans) and lidar was very very useful

Devis Tuia (devis.tuia@epfl.ch)

2022-05-13 05:15:35

even by eyes often you don’t see where a tree finishes and where the next one starts and forest is inherently 3D. It’s a though problem, there are a lot of papers in journals like Remote Sensing of Environment and entire confs like silvilaser or forestsat

Kasirat (kasirat_turfi@hotmail.com)

2022-05-13 05:26:59

Yes, it is very difficult to even tell apart one tree from the other by human eyes (I couldn't do it for the life of me). Looking for ways to find a solution. I am not looking for over 90% accuracy, anything over 70% would be good. Do not have high hopes with Mask Rcnn on crowded trees.

Daniel Davila (daniel.davila@kitware.com)

2022-05-13 10:01:28

One idea would be to bootstrap a larger training set with someone else's model. Have you looked into some of the open source tree inventory models out there? TreeTect comes to mind, there are a few others too.

https://github.com/krakchris/TreeTect

Is the goal to discern individual trees or just segment out the region in the image that is or isnt tree?

Ben Weinstein (benweinstein2010@gmail.com)

2022-05-13 10:04:47

I can't claim to be an expert because data vary so widely, but we have spent several years on the problem. A couple observations. 1) Because there is massive intra-class variance inside of 'tree', you need alot of training data, especially if you want to generalize to large forests, or across geographies. Pretraining is key (https://www.sciencedirect.com/science/article/pii/S157495412030011X). 2) Really consider if you need semantic segmentation through mask rcnn or if bounding boxes are sufficient. We have found that the extra effort for training a pixel level model is not worth it. Many trees can be approximated by 4 point boxes and that does not inhibit downstream analysis. It allows you to annotate many more images faster. We use torchvision's retinanet and have released a pretrained model that I'd love you to try. Finetuning it to your data usually works well. https://deepforest.readthedocs.io/. 3) This one is a bit controversial, but outside of using the unsupervised LiDAR algorithms for pretraining in the RGB, i have never seen a LiDAR algorithm that outperforms RGB alone. The caveat here is that i think it depends alot on LiDAR point density. The majority of LiDAR papers use hundreds of points per meter through intense data collection of a small area. When you want to monitor millions of trees, this level of data collection is usually not possible. NEON LiDAR data averages about 4 points per meter over large areas. At this scale, the LiDAR is not that useful, and a canopy height model is often helpful. I do think any multi-spectral data may be useful, we have shown it helps in tropical forests (https://ieeexplore.ieee.org/abstract/document/9387530). 4). If I had to put money on it, the future is in massive pretrained models, after significant research on scaling across resolution. What resolution is your data?

sciencedirect.com

Cross-site learning in deep learning RGB tree crown detection

Tree crown detection is a fundamental task in remote sensing for forestry and ecosystem ecology. While many individual tree segmentation algorithms ha…

Original URL: https://www.sciencedirect.com/science/article/pii/S157495412030011X

ieeexplore.ieee.org

Multisensor Data Fusion for Improved Segmentation of Individual Tree Crowns in Dense Tropical Forests

Automatic tree crown segmentation from remote sensing data is especially challenging in dense, diverse, and multilayered tropical forest canopies, and tracking mortality by this approach is even more difficult. Here, we examine the potential for combining airborne laser scanning (ALS) with multispectral and hyperspectral data to improve the accuracy of tree crown segmentation at a study site in French Guiana. We combined an ALS point cloud clustering method with a spectral deep learning model to achieve 83% accuracy at recognizing manually segmented reference crowns (with congruence >0.5). This method outperformed a two-step process that involved clustering the ALS point cloud and then using the logistic regression of hyperspectral distances to correct oversegmentation. We used this approach to map tree mortality from repeat surveys and show that the number of crowns identified in the first that intersected with height loss clusters was a good estimator of the number of dead trees in these areas. Our results demonstrate that multisensor data fusion improves the automatic segmentation of individual tree crowns and presents a promising avenue to study forest demography with repeated remote sensing acquisitions.

Original URL: https://ieeexplore.ieee.org/abstract/document/9387530

🌟 Carl Boettiger

👍 Kasirat

Ben Weinstein (benweinstein2010@gmail.com)

2022-05-13 10:08:52

We have annotations here as well. https://github.com/weecology/NeonTreeEvaluation If you would like to use them for your method.

weecology/NeonTreeEvaluation

Benchmark dataset for tree detection for airborne RGB, Hyperspectral and LIDAR imagery

Stars

Language

Python

Devis Tuia (devis.tuia@epfl.ch)

2022-05-13 10:29:53

+1 for the Lidar point cloud argument, with a few returns per square meter it get complicated. to segment at scale

👍 Sara Beery

Jon Van Oast (jon@wildme.org)

2022-05-13 13:39:48

@Jason Holmberg (Wild Me) - werent we just talking about this a couple days ago?

Ben Weinstein (benweinstein2010@gmail.com)

2022-05-13 14:09:54

https://www.mdpi.com/2072-4292/14/6/1317 here is an example of that LiDAR challenge. A extremely thorough well done paper, probably a year's worth of diligent work. Is only about 5% better than a completely off the shelf RCNN that you could run in a couple days. I would be willing to guess that if that's true, than the pretraining we've done on deepforest is probably equal to that, and this was in higher resolution LiDAR data. Its strange, and I do hope that eventually a massive, multi-sensor with RGB/CHM/HSI pretrained algorithm will outperform, but I think the labels are lacking to truly evaluate it.

MDPI

Individual Tree Detection in Urban ALS Point Clouds with 3D Convolutional Networks

Since trees are a vital part of urban green infrastructure, automatic mapping of individual urban trees is becoming increasingly important for city management and planning. Although deep-learning-based object detection networks are the state-of-the-art in computer vision, their adaptation to individual tree detection in urban areas has scarcely been studied. Some existing works have employed 2D object detection networks for this purpose. However, these have used three-dimensional information only in the form of projected feature maps. In contrast, we exploited the full 3D potential of airborne laser scanning (ALS) point clouds by using a 3D neural network for individual tree detection. Specifically, a sparse convolutional network was used for 3D feature extraction, feeding both semantic segmentation and circular object detection outputs, which were combined for further increased accuracy. We demonstrate the capability of our approach on an urban topographic ALS point cloud with 10,864 hand-labeled ground truth trees. Our method achieved an average precision of 83% regarding the common 0.5 intersection over union criterion. 85% percent of the stems were found correctly with a precision of 88%, while tree area was covered by the individual tree detections with an F1 accuracy of 92%. Thereby, we outperformed traditional delineation baselines and recent detection networks.

Original URL: https://www.mdpi.com/2072-4292/14/6/1317

Kasirat (kasirat_turfi@hotmail.com)

2022-05-13 20:37:49

@Daniel Davila thankyou for this link, I have not seen it before. I am working with forest data, my search has been limited, haven't thought about urban tree data.

Kasirat (kasirat_turfi@hotmail.com)

2022-05-13 20:51:58

@Ben Weinstein I have been pouring over this new paper last few days, the one you have linked here on urban individual tree detection from lidar point cloud.

I have not started working on this project too long ago, new to lidar point cloud. My actual data is ALS point cloud, and I want to segment individual trees. There hasn't been a ton of study on application of deep learning to segment individual trees in point cloud. I am going about it systematically, trying out the projection based method, ie. taking the bird's eye view, getting the 2D raster, applying image based instance segmentation. I had hoped that masks would better delineate a tree than bounding boxes. However, at this stage, I can appreciate your conclusion on masks not really improving on bounding box situation, specially when the trees are dense.
I may have to abandon the Mask-RCNN idea, and look at 3D object detection/3D instance segmentation algorithms.
Another incentive to first start with a Mask-RCNN was to establish a base-line, to compare with another approach.

Kasirat (kasirat_turfi@hotmail.com)

2022-05-13 20:55:34

@Jon Van Oast if you and other people are working on point cloud of forests, would love to know more about your work!
I am interested in "point cloud", "deep learning" and "segmentation"!

👍 Jon Van Oast

Jon Van Oast (jon@wildme.org)

2022-05-13 20:58:20

@Kasirat - at this point we are just at the very beginning of wondering what work is out there. so this thread is great!

👍 Kasirat

Fernando Pérez (fernando.perez@berkeley.edu)

2022-05-13 18:56:47

Hi everyone! You may have already seen this, but in case it didn’t make it on your radar - our DS4E Initiative at Berkeley is hiring an executive director!

I hope some of you may find this position interesting, and please feel free to share with colleagues. You can find more about the initiative here, but I’m happy to answer any questions you might have.

Nature Careers

Executive Director | Nature Careers

Executive Director, with University of California Berkeley (UC Berkeley). Apply Today.

Original URL: https://www.nature.com/naturecareers/job/executive-director-university-of-california-berkeley-uc-berkeley-758048

Berkeley News

New UC Berkeley center will apply data science to solving environmental challenges

Open science and collaboration with communities is at the heart of the Eric and Wendy Schmidt Center for Data Science and Environment

Original URL: https://news.berkeley.edu/2022/03/23/new-uc-berkeley-center-will-apply-data-science-to-solving-environmental-challenges/

‼️ Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Sara Beery

💚 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Suzanne Stathatos, Sara Beery

Monty Ammar (montyx23@gmail.com)

2022-05-15 16:49:47

Hey everyone, im at the University of Kent doing a masters in Conservation Biology.

I’ve been working on a review titled: “Machine Learning in Conservation Science: a review”. It started out as a uni project, but I’m gathering some academics from our department who have experience in ML & conservation to collaborate and I aim to submit to ‘Biological Conservation’ for publication by the end of the summer.

I know there are already a number of great reviews (@Devis Tuia @Ben Weinstein @Sara Beery @Benjamin Kellenberger and im sure some other authors on this slack too) on ML in ecology and a fantastic perspectives paper in Nature. I envisage this review article covering a slightly different angle than the ones mentioned. Mainly because it will focus on the progress of ML in conservation science holistically rather than individually the ecology, deep learning, CV, or SDM applications. E.g. tracking illegal wildlife trade on social media/internet, optimising ranger patrols, automated biodiversity assessments.

The aim is to inform practitioners and researchers from all corners of conservation science about the range of problems ML is tackling in this field now.

I’d like to Invite people from here with experience in ML in conservation science (i.e bioacoustics, CV, decision making/ management optimisation etc) to collaborate on the review by reviewing a few studies and writing a couple paragraphs on them to go in the manuscript!

Please reach out to me if you want to help! It seems crazy to write a review on this topic and not get input from this group! 😁

🙌 Sara Beery, Declan, Ted Schmitt, Merlin Bleile, Carly Batist, Sinan Robillard

👍 Rita Pucci

Ted Schmitt (teds@allenai.org)

2022-05-16 19:17:34

Please, please be sure to post a link to the review here once it is published.

👍 Monty Ammar, Sara Beery

Monty Ammar (montyx23@gmail.com)

2022-05-16 20:13:39

Definitely will @Ted Schmitt 👌

Devis Tuia (devis.tuia@epfl.ch)

2022-05-17 07:16:49

Sounds like a relevant review, very complementary to our perspective paper! If you want have a chat on do’s and don’ts when writing a paper like this, just PM me!

❤️ Monty Ammar

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2022-05-17 08:04:49

I second that; happy to talk about it!

❤️ Monty Ammar

Sara Beery (sara.beery@gmail.com)

2022-05-17 10:18:32

Agreed 🙂

❤️ Monty Ammar

Monty Ammar (montyx23@gmail.com)

2022-05-17 12:32:29

Perfect. 🙏

Victor Anton (victor@wildlife.ai)

2022-05-24 23:38:44

@Monty Ammar it's not a scientific article but sharing here a recent report on the use of AI for the NZ environment in case it's of any relevance for your review https://aiforum.org.nz/2022/05/18/ai-for-the-environment-in-aotearoa-new-zealand/

AI Forum

AI for the Environment in Aotearoa New Zealand - AI Forum

Learn how AI can be used as a tool for positive environmental outcomes. Discover the opportunities and explore the possibilities.

Written by

AI Forum

Est. reading time

3 minutes

Original URL: https://aiforum.org.nz/2022/05/18/ai-for-the-environment-in-aotearoa-new-zealand/

Monty Ammar (montyx23@gmail.com)

2022-05-25 05:17:01

Thank you @Victor Anton 🙂

👍 Victor Anton

Lukas Picek (lukaspicek@gmail.com)

2022-05-17 05:57:53

Dear All,

I want to invite you to participate in the LifeCLEF 2022 Snake ID Challenge! We need humans (novices and experts) to identify the same images to compare humans with AI-based algorithms. This project will help you practice your snake ID skills and will also provide valuable data on which species are most commonly misidentified as which other species.

Through June 15th, you will identify 150 images of snakes from around the world. We'll publish bios of the top 3 IDers, and the top IDers get a book of their choice.

If you decide to join, you will be shown a picture of a snake and its country of origin. Your task is to recognize the snake species. Use of google search or citizen science platform to identify it is expected. Any existing "tool: is allowed with one exception -- Machine Learning-Based systems, e.g., Google Image Search, iNaturalist recognition tool, HerpMapper recognition toll etc.

More detail and link to the snake ID page at the end of the consent form/survey: https://forms.gle/awongX8Fk3tX2zGG6
Please be sure to use the same email address for the consent form and the lab.citizenscience.ch website where the snake IDs are so that we can match your IDs with your responses to the survey.

Thank you for considering your participation!

👍 Oisin Mac Aodha, Leonardo Viotti, Dan Morris

🐍 Sara Beery, Lukas Picek, Carly Batist

Rita Pucci (rita.pucci85@gmail.com)

2022-05-18 11:04:10

Hi everyone I am Rita from the University of Udine (north Italy),
I have studied ML and I am almost new to the world of interdisciplinary works in the field of Computer Vision and Conservation. I really would like to be more in this world so here I am 😄
Hope to contribute to this field and collaborate for the best!
Nice to meet all of you, I know some of you from literature and from social media (well.. it is true..). I am willing to collaborate and in helping other researchers if I can!

👋 Sara Beery, Felipe Parodi, Jason Holmberg (Wild Me), Devis Tuia, Fagner Cunha, Omiros Pantazis, Monty Ammar, Jon Van Oast, Akronix, Ritwik, Benjamin Kellenberger

Greg Lipstein (greg@drivendata.org)

2022-05-18 15:28:01

Hi all! - If anyone is looking for an accessible way to learn or practice with computer vision for conservation, a couple data scientists on our team just put out a practice wildlife image classification challenge with some camera trap images and 8 classes.

The data comes from partners at the Wild Chimpanzee Foundation and the Max Planck Institute for Evolutionary Anthropology. There is also an introduction to image classification using camera trap images blog post that walks through an initial approach to the challenge.

Please feel free to share with any students, data enthusiasts, etc. who might be interested in getting some practice in this field! Thanks!

DrivenData

Conser-vision Practice Area: Image Classification

Looking for a great way to start working with computer vision? This competition features a small dataset of wildlife captured by camera traps used in conservation research.

Original URL: https://www.drivendata.org/competitions/87/competition-image-classification-wildlife-conservation/

❤️ Suzanne Stathatos, Felipe Parodi, Sara Beery, Fagner Cunha, Stephanie O'Donnell, Dongmin (Dennis) Kim, Rita Pucci, Carly Batist, nyakundi lamech

😎 Jon Van Oast, Sara Beery

👏 Dan Morris

Reece Rhinehart (rhin0098@pacificu.edu)

2022-05-18 16:27:12

Thanks for this!

Sara Beery (sara.beery@gmail.com)

2022-05-19 10:47:35

@gvanhorn is giving a talk today on Merlin Bird ID!!

https://ecornell.cornell.edu/keynotes/overview/K051922/?utm_content=buffer6bab8&utm_medium=social&utm_source=twitter.com&utm_campaign=buffer

eCornell

What’s That Bird?

Merlin Bird ID is a free app from the Cornell Lab of Ornithology that helps you identify the birds you see and hear around the world. This feature-rich bird identification app provides photo ID, sound ID, customized bird lists based on location, and more. In this one-hour webcast, Cornell Lab of Ornithology’s Grant Van Horn, Jessie Berry, and Drew Weber describe how they used the power of citizen scientists, artificial intelligence, and machine learning to create the most advanced bird identification app available.We’ll walk through how Merlin works its magic and share how Merlin can be used to inspire curiosity and advance conservation. Join us for a live demonstration of the app and discover all the features available in Merlin. Image Credit: ©Brad Imhoff/Macaulay Library

Original URL: https://ecornell.cornell.edu/keynotes/overview/K051922/?utm_content=buffer6bab8&utm_medium=social&utm_source=twitter.com&utm_campaign=buffer

🐦 Oisin Mac Aodha, Andrés C Rodríguez, Justin Kay, Sunnie S. Y. Kim, Jason Parham, Sophia Abraham

👍 Kakani Katija, Jason Parham, Dan Morris, Sophia Abraham, Victor Anton

🎉 Jon Van Oast, Suzanne Stathatos, Sophia Abraham

Suzanne Stathatos (suzanne.stathatos@gmail.com)

2022-05-20 13:36:55

Our slack founder, @Sara Beery, defended her thesis today 🎉 I could not be more proud of all she's done and excited to see where she’ll go!

💕 Jon Van Oast, Declan, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Stefan Schneider, Jason Holmberg (Wild Me), Sophia Abraham, Daniel Davila, Lauren E. Gillespie, Chris Yeh, David, Elijah Cole (Deactivated)

🍾 Jon Van Oast, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Ando Shah, Ștefan Istrate, Stefan Schneider, Jason Holmberg (Wild Me), Omiros Pantazis, Frederic Fol Leymarie, Peter Bull, Casey Youngflesh, Akronix, Sophia Abraham, Nikhil Vytla (he/him), Daniel Davila, Dhruv Sheth, Mitch Fennell, Peter van Lunteren, Talia Speaker, Lauren E. Gillespie, Ted Schmitt, Chris Yeh, Carly Batist, Sicily Fiennes, David, Antón Álvarez

🎉 Jes Lefcourt, Declan, Justin Kay, Kakani Katija, Thor Veen, Justine Boulent, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Beckett Sterner, Stefan Schneider, Avi Sundaresan, Jason Holmberg (Wild Me), Sunnie S. Y. Kim, Olivier Gimenez, Emily Huong, Fagner Cunha, Peter Bull, Alex Brace, Yihang She, Timm Haucke, Grace Hansen, Matt Weldy, Dan Morris, Sophia Abraham, Nikhil Vytla (he/him), Daniel Davila, Dhruv Sheth, Hemal Naik, Thijs, Riccardo de Lutio, Rowan Converse, Lucia Gordon, Kasirat, Kewal Shah, Emilio Luz-Ricca, Marta Skreta, Nicholas Osner, Lloyd Hughes, Talia Speaker, Joanna Turner, Lauren E. Gillespie, Stephanie O'Donnell, Dongmin (Dennis) Kim, Carly Batist, Jacob Kamminga, Ritwik, Sachith Seneviratne, David, Cody Kupferschmidt, Juan Sebastián Cañas Silva

❤️ Monty Ammar, Ben Weinstein, Sophia Abraham, Nikhil Vytla (he/him), Daniel Davila, Talia Speaker, Chris Yeh, David, Juan Sebastián Cañas Silva

👍 Matt Hron, Ivory lu, Chris Yeh, David, Barry Brook

Sara Beery (sara.beery@gmail.com)

2022-05-20 14:17:45

Thank you!!!!

Stefan Schneider (sschne01@uoguelph.ca)

2022-05-20 14:21:23

Congrats Sara! Your defence was a great ride through all the amazing work you've done! And only the start of an incredible career ahead 😄

❤️ Sara Beery

Kakani Katija (kakani@mbari.org)

2022-05-20 14:23:37

Congratulations!!!! Can't wait to hear what's next.

❤️ Sara Beery

Alex Borowicz (alex.borowicz@stonybrook.edu)

2022-05-20 15:12:07

Such a good talk! Congratulations, Doctor!

❤️ Sara Beery

Chris Lang (chrislang@ucsb.edu)

2022-05-20 15:36:54

Congratulations Sara!! I had to leave at 10 but any chance you can share the recording?

Sara Beery (sara.beery@gmail.com)

2022-05-20 15:40:08

Yes! I'll post the recording :)

👍 Casey Youngflesh

🙌 Nikhil Vytla (he/him), Talia Speaker

Monty Ammar (montyx23@gmail.com)

2022-05-20 17:33:39

Congratulations Sara 👏

Holger Klinck (hk829@cornell.edu)

2022-05-20 14:19:18

Congrats, Sara. You are a rockstar 🙂

❤️ Sara Beery

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2022-05-20 14:22:37

Congrats, Dr. Beery!

❤️ Sara Beery

Jon Van Oast (jon@wildme.org)

2022-05-20 14:30:29

very exciting!! congratulations. 🎉

❤️ Sara Beery

Fagner Cunha (fagner.cunha@icomp.ufam.edu.br)

2022-05-20 14:49:45

Congrats, @Sara Beery! 🎉🎉🎉

❤️ Sara Beery

Jake Wall (walljcg@gmail.com)

2022-05-21 06:05:08

Great job Sara!!

❤️ Sara Beery

Kasirat (kasirat_turfi@hotmail.com)

2022-05-21 16:50:17

Congratulation @Sara Beery! Would love to watch the presentation if there is a link to the recording 😄

❤️ Sara Beery, Zac Winzurk

Rita Pucci (rita.pucci85@gmail.com)

2022-05-23 03:44:07

congratulation for you PhD!!

❤️ Sara Beery

Hannah Kerner (hkerner@umd.edu)

2022-05-23 11:50:04

New funding opportunity

Machine learning can help communities around the world understand, mitigate, and adapt to climate change. However, a lack of ground truth data limits the effective use of machine learning in low- and middle-income communities that are disproportionately impacted by climate change. Solutions created by and for those communities are crucial to address the climate crisis globally.

To address this need, a group of philanthropies and data scientists have created Lacuna Fund. The Fund is the world’s first collaborative effort to provide data scientists, researchers, and social entrepreneurs in low- and middle-income contexts globally with the resources they need to produce training and evaluation datasets that address urgent problems in their communities.

Lacuna Fund just launched its most recent call for proposals focused on machine learning datasets for equitable climate outcomes in climate & health. Proposals are due on 17 July. See the full announcement, with more details on eligibility and selection, on the Lacuna Fund Apply page.

Lacuna Fund

Home - Lacuna Fund

Lacuna Fund is the world’s first collaborative effort to provide data scientists, researchers, and social entrepreneurs in low- and middle-income contexts globally with the resources they need to produce labeled datasets that address urgent problems in their communities.

Original URL: https://lacunafund.org/

👍 Sara Beery, Jason Holmberg (Wild Me), Peter van Lunteren, Justin Kay, Suzanne Stathatos, Matt Weldy, Kakani Katija, Dan Morris, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Ivan Zvonkov, Rita Pucci, Merlin Bleile, Kasirat, Carly Batist, Caleb Robinson

Victor Anton (victor@wildlife.ai)

2022-05-24 23:50:25

Two post-doctoral opportunities for "AI in ecology" in Utrecht, Netherlands (PS I don't have any links with the institution or know more details about the positions, I just came across it) https://www.iamexpat.nl/career/jobs-netherlands/research-academic/two-postdoctoral-positions-ai-ecology-10-fte/391358

IamExpat

Two Postdoctoral positions AI in ecology (1.0 FTE)

Utrecht University is hiring a Two Postdoctoral positions AI in ecology (1.0 FTE) in Utrecht. Are you an expat looking for jobs in English? Apply now!

Original URL: https://www.iamexpat.nl/career/jobs-netherlands/research-academic/two-postdoctoral-positions-ai-ecology-10-fte/391358

👍 Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Oisin Mac Aodha, Rita Pucci, Déva Sou, Stefan Schneider, Jason Holmberg (Wild Me)

😎 Carl Boettiger

Amir Patel (amir.patel@uct.ac.za)

2022-05-29 15:58:08

👋 Hi everyone! I'm new here 🙂 Looking forward to working and meeting you all!

👍 Peter van Lunteren, Lloyd Hughes

😍 Sara Beery

👋 Benjamin Kellenberger, Rita Pucci

Amir Patel (amir.patel@uct.ac.za)

2022-05-29 16:01:34

My lab is looking for a post-doc in multi-sensor data fusion for wildlife biomechanics. Please see advert attached! We are based in beautiful Cape Town, South Africa if you would like a change of scenery with lots of cool animals 😁 🐆🦁

🎉 Sara Beery, Oisin Mac Aodha

👍 Benjamin Kellenberger

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2022-05-30 04:28:26

Hey Amir - do you want us to pop this on WILDLABS as well, so the broader conservation tech community can see it?

Amir Patel (amir.patel@uct.ac.za)

2022-06-01 04:14:21

@Stephanie O'Donnell yes, please! 🙂

👍 Stephanie O'Donnell

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2022-06-01 05:00:24

Does it have a close date?

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2022-06-01 05:00:53

I've put to close at 30June, but let me know if I need to change that! https://www.wildlabs.net/career-opportunity/post-doc-multi-sensor-fusion-animal-biomechanics

wildlabs.net

Post-doc in Multi-Sensor Fusion for Animal Biomechanics | WILDLABS

The African Robotics Unit is looking for a Postdoctoral Research Associate to join A/Prof. Amir Patel’s lab at UCT. The Postdoc will investigate how sensor fusion can be leveraged to understand the locomotion of wildlife in South Africa.

Original URL: https://www.wildlabs.net/career-opportunity/post-doc-multi-sensor-fusion-animal-biomechanics

Amir Patel (amir.patel@uct.ac.za)

2022-06-02 08:48:20

thank you so much Stephanie! 30 June is perfect 🙂

🙌 Stephanie O'Donnell

slackbot

2022-05-31 05:43:18

This message was deleted.

Déva Sou (soudeva974@gmail.com)

2022-05-31 07:17:11

Hi Devis, thanks for the feedback!

Alayna Van Dervort (av@thebigwild.com)

2022-05-31 18:30:38

https://crcs.seas.harvard.edu/get-involved

crcs.seas.harvard.edu

Get Involved

Join the CRCS Lab

Original URL: https://crcs.seas.harvard.edu/get-involved

❤️ Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Sara Beery, Alexander Robillard, Dhruv Sheth

Alayna Van Dervort (av@thebigwild.com)

2022-05-31 18:30:43

The Harvard Center for Research on Computation and Society (CRCS) is currently accepting applications for postdoctoral fellows

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-04 06:05:14

https://techcommunity.microsoft.com/t5/internet-of-things-blog/wildlife-monitoring-and-conservation-with-azure-percept/ba-p/3390910

TECHCOMMUNITY.MICROSOFT.COM

Wildlife Monitoring and Conservation with Azure Percept

BearID Project is a non-profit organization staffed by conservation scientists and volunteer software engineers. Our aim is to progress the field of conservation technology by developing individual identification software applied to camera trap data for use in noninvasive wildlife monitoring. Our ...

Original URL: https://techcommunity.microsoft.com/t5/internet-of-things-blog/wildlife-monitoring-and-conservation-with-azure-percept/ba-p/3390910

👍 Justin Kay, Alexander Robillard, Ed Miller

👀 Alexander Robillard, Rita Pucci, Crystal Huang

Justin Kay (justinkay92@gmail.com)

2022-06-04 12:48:13

If anyone here has experience with these I'd love to hear how they stack up to Jetson / Coral options for edge

Ed Miller (ed@hypraptive.com)

2022-06-05 19:19:15

For ease of use, the Percept is pretty good. I haven't had a chance to compare performance across these platforms. @Henrik Cox (Sentinel) & @Sam Kelly, did you do any performance or efficiency comparisons across Jetson, Coral, etc.?

Henrik Cox (Sentinel) (henrik@conservationxlabs.org)

2022-06-06 16:56:17

We tried out the Jetson a while ago and have settled on the Coral since. Still using to this day

Peter Griggs (peter@deepai.org)

2022-06-06 13:01:13

Hi All! I'm looking for individuals or teams that are interested in trying out our computer vision platform for detecting species in camera trap imagery. Anyone can use it, and it can be quickly taught to find new species in your camera trap imagery after seeing only dozens of images.

Our goal is to reduce the time teams spend reviewing camera trap imagery by 80-90%

There's a free community version we host that anyone can start training on their own camera trap images here: https://deepai.org/zendo

I'd be happy to chat with anyone interested as well!

❤️ Sara Beery, Alexander Robillard

👍 Talia Speaker, Carly Batist, Alexander Robillard, Dan Morris

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-06 14:29:41

Just out of curiosity, what’s the difference between this and other software ou there for camera trap image processing, (Wildlife Insights, Zamba, MegaDetector, Trap Tagger, Conservation AI, MLWIC2, FASTCAT, etc? Would be awesome to have a comparisons sheet of all the camera trap image processing software since I feel like there are a lot popping up recently! There is this one guide but it doesn’t have all the software options available (just 3 I believe) - https://ai-camtraps.netlify.app/index.html. I don’t mean to sound negative about a new software, it’s obviously great that there are different options; just that it would be good to have pro’s/con’s of each for people to see which is best for their use case! 🙂

👍 Talia Speaker, Peter Griggs, Alexander Robillard

Talia Speaker (talia.speaker@wildlabs.net)

2022-06-06 14:44:48

This camera trap solution comparison doc by @Petar Gyurov is a great start on that Carly! https://www.notion.so/Camera-Trap-Pipeline-Solution-Comparison-2eac80825c4941b0b2b5fad3daea1cc3

Petar's Notion on Notion

Camera Trap Pipeline - Solution Comparison

This page compares known camera trap solutions that aid with detection and species classification. The table below is a summary of key features that would aid the typical workflow for an ecologist.

Original URL: https://www.notion.so/Camera-Trap-Pipeline-Solution-Comparison-2eac80825c4941b0b2b5fad3daea1cc3

👍 Carly Batist, Stephanie O'Donnell

😊 Petar Gyurov

Talia Speaker (talia.speaker@wildlabs.net)

2022-06-06 14:48:26

Also Dan's helpful running resource for overviews of solutions: https://agentmorris.github.io/camera-trap-ml-survey/

Camera Trap ML Survey

Overview

Everything I know about machine learning and camera traps.

Original URL: https://agentmorris.github.io/camera-trap-ml-survey/

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-06 14:51:02

Oh awesome hadn’t heard of these yet, thanks for sharing! Will def add these to the next update of the Conservation Tech Directory. I think the only ones I can think of that aren’t on Peter’s list are TrapTagger, FASTCAT cloud, ConservationAI, and obviously this new one Peter mentioned (Zendo).

🙌 Talia Speaker, Peter Griggs, Rita Pucci

Peter Griggs (peter@deepai.org)

2022-06-06 15:22:17

Oh these are great lists! Thanks for sharing.

The main difference between our platform, Zendo, and other options is that you can train it on your own camera imagery really quickly, and get a highly accurate model fit to your dataset and species of interest.

That model you've trained is hosted behind the scenes, and accessible for use at scale within the platform, or via an API.

You can then collaborate on and review new images within the platform, or export everything to a CSV file.

👍 Talia Speaker, Carly Batist

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-08 03:57:46

https://www.straitstimes.com/singapore/environment/new-app-launched-to-help-in-fight-against-illegal-trade-of-shark-and-ray-fins

The Straits Times

New app launched to help in fight against illegal trade of shark and ray fins

SINGAPORE - A new mobile application launched yesterday (June 7) will leverage artificial intelligence (AI) to help enforcement officers streamline their fight against illegally traded shark and ray fins coming through Singapore. Read more at <a href="http://straitstimes.com">straitstimes.com</a>.

Original URL: https://www.straitstimes.com/singapore/environment/new-app-launched-to-help-in-fight-against-illegal-trade-of-shark-and-ray-fins

👏 Rita Pucci, Jason Holmberg (Wild Me), Alexander Robillard

🎉 Jon Van Oast, Dan Morris, Alexander Robillard

Thiago Bicudo (bicudotks@gmail.com)

2022-06-11 13:12:05

https://twitter.com/mclduk/status/1535166475518058497?t=RtcNBPuUwNCyjTMPD4XjRw&s=08

Looking for a postdoc opportunity using AI, sound and image recognition for nature conservation? We're about to advertise 2 x 2-year postdocs here in the beautiful Dutch city of Leiden at @Naturalis_Sci. Feel free to message me/email me if you want to know more

twitter

Dan Stowell (https://twitter.com/mclduk/status/1535166475518058497)

Looking for a postdoc opportunity using AI, sound and image recognition for nature conservation? We're about to advertise 2 x 2-year postdocs here in the beautiful Dutch city of Leiden at <a href="https://twitter.com/Naturalis_Sci">@Naturalis_Sci</a>. Feel free to message me/email me if you want to know more

Original URL: https://twitter.com/mclduk/status/1535166475518058497?t=RtcNBPuUwNCyjTMPD4XjRw&s=08

❤️ Thiago Bicudo, Alexander Robillard, Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Rita Pucci, Subhransu Maji

👍 Carly Batist, Alexander Robillard, Eelke

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-11 14:50:28

ha - great minds think alike, I posted it in the jobs channel today too😂

Thiago Bicudo (bicudotks@gmail.com)

2022-06-11 14:55:41

Thanks Carly, I'm new here and didn't realize that there is a job channel. Thank you so much 😘

😲 Rita Pucci

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-11 14:56:45

Yes definitely check out the other channels! The ‘jobs’, ‘upcomingevents’ and ‘newpapers’ are particularly helpful for me in staying up to date with what’s going on in the field 🙂

❤️ Thiago Bicudo

Devis Tuia (devis.tuia@epfl.ch)

2022-06-12 13:32:01

I wish! I always enjoyed visiting Naturalis during the Dutch part of my life.. say hi to Rutger Vos, if he is still there!

Rita Pucci (rita.pucci85@gmail.com)

2022-06-14 03:21:52

@Thiago Bicudo, which job channel ??

Rita Pucci (rita.pucci85@gmail.com)

2022-06-14 04:49:49

ah, thanks, it didn't pop in the channels list before!

👍 Devis Tuia

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-14 06:03:08

not all of them necessarily appear at first - when I first joined I had the same issue! I had to go to the “+” button next to channels --> ‘browse channels’ and then I saw a bunch of other ones I joined

👍 Rita Pucci

Devis Tuia (devis.tuia@epfl.ch)

2022-06-15 02:50:06

exactly… it’s a discovery process 😛

😂 Carly Batist, Sara Beery, Rita Pucci

Patrizia Paci (pp4649@open.ac.uk)

2022-06-15 11:18:26

Hi everyone, I am new in the channel and just wanted to greet all of you! My name is Patrizia and I am an associate lecturer in Evolutionary Biology and researcher in Animal-Computer Interaction at the Open University in the UK. I developed an interest in Machine Learning. Thus, I very recently visited the Machine Vision group at TU Delft in the Netherlands to work on training algorithms and annotations to recognise animal activities on videos. I am interested in ML applications to bioacoustics too. I would like to continue my research in ML for animal activity recognition, integrating behavioural ecology and ML. Happy to be part of your community and share with you knowledge, insight, challenges, etc. Special thanks to Silvia Zuffi who invited me to this channel :)

🎉 Stefan Schneider, Sara Beery, Oisin Mac Aodha, Alexander Robillard, Stephanie O'Donnell, Bilgenur Baloglu, Rita Pucci

👋 Benjamin Kellenberger, Jon Van Oast, Declan, Omiros Pantazis

Subhransu Maji (smaji@cs.umass.edu)

2022-06-15 11:23:55

Welcome! I’m in Delft too on a sabbatical!

Patrizia Paci (pp4649@open.ac.uk)

2022-06-15 12:25:48

Hello Subhransu, I loved Delft!

👍 Subhransu Maji

Bilgenur Baloglu (bilgenurb@gmail.com)

2022-06-15 13:10:30

Hi everyone! I am also new in the channel and really wanted to join after hearing about it at Sara Beery's talk. This is Bilgenur, I am a lecturer at USC (teaching machine learning and genomic data analysis) and a bioinformatics scientist at Thermo Fisher. I am interested in starting my company for eDNA based monitoring, focusing on water (freshwater and ocean). My PhD work was on biomonitoring of Singapore's aquatic ecosystems using DNA sequencing technologies and my postdoc work took me all the way to sub-arctic Canada! Really happy to join you all, and look forward to future conversations and collaborations.

👋 Sara Beery, Declan, Omiros Pantazis, Déva Sou, Patrizia Paci, Rita Pucci

✌️ Burak Ekim

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-15 13:31:23

Assuming you know about NatureMetrics? If you wanted to do eDNA based aquatic monitoring they’re doing that already!

Bilgenur Baloglu (bilgenurb@gmail.com)

2022-06-15 13:41:18

Yes! Very aware of them since the beginning of NatureMetrics :) The niche market or the customer base I am thinking about is a little different than theirs.

👍 Carly Batist

Ando Shah (ando@berkeley.edu)

2022-06-15 23:43:01

Hi Bilgenur! Welcome! A project Im working on involves using eDNA within a very large MPA to assess biodiversity, and working with NatureMetrics is problematic for a few reasons. We would welcome new players in this field, and in general our team would be happy to chat more! If this is of interest, please DM me and we can chat

❤️ Bilgenur Baloglu, Sara Beery

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-16 07:04:07

I’d also recommend posting to the WILDLABS eDNA group! Lots of people working on really cool stuff there. https://wildlabs.net/groups/edna-genomics

wildlabs.net

eDNA & Genomics | WILDLABS

eDNA is a molecular conservation tech tool that can be used to detect species presence in samples taken directly from the environment. To date, eDNA has been used for species detection, biomass estimation, diet analysis, reconstruction of past flora and fauna, and wildlife disease detection. Still a relatively new area of conservation tech, eDNA is in a phase of rapid innovation and growth, with improved ease of use and more accessibility allowing this technology to find new uses in the field and lab.

Original URL: https://wildlabs.net/groups/edna-genomics

👍 Bilgenur Baloglu

Bilgenur Baloglu (bilgenurb@gmail.com)

2022-06-16 14:44:22

Hi Ando! Thank you so much for your response. I would be happy to hear more about your project!

Bilgenur Baloglu (bilgenurb@gmail.com)

2022-06-16 14:46:15

Will do, thanks for bringing this group to my attention, Carly!

👍 Carly Batist

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-23 19:08:07

Just came across this and figured it may of interest as well - https://www.ednacollab.org/ (they also have a grants program)

❤️ Bilgenur Baloglu

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2022-06-17 01:20:13

EAAMO Doctoral Consortium applications out and due July 8! A fantastic second-year conference growing out of a wonderful and supportive research community (Mechanism Design for Social Good)

EAAMO is October 6–9 in Washington DC

The second ACM conference on Equity and Access in Algorithms, Mechanisms, and Optimization (EAAMO'22) aims to highlight work where techniques from algorithms, optimization, and mechanism design, along with insights from the social sciences and humanistic studies, can help improve equity and access to opportunity for historically disadvantaged and underserved communities.

https://eaamo.org/doctoral_consortium/

👍 Ayan Mukhopadhyay

Thijs (thijs@q42.nl)

2022-06-17 05:53:24

Wednesday was an exciting (and a bit terrifying) day. Sorry, I just feel the need to share this with this group of beautiful people 🤩

I gave a TEDx talk! It's about how I've been using my tech skills for good / conservation.

My title is: From code to conservation: a nerd's search for meaning

Hopefully the recording will come online in a couple of weeks so I can also share the video here (if that's okay) 🙏

🙌 Stephanie O'Donnell, Ștefan Istrate, Carly Batist, Catherine Villeneuve, Omiros Pantazis, Yihang She, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Kakani Katija, Declan, Bilgenur Baloglu, Suzanne Stathatos, Mark Jordan, Caleb Robinson, nyakundi lamech, Déva Sou, Catherine Wang, Rita Pucci, Crystal Huang

💕 Jon Van Oast

👏 Rita Pucci

👍 Jan Kees

Thijs (thijs@q42.nl)

2022-06-17 06:04:03

I talk about the effect of climate change on the African rainforest. Especially the impact it has on forest elephants and why that is causing more Human-Elephant-Conflicts. Obviously not in much detail, because I only had 14 minutes to talk 🙂

I can tell you that this was by far one of the coolest but also scariest talks I've given so far. 😇

💚 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Caleb Robinson, Rita Pucci, Déva Sou

Caleb Robinson (calebrob6@gmail.com)

2022-06-18 09:50:40

Congrats @Thijs!! I can't wait to watch!

Devis Tuia (devis.tuia@epfl.ch)

2022-06-18 20:10:48

congrats @Thijs! I gave one last year in Martigny and I agree 100%: super scary but also very rewarding

👍 Déva Sou

Thijs (thijs@q42.nl)

2022-06-19 07:46:01

Thanks Caleb!

Thijs (thijs@q42.nl)

2022-06-19 07:46:28

Cool @Devis Tuia I'll try to look up your talk on YouTube 👍

Thijs (thijs@q42.nl)

2022-06-19 07:48:33

Ah it's in french, I don't understand that 😇

Devis Tuia (devis.tuia@epfl.ch)

2022-06-20 11:33:51

😄

Colin Donihue (colindonihue@gmail.com)

2022-07-19 10:00:36

Hey @Thijs! I’d love to see your talk. I looked for it online but couldn’t find it, do you know if it’s been posted? Thanks!

Thijs (thijs@q42.nl)

2022-08-01 05:25:50

@Colin Donihue The talk is not online yet, this usually takes 4-8 weeks for editing and reviewing by TED. Once it's online I'll post it here 👍

👍 Colin Donihue

Devis Tuia (devis.tuia@epfl.ch)

2022-06-18 20:11:53

In case you are around at CVPR2022, please let me know! It would be super cool to have a AIforConservation meetup (for example for lunch once)! you can find me for sure tomorrow (Sunday) at the Earthvision workshop: https://www.grss-ieee.org/events/earthvision-2022/

GRSS-IEEE

EARTHVISION 2022 - GRSS-IEEE

EARTHVISION 2022 June 19th, New Orleans, Louisiana – hybrid/virtual in conjuction with the Computer Vision and Pattern Recognition (CVPR) 2022 Conference Aims and Scope Important Dates People Challenge Sponsors Submission Program CVPR 2022 Aims and Scope Earth Observation (EO)/Remote Sensing is an ever-growing field of investigation where computer vision, machine learning, and signal/image processing meet. The […]

Est. reading time

14 minutes

Original URL: https://www.grss-ieee.org/events/earthvision-2022/

🙌 Hannah Kerner, Sara Beery, Oisin Mac Aodha, Kakani Katija, Sophia Abraham, Stephanie O'Donnell, Rita Pucci, Helena Russello, Suzanne Stathatos, Kasirat, Dhruv Sheth

Devis Tuia (devis.tuia@epfl.ch)

2022-06-18 20:12:49

(room 219, from 8:30 on)

Frederic (frederic@apic.ai)

2022-06-19 20:35:02

Hi @Devis Tuia i am at CVPR, too. As probably a few others from this slack channel.

A few are probably attending the CV4Animals workshop tomorrow. How about we get something for dinner afterwards with anyone who is interested?

❤️ Sophia Abraham, Helena Russello

Devis Tuia (devis.tuia@epfl.ch)

2022-06-20 11:33:38

Cool! Let’s try lunch first at the cv4animals break? Like we can meet at the end of the session just outside of the cv4animals room? 11h45?

Daniel Davila (daniel.davila@kitware.com)

2022-06-20 12:18:43

My colleague, Matt Dawkins, will be giving a talk at the CV4Animals workshop in a little while here. We have a lot of team members from Kitware milling about the conf and are at booth #1522 in the exhibition, if anyone wants to meet up! Have a good week yall.

Devis Tuia (devis.tuia@epfl.ch)

2022-06-20 16:05:12

@Frederic let’s discuss at the end of the sessions of cv4animals. I am in the room btw 😉

Frederic (frederic@apic.ai)

2022-06-20 16:07:22

sorry did not see your message. Lets meet afterwards, to discuss it.

👍 Devis Tuia

Rita Pucci (rita.pucci85@gmail.com)

2022-06-21 03:55:39

Yesterday's Workshop was fantastic!! 😄 I met virtually lots of people, thanks for visiting my poster!

🙌 Silvia Zuffi, Sara Beery, Jason Holmberg (Wild Me), Alexander Robillard, Frederic, Helena Russello, Sophia Abraham

🎉 Jon Van Oast, Sophia Abraham

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-21 20:26:24

New paper on transfer learning with passive acoustic data out, led by@Emmanuel Dufourq! Happy to have played a small co-author part but he deserves the recognition for all the programming heavy lifting! https://www.sciencedirect.com/science/article/pii/S1574954122001388

sciencedirect.com

Passive Acoustic Monitoring of Animal Populations with Transfer Learning

Progress in deep learning, more specifically in using convolutional neural networks (CNNs) for the creation of classification models, has been tremend…

Original URL: https://www.sciencedirect.com/science/article/pii/S1574954122001388

🎉 Jon Van Oast, Declan, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Suzanne Stathatos, Dan Morris, Marcus Lapeyrolerie, Juan Sebastián Cañas Silva

👏 Rita Pucci

👀 Alexander Robillard

Sara Beery (sara.beery@gmail.com)

2022-06-22 16:18:20

Auto Arborist has been released!! This fine-grained multiview dataset contains over 2 million trees belonging to over 300 genus-level categories in 23 cities across the US and Canada built to foster the development of robust methods for large-scale urban forest monitoring

https://twitter.com/sarameghanbeery/status/1539703332100521984

twitter

Sara Beery (https://twitter.com/sarameghanbeery/status/1539703332100521984)

#AutoArborist has been released!! This fine-grained multiview dataset contains over 2 million trees belonging to over 300 genus-level categories in 23 cities across the US and Canada built to foster the development of robust methods for large-scale urban forest monitoring 🌲🌲🌲 <https://twitter.com/GoogleAI/status/1539680204594982912>

Original URL: https://twitter.com/sarameghanbeery/status/1539703332100521984

twitter

Google AI (https://twitter.com/GoogleAI/status/1539680204594982912)

Introducing the Auto Arborist Dataset, a multiview urban tree classification dataset that consists of ~2.6M trees and >320 genera, which can aid in the development of models for urban forest monitoring. Check it out at → <https://goo.gle/3QytK5w>

🎉 Carly Batist, Justin Kay, Declan, gvanhorn, Nikhil Vytla (he/him), Arjun Subramonian (they/them), Stefan Schneider, Daniel Grzenda, Ando Shah, Sachith Seneviratne, Suzanne Stathatos, Yihang She, Emilio Luz-Ricca, Alan Papalia, Vivek Mishra, Chris Yeh

😎 Jon Van Oast, Frederic, Ben Weinstein

🌳 Stefan Schneider, Daniel Grzenda, Frederic, Dan Morris, Oisin Mac Aodha, Catherine Villeneuve, Alan Papalia

🌴 Stefan Schneider, Daniel Grzenda, Helena Russello, Catherine Villeneuve, Alan Papalia, Riccardo de Lutio

🌲 Stefan Schneider, Daniel Grzenda, Frederic, Robin Zbinden, Catherine Villeneuve, Alan Papalia

🙌 Carl Boettiger

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2022-06-23 11:14:12

EarthRanger's Conservation Technology Award is providing two grants — $15K each — to organizations that are developing tech for wildlife. Applications due August 31 https://www.earthranger.com/conservation-tech-award

@Sara Beery @Jake Wall @Tanya Birch @Stephanie O'Donnell @Jes Lefcourt

👀 Stephanie O'Donnell, Jason Holmberg (Wild Me), Declan, Alexander Robillard

🎉 Carly Batist, Fadel Seydou

aruna (arunas@mit.edu)

2022-06-23 11:17:51

Folks, does anyone have connections to individuals in the following orgs. Please let me know if you do, so I can DM you for a small favour.

350.org
AfD
APANO
Citizens' Climate Lobby
Climate Action Rhode Island
Climate Justice Alliance
Climate Pop Up
Creative Action Network
EARTHJUSTICE
EDF
Environmental Defence Canada
Environmental Defense Fund
ENVIRONMENTAL WORKING GROUP
Extinction Rebellion Delaware
Extinction Rebellion Malvern
Extinction Rebellion NL
Extinction Rebellion Québec
Extinction Rebellion Skåne
Extinction Rebellion UK
Fridays for Future
Fridays For Future Toronto
Friends of the Earth
Friends of the Earth Europe
Friends of the Earth Malta
Friends of the Earth Scotland
Greenpeace UK
Greenpeace USA
HEALTHY ENVIRONMENT ALLIANCE OF UTAH
Israel Action Network
LPAC Action Network
Michigan Climate Action Network
MTÜ Loodusvõlu
Natural Resources Defense Council
NRDC
RAINFOREST ACTION NETWORK
Rana
Sunrise
the Environmental Working Group
Toronto Climate Action Network
Union of Concerned Scientists
USA Justice Alliance
World Wildlife Fund

350

350.org: A global campaign to confront the climate crisis

We are standing up to the fossil fuel industry to stop all new coal, oil and gas projects and build clean energy future for all.

Original URL: http://350.org

👍 Helena Russello, Sara Beery

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-23 15:12:02

Specific departments or areas? Some of those are very large & multi-faceted so I would have a better idea if I know someone with a bit more context/detail

Thijs (thijs@q42.nl)

2022-06-23 15:16:46

Yeah, I do know some people at some of these organisations. But it's quite a long list. Maybe you can elaborate a bit on your intentions?

aruna (arunas@mit.edu)

2022-06-23 15:17:13

Thanks @Carly Batist and @Thijs. I am looking for someone dealing with Facebook advertisements at these orgs.

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-23 15:17:48

so someone in the marketing or comms departments then?

Thijs (thijs@q42.nl)

2022-06-23 15:18:07

And lets assume you find people at these departments, what's your question?

aruna (arunas@mit.edu)

2022-06-23 15:18:09

That sounds perfect! 🙂

aruna (arunas@mit.edu)

2022-06-23 15:18:22

I am interested in talking to them about how they advertise on Facebook. 🙂

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-23 15:41:39

Do you work for an org that is trying to do this? Just curious how you came up with this list

aruna (arunas@mit.edu)

2022-06-23 15:45:04

I am a PhD student at MIT, looking at how environmental orgs communicate about CC.

aruna (arunas@mit.edu)

2022-06-23 15:47:11

This list is from my initial survey of adv orgs. Happy to talk to other adv orgs who advertise/post on FB too!

Sara Beery (sara.beery@gmail.com)

2022-06-25 06:26:00

NVIDIA hardware grant program is back: https://mynvidia.force.com/HardwareGrant/s/Application

❤️ Sophia Abraham, Suzanne Stathatos, Jan Kees, Jason Parham, Carly Batist, Frederic, Carl Boettiger, David

🎉 Jon Van Oast

Ankur Kalra (ankur@hoplabs.com)

2022-06-27 14:25:56

Hi folks! New to the Slack, but met @Sara Beery at CVPR 22 and learned of this community. Excited to get to know folks and their projects! I'm at Hop Labs, which is an ML/CV consulting firm -- we do a lot of work for corporate clients, but we have a pro-bono program for non-profit research teams: https://www.hoplabs.com/pro-bono-work. Please do not hesitate to DM or email if your project could use some free help. Thanks!

❤️ Sara Beery, Stephanie O'Donnell, Jason Holmberg (Wild Me), Suzanne Stathatos, Carly Batist, Talia Speaker, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

👋 gvanhorn, Jason Holmberg (Wild Me), Jon Van Oast, Dan Morris, David

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-30 12:04:41

Climate Change AI Summer School - call for mentors
Description: https://docs.google.com/document/d/1C1bfCesbAqMNmfEJzn7QETv2pNYjks3ZQl9CUqXO3o0/edit
Application form: https://docs.google.com/forms/d/1y3DoURFesd8k1NQN1mZSOqYgmbJxwpN4syIFfS_bpDs/viewform?edit_requested=true

👍 Sara Beery, Daniel Spokoyny, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Emily Lines

:thumbsup_all: Frederic Fol Leymarie

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-06-30 12:06:27

disregard my emails highlighting…(just noticed that🤦‍♀️

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-07-06 11:48:24

🗣️🔊🎶 For those interested in AI for bioacoustics/ecoacoustics applications 🗣️ 🔊🎶

The Bioacoustics Stack Exchange has reached the Beta stage and is LIVE! It is also in a critical stage, and we need your help to keep it alive.
We have only 2 weeks to prove we have the community to support the site so please help as you can! If you want more information on how to help or get started with Stack Exchange, check out this helpful Youtube tutorial made by Selene Fregosi.

There have been LOTS of questions on AI/ML for bioacoustics recently so would be great to have input/users from the tech side contributing as well!

How can you help?

JOIN: https://bioacoustics.stackexchange.com/

VISIT: Make it a habit to check in on the site regularly. We need >>500 visits per day!

VOTE: We need to show participation! Vote on all good Q/A

ASK: Have a new bioacoustics-related question? What about your older questions? You can even ask/answer your own question! These questions will be archived for future use, so ALL good questions are welcome! We need >> 10 questions per day!

ANSWER: Have an answer to one of the questions, or additional information? Share your expertise where you can! Is there a question you are often asked? You can pair up with someone to ask/answer it again to be archived! If someone asks you a question-- tell them you will answer it on the Stack Exchange! We need ~2.5 answers per question.

INVITE & SHARE: Invite others to join! Share on social media, with your collaborators, etc.!

bioacoustics.stackexchange.com

Bioacoustics Stack Exchange

Q&A for people interested in the studies of non-human animal sounds and the impacts of sounds on animals

Original URL: https://bioacoustics.stackexchange.com/

❤️ Sara Beery, Declan, Alexander Robillard, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Jan Kees

😎 Jon Van Oast

🔊 Stefan Schneider, Elijah Cole (Deactivated)

🙌 Antón Álvarez

Yves Bas (yves.bas@gmail.com)

2022-07-08 09:33:11

Hello, a TREE paper on new techs and insect monitoring with an important focus on AI, computer vision and acoustics: https://www.sciencedirect.com/science/article/pii/S0169534722001343

sciencedirect.com

Emerging technologies revolutionise insect ecology and monitoring

Insects are the most diverse group of animals on Earth, but their small size and high diversity have always made them challenging to study. Recent tec…

Original URL: https://www.sciencedirect.com/science/article/pii/S0169534722001343

🔊 Oisin Mac Aodha, Alan Papalia, Suzanne Stathatos, Jason Holmberg (Wild Me), Sara Beery, Belen Saavedra

🐜 Carly Batist, Stefan Schneider, Stephanie O'Donnell, Jason Holmberg (Wild Me), Déva Sou, Tarun

👍 Michael Bunsen

Sara Beery (sara.beery@gmail.com)

2022-07-11 20:17:31

NSF funding opportunity: https://beta.nsf.gov/funding/opportunities/partnership-advance-conservation-science-and-practice-pacsp.

Beta site for NSF - National Science Foundation

Partnership to Advance Conservation Science and Practice (PACSP)

Original URL: https://beta.nsf.gov/funding/opportunities/partnership-advance-conservation-science-and-practice-pacsp

💯 Carly Batist, Jason Holmberg (Wild Me), Subhransu Maji, Stephanie O'Donnell, Ben Weinstein, Olivier Gimenez

George Darrah (george.darrah@systemiq.earth)

2022-07-14 10:39:58

Hey everyone - recently posted my take on the role of biodiversity monitoring tech in scaling up private sector investment into nature... would be super interested in expert feedback! thanks 🙏 https://www.linkedin.com/posts/georgedarrah_we-need-to-bring-biodiversity-to-the-bloo[…]e59v?utm_source=linkedin_share&utm_medium=member_desktop_web

linkedin.com

George Darrah on LinkedIn: We need to bring biodiversity to the Bloomberg terminal — Systemiq | 12 comments

My thoughts on why scaling biodiversity monitoring technology will play a key role in unlocking $$$ of private sector finance for biodiversity. We need... 12 comments on LinkedIn

Original URL: https://www.linkedin.com/posts/georgedarrah_we-need-to-bring-biodiversity-to-the-bloomberg-activity-6950070143146209280-e59v?utm_source=linkedin_share&utm_medium=member_desktop_web

👍 Stephanie O'Donnell, Sara Beery, Suzanne Stathatos, Alexander Robillard, Talia Speaker, Carly Batist, nyakundi lamech

😎 Jon Van Oast, Alexander Robillard

:first_place_medal: Jan Kees

Jan Kees (jankees.schakel@sensingclues.org)

2022-07-15 02:27:57

thx @George Darrah, very interesting read!

Frederic Fol Leymarie (ffl@dynaikon.com)

2022-07-15 07:26:47

Monitoring: yes; protecting: YES; rewilding: even BETTER (leaving these to the private sector .. is not enough).

George Darrah (george.darrah@systemiq.earth)

2022-07-18 09:55:25

Agreed - but my sense is that without private sector funding and significant change to taxation we will not turn the tide on biodiversity losses

Frederic Fol Leymarie (ffl@dynaikon.com)

2022-07-18 12:44:16

So lobbying both industry and government is needed, and keep raising awareness in the public and even in the academic spheres.

nyakundi lamech (lamechongondi88@gmail.com)

2022-07-15 15:28:08

Hi Everyone, my name is Lamech, and I am relatively new to this forum. I recently completed my undergraduate in statistics and have been since working on a startup I helped co-found. Our pioneer tool, E-Savior uses computer vision algorithms to reduce human-animal conflict through monitoring animal movements. We are currently using it to monitor elephants and hope to scale it up to other species in the future. My broad research interests are computational science in conservation. Next week I will be attending the APAC conference in Kigali, Rwanda, and would be happy to meet folks from this forum for a chat to learn more about all the exciting projects that you are all working on. I am also open to exciting roles in this field, should you have any! I can’t wait to meet those who will be attending APAC. Please email me on lamechongondi88@students.uonbi.ac.ke

🐘 Jason Parham, Subhransu Maji, Fadel Seydou, Sara Beery, Malte Pedersen, nyakundi lamech, Alexander Robillard

👍 Benjamin Kellenberger, nyakundi lamech, Alexander Robillard

Sara Beery (sara.beery@gmail.com)

2022-07-18 13:01:17

Hey everyone! Bit of personal news from your friendly neighborhood AI for Conservation founder - I've accepted a faculty position at MIT, starting fall 2023! I'll spend the next year at Google working on multimodal/multiview urban forest monitoring (see Auto Arborist), but I'll be hiring PhD students and a PostDoc this fall to start with me next year! If you're looking for an academic position at the intersection of CV/ML, biodiversity, and the environment please reach out!! I'm looking to hire students from diverse backgrounds to work collaboratively on interdisciplinary research projects.

twitter

Sara Beery (https://twitter.com/sarameghanbeery/status/1548383002496929792)

We're moving to Boston!!! I'll be starting as an Assistant Professor at <a href="https://twitter.com/MITEECS">@MITEECS</a> in Fall 2023, where my lab will focus on computer vision, biodiversity, and the environment🌲🐘🦅:ladybug:🐟 Until then I'll be at <a href="https://twitter.com/GoogleAI">@GoogleAI</a> working on monitoring urban forests with #AutoArborist 🌳🌲🌴 <https://pbs.twimg.com/media/FXz2ImxUcAEtTeR.jpg>

Original URL: https://twitter.com/sarameghanbeery/status/1548383002496929792

Google AI Blog

Mapping Urban Trees Across North America with the Auto Arborist Dataset

Posted by Sara Beery, Student Researcher, and Jonathan Huang, Research Scientist, Google Research, Perception Team Over four billion peop...

Original URL: https://ai.googleblog.com/2022/06/mapping-urban-trees-across-north.html

🎉 Avi Sundaresan, Peter Bull, Jason Holmberg (Wild Me), Talia Speaker, aruna, Mark Goldwater, Declan, Daniel Grzenda, Suhail Alnahari, Suzanne Stathatos, Jon Van Oast, Alan Papalia, Daniel Spokoyny, Fagner Cunha, Ted Schmitt, Lucia Gordon, Justin Kay, Felipe Parodi, Carly Batist, Amrita Gupta, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Leonardo Viotti, Alexander Robillard, Matt Weldy, Lukas Picek, Elijah Cole (Deactivated), Bilgenur Baloglu, Lauren E. Gillespie, Dan Morris, Belen Saavedra, Stefan Schneider, Mitch Fennell, Blair Costelloe, Hemal Naik, Liyuan Zhu, Riccardo de Lutio, Yihang She, Diego Marcos, nyakundi lamech, Agnethe Seim Olsen, Cameron Trotter, Colin Donihue, Déva Sou, Oliver Broadrick, Dongmin (Dennis) Kim, Ando Shah, Timm Haucke, Alex Borowicz, Georgia Atkinson, Dhruv Sheth

👏 Malte Pedersen, Peter Bull, Jason Holmberg (Wild Me), Oisin Mac Aodha, aruna, Mark Goldwater, Daniel Grzenda, Jon Van Oast, Alan Papalia, Subhransu Maji, Emilio Luz-Ricca, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Alexander Robillard, Lukas Picek, Lauren E. Gillespie, Belen Saavedra, Stefan Schneider, Kalyan Nadimpalli, Liyuan Zhu, Omiros Pantazis, nyakundi lamech, Atriya Sen, Thijs, Dongmin (Dennis) Kim, Timm Haucke, Dhruv Sheth, Chris Yeh, Antón Álvarez, Lloyd Hughes, Andrew Schulz, David

❤️ Suzanne Stathatos, Ben Weinstein, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Alexander Robillard, Lukas Picek, Eddie Zhang, Belen Saavedra, Stefan Schneider, Mark Goldwater, Liyuan Zhu, nyakundi lamech, Dongmin (Dennis) Kim, Dhruv Sheth, Andrew Schulz, David

🙌 Sean P. Rogers, Dhruv Sheth, Andrew Schulz

Peter Bull (peter@drivendata.org)

2022-07-18 13:02:49

Amazing, congrats!! 🎊

❤️ Sara Beery

nyakundi lamech (lamechongondi88@gmail.com)

2022-07-19 07:10:13

Great !👏

❤️ Sara Beery

Déva Sou (soudeva974@gmail.com)

2022-07-19 12:30:14

Congratulations!

❤️ Sara Beery

Oliver Broadrick (obroadrick@gwmail.gwu.edu)

2022-07-19 19:48:48

Hi all, I'm an MS student studying CV at GW, advised by Prof Robert Pless. I love nature and conservation (fly fishing, hiking...), and I was recently introduced to @Sara Beery’s amazing work! That led me to finding this slack... so, hi!

Among other things, I'm currently playing with pictures of trout. If you have trout pictures, send them my way! 🙂

🐟 Sara Beery, Suzanne Stathatos, Kakani Katija, Jason Holmberg (Wild Me), Declan, Déva Sou, Alexander Robillard, Belen Saavedra

👋 gvanhorn

Oliver Broadrick (obroadrick@gwmail.gwu.edu)

2022-07-19 19:52:26

Joined, thanks!

Sean P. Rogers (sean.rogers@uvm.edu)

2022-07-20 11:17:33

Hi Everybody, @Dan Morris introduced me to this great community. I'm a PhD student in Complex Systems and Data Science at the University of Vermont studying applications of ML and NLP to identifying wildlife exploitation on social media. I hold a general interest in conservation technology and environmental security issues. I'm looking forward to learning with and interacting with you all. 🐢

👋 gvanhorn, Declan, Sara Beery, Dan Morris, Yihang She, Oliver Broadrick, Carly Batist, Suzanne Stathatos, Alex Borowicz, Belen Saavedra, Benjamin Kellenberger

Ryan Feng (ryan.feng16@gmail.com)

2022-07-20 12:36:45

Hi all, just learned about this from coming across Sara’s website! I'm a robotics M.S. student at UMich. I know a lot of y'all will be from more CV/DS backgrounds, but conservation is something I feel pretty strongly about and I’m looking forward to learning more about where robotics can play a part!

🤖 Sara Beery, Declan, Carly Batist, Oliver Broadrick, Suzanne Stathatos, Jason Holmberg (Wild Me), Alan Papalia, nyakundi lamech

👋 Benjamin Kellenberger, Yihang She, Jaanak

Muskan Sachdeva (muskan@meta-lynx.com)

2022-07-21 07:57:01

Hi Everyone,
I am an MBA student working with Metalynx a data curation start-up based out of London. Our platform helps users visualise and curate image data sets to uncover biases and corrupted data, and efficiently choose which images to label. I found this slack channel following @Sarah Zimmer Beery’s amazing work in conservation. Additionally @Dan Morris’s github page and @Petar Gyurov’s notion page have immensely helpful for me to understand AI in conservation landscape.

I would be extremely grateful if could have a short 15-20 min chat with anyone working with computer vision to better understand the workflow while using CV in conservation and issues encountered in development using large image data sets.

👋 Sara Beery, Oliver Broadrick, Jason Holmberg (Wild Me), nyakundi lamech, Jaanak

Jaagat P. (jaagatp05@gmail.com)

2022-07-22 14:00:30

Hi All! I hope this finds you well. My name is Jaagat and I am a rising junior in high school (probably didn't expect to see one here 😅) who's truly passionate about leveraging AI for social good/conservation and would be more than grateful to connect with you all. While I still have a lot to learn, I can't wait to see the potential that AI, robotics, and ML hold for our future! It's a true privilege to join such a community!

👍 Jason Holmberg (Wild Me), Justin Kay

👋 Declan, Jon Van Oast, Belen Saavedra, Oliver Broadrick, Sara Beery, Andy Viet Huynh, Carly Batist, nyakundi lamech, Eddie Zhang, Carl Boettiger, Jaanak

👶 Ryan Feng

🚀 Carl Boettiger

Marconi Campos (marconi@rfcx.org)

2022-07-22 16:26:42

Hi everyone, I am new here on the channel and just wanted to greet you all! My name is Marconi Campos; Im a tropical ecologist committed to improving wildlife detection, monitoring, and conservation. Im currently the chief scientist in Rainforest Connection (https://rfcx.org), where we have been working hard combining acoustic monitoring, AI, and occupancy models to understand how fauna is responding to both natural and human disturbances all around the world. I`m interested in ML applications to bioacoustics but also eager to learn and combine different approaches to improve biodiversity monitoring. Happy to be here and excited to learn a lot and share some ideas ; )

🌿 Suzanne Stathatos, Oliver Broadrick, Sara Beery, Declan, Andy Viet Huynh, Carly Batist, Jaagat P., nyakundi lamech, Jaanak, Julia Marisa Sekula

😎 Jon Van Oast

🎉 Dan Morris, Justin Kay, Eddie Zhang

👋 Carl Boettiger

Jaanak (jaanak007@gmail.com)

2022-07-22 20:56:37

Hi all! My name is Jaanak Prashar, and I am a rising junior in high school (definitely not Jaagat's fraternal twin). I hold a strong passion for computational biology, molecular genetics, and combining artificial intelligence with genetics and health equity. I also have a passion for philosophy, particularly the philosophy of the RNA world in genetics! It is a great pleasure and a privilege to be here, and I am excited to learn a lot!

😁 Sara Beery, Justin Kay, nyakundi lamech, Suzanne Stathatos

👬 Sara Beery, Jaagat P., Suzanne Stathatos, Jaanak

👋 Peter van Lunteren, Mark Goldwater, Eddie Zhang, Déva Sou

🚀 Carl Boettiger

Jaanak (jaanak007@gmail.com)

2022-07-25 14:54:32

Hi all,

I hope this message finds you well. My twin brother Jaagat and I are currently looking for any research opportunities in the fields of machine learning and data science and/or their practical applications (i.e. genomics, ecology, etc.). We would be grateful for any research opportunities available (unpaid) either during the summer, throughout the year, or even next summer. We both would also be more than happy to share our resumes as well as past research I have conducted and presented at a conference. Have a lovely day!

Best,

The twins 👬

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-07-25 14:56:12

Where are you based geographically now? That would help with identifying projects

Jaanak (jaanak007@gmail.com)

2022-07-25 14:57:18

We are both located in Texas, and we are currently seeking any remote opportunities!

Mark Goldwater (mgoldwater@whoi.edu)

2022-07-25 15:00:21

Try your best to not settle for unpaid internships. You deserve money for your efforts!

💯 Declan, Ștefan Istrate, Alexander Robillard, Andy Viet Huynh, Antón Álvarez, Fridah Nyakundi

👍 Carly Batist, Sara Beery, Yves Bas, Elijah Cole (Deactivated), Andy Viet Huynh, Omiros Pantazis, Jaanak, Jaagat P.

Jaanak (jaanak007@gmail.com)

2022-07-25 15:01:07

We will be happy to just gain exposure and learn at this point of time, but maybe in the future! 🙂

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-07-25 15:02:30

If you live in a town with a university, see if there are any labs or professors working in the AI for good space and reach out. Many professors I know have high school students working in their labs or work with high school programs so it’s worth a shot to ask!
You might also check out the Conservation Tech Directory to scope out possible organizations/companies/projects to reach out to.

👍 Jaanak, Jaagat P.

Daniel Davila (daniel.davila@kitware.com)

2022-07-25 20:05:28

I have to echo Mark, it's actually more and more common for companies to take on high school talent. If you are contributing value to the enterprise, you deserve compensation!

That being said, what part of texas? There are a ton of great universities out there, the UT system for example, with a campus in every sector. There is a very large research institute in san antonio, SWRI, that may have programs for your level. You could also just jump into an open source citizen science project if you have your heart set on free labor, there are so many out there.

➕ Mark Goldwater

👍 Jaanak, Carly Batist, Jaagat P.

Jaanak (jaanak007@gmail.com)

2022-07-28 00:41:20

Thank you all very much for the guidance; I greatly appreciate it. I will also definitely consider looking around for paid opportunities as well!! 🙂

✅ Jaagat P.

Julia Marisa Sekula (jmsekula@stanford.edu)

2022-07-26 09:46:52

Hello Everyone!! So great to find a community of people working on similar issues. I’m here as @Lauren E. Gillespie was kind enough to invite me! I’m an MBA/Msc in Nature-Co Design and Climate Tech student at Stanford University. My background is 6 years in financial markets (distressed + special situations) and the last 3 years I spent in policy and climate tech in the Amazon Rainforest in Brazil :flag_br: (which is where I’m from - any Brazilians or Latin Americans please reach out). I’m particularly interested in genome sequencing and finding ways to collect, systematize, understand nature’s intelligence (and in our current systems, quantify nature’s economic value). I’m also currently developing Obvious Venture’s Synthetic Biology for Climate thesis. If anyone is looking at the Amazon and/or wants to talk ideas - I’d love to chat.

👋 Sara Beery, Andy Viet Huynh, Catherine Villeneuve, Marconi Campos, Jason Holmberg (Wild Me), Jaagat P., Mark Goldwater, Lauren E. Gillespie, Daniel Spokoyny, Suzanne Stathatos, Declan, Emilio Luz-Ricca, Jaanak, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, nyakundi lamech

🌎 Belen Saavedra, Jason Holmberg (Wild Me), Jaagat P., Andy Viet Huynh, Lauren E. Gillespie, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

👏 Ben Wilcox

Toryn Schafer (tschafer@tamu.edu)

2022-07-26 17:28:04

Hello, I've been lurking for a few months after learning about @Sara Beery’s work through the academic job market. I am an incoming assistant professor in the Statistics department at Texas A&M University. I'm broadly interested in machine learning in ecology and focus primarily on movement ecology using inverse reinforcement learning. Sara and I are actually scheduled in the same symposium for the upcoming TWS Annual Conference 2022. I hope to be more active in this community!

👋 Suzanne Stathatos, Sara Beery, Declan, Carly Batist, Andy Viet Huynh, Elijah Cole (Deactivated), Dan Morris, Jaagat P., Belen Saavedra, Devis Tuia, Dhruv Sheth, Emilio Luz-Ricca, Armin Bazarjani, Jaanak, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Kaiyang, nyakundi lamech, Marcus Lapeyrolerie

Suzanne Stathatos (suzanne.stathatos@gmail.com)

2022-07-26 17:34:30

Also, huge congrats on being done with the job market and landing a spot at A&M!

➕ Sara Beery

Toryn Schafer (tschafer@tamu.edu)

2022-07-26 17:42:42

Thank you!

Sara Beery (sara.beery@gmail.com)

2022-07-27 10:10:04

New issue of JMLR out with a focus on climate change!! https://www.jmlr.org/special_issues/climate_change.html

💜 Arjun Subramonian (they/them), Lukas Picek, Mark Goldwater, Jason Holmberg (Wild Me), Subhransu Maji, Andy Viet Huynh, Declan, Gabriel Tseng, Jaagat P., Armin Bazarjani, Carl Boettiger, Catherine Villeneuve, Eddie Zhang, Jaanak, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

🎉 Eddie Zhang, Kaiyang

Andrew Hartnett (a.t.hartnett@gmail.com)

2022-07-27 11:09:21

Hi everyone. I am a ML engineer currently working on models to understand and predict the behavior of roadway agents for autonomous driving. Previously I worked on problems at the intersection of physics, machine learning, and collective animal behavior (mostly shoaling fish). I’m exited to join this group to follow along with all your incredible conservation work, and to contribute on the ML side where possible.

👋 Sara Beery, Jason Holmberg (Wild Me), Toryn Schafer, Stefan Schneider, Andy Viet Huynh, Omiros Pantazis, Suzanne Stathatos, Emilio Luz-Ricca, Jaagat P., Declan, Jaanak, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, nyakundi lamech, Dhruv Sheth

🐘 Blair Costelloe

Lukáš Adam (lukas.adam.cr@gmail.com)

2022-07-27 11:10:55

I’m exited to join this group 😄 Anyway, welcome 🙂

Andrew Hartnett (a.t.hartnett@gmail.com)

2022-07-27 11:14:06

PS … my image was the result of asking DALLE-2 to produce “a.i. for wildlife conservation digital art”

Andrew Hartnett (a.t.hartnett@gmail.com)

2022-07-27 11:14:56

Thread Reply:

🤖 Stefan Schneider, Andy Viet Huynh, Kakani Katija, Carly Batist, Dhruv Sheth

🐘 Stefan Schneider, Andy Viet Huynh, Belen Saavedra, nyakundi lamech, Dhruv Sheth

👏 Declan, Dhruv Sheth

Ben Weinstein (benweinstein2010@gmail.com)

2022-07-27 15:57:53

Anyone on the statistical side of machine learning want to give some thoughts here? I've been thinking and talking (@Sara Beery, @Brad Pickens, and others) about how the outputs of deep learning models will be integrated with existing policy decision making tools. If we count a million trees in an area and classify each to species, what are the types of uncertainty to attach to point count estimates? When are these intervals useful? My background in bayesian ecology work (w/@Heather Lynch) doesn't fully jive with the outputs of these models. Maybe @Bistra Dilkina? For those who want to read more, I enjoyed (https://arxiv.org/pdf/2104.12953.pdf, https://proceedings.neurips.cc/paper/2019/file/8558cb408c1d76621371888657d2eb1d-Paper.pdf). Here is my summary.

Developing an accurate model for species prediction is the first step in operationalizing broad scale surveys. To illustrate how to convert the classification model into ecological information, we choose a simple task of counting the number of individuals of each species within a broad area. By assigning the species with the highest confidence score for each prediction, we can count the frequency of each class. However, this point estimate does not incorporate any uncertainty in these predicted counts. It gives no indication of the relative confidence among classes, and no way of assessing the reliability of predictions for future use. This makes it difficult to integrate into existing decision-making frameworks at the policy level. While creating prediction intervals for deep learning classification scores remains an open area of research (Ovadia et al. 2019, Lai et al. 2021, Abdar et al. 2021), there are three main avenues of uncertainty that can be assessed, 1) 'epistemic' uncertainty is the process noise from model training and optimization and can be assessed by training the model with different starting initializations and using the range of predictions as a confidence interval, 2) 'aleatoric' uncertainty is the process noise inherent in the data and can be assessed through cross-validation of the training/test data, 3) the per sample uncertainty measured by the calibrated confidence score given by the model. While a truly Bayesian perspective to deep learning is either computationally difficult or requires significant alteration to existing models (Kendall and Gal 2017), the confidence scores outputted by most deep learning systems can be calibrated to better reflect the probability of correct classification (Guo et al. 2017). After calibration, we can simulate from these probabilities and multiply each misclassification by the corresponding row in evaluation confusion matrix to create coarse confidence intervals for counts at broad scales. We compare these measures of uncertainty and discuss their relevance for downstream applications and management frameworks.

👍 Casey Youngflesh, nyakundi lamech

Heather Lynch (heather.lynch@stonybrook.edu)

2022-07-27 16:17:36

@Ben Weinstein I don't have a good answer for you but your own application is identical to one of ours, which is identifying seal species classified from a computer vision model to get a good estimate of species-specific abundance. There are two layers of uncertainty: 1) the classification of an seal (i.e. is there anything here at all?), and 2) its classification to species (if yes, what is it?). We've had a working group here at IACS of people interested in this exact question at the interface of ML and Bayesian inference because we are now combining ML classification with population models. Unfortunately, the two disciplines are quite far apart and a common understanding of error propagation methods is difficult to come by.

👍 Ben Weinstein, Sara Beery, Casey Youngflesh, Olivier Gimenez, Devis Tuia

Beckett Sterner (bsterne1@asu.edu)

2022-07-28 12:59:50

Don't forget that species classifications are often disputed and have substantial consequences for conservation planning/decisions! Value choices and uncertainty enter into the classification model from multiple inputs sources

👍 Sara Beery

Luke McEachron (luke.mceachron@gmail.com)

2022-08-02 09:51:03

@Ben Weinstein When thinking about “When are these intervals useful?”, could you approach this as a model selection problem? Specifically, could you allow an end-user to make various model assumptions to predict occurrence, then demonstrate performance relative to alternative models within a tool framework? Happy to discuss more.

Paul Allin (allinpaul@gmail.com)

2022-08-26 07:40:49

For my research I am looking at something similar, large terrestrial mammals in savanna biome. The first portion will be focused on establishing accuracy and precision from known populations. Once this has been determined it should be possible to scale up (assuming similar characteristics) and gather reliable data on population sizes and distributions. For vegetation I assume you're using satellite imagery and EVI/NDVI for some kind of semi-supervised classification?

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-07 00:30:39

I wanted to add this paper here because it was thoughtfully done.

👍 Sara Beery

Jitendra Hushare (jhushare@doc.govt.nz)

2022-08-01 20:27:56

Hello, I am working as an IT (Enterprise) architect with the Department of Conservation (DOC), New Zealand government. In the past, DOC staff captured thousand/millions of images that are being processed manually or semi-manually in isolated pockets/teams. We are eagerly looking for a working and implemented ML/AI-based enterprise solution for our staff to identify the animals, especially cats, mice, stoats, and possums (from the images) to meet our target of predator-free NZ by 2050. We are open to collaborating, adopting, developing and investing in such a proven suitable solution. Will greatly appreciate your thoughts, ideas and leads to help us solve this enormous problem. Feel free to connect or email me jhushare@doc.govt.nz. Thank you so much.

👍 Kakani Katija, Henrik Cox (Sentinel), Sara Beery, Rio Akbar, Andy Viet Huynh

Muskan Sachdeva (muskan@meta-lynx.com)

2022-08-02 08:30:12

Hi Jitendra,
I should be able to help you with that, I am working on a Data curation platform; sending you an email with the information

😎 Jason Holmberg (Wild Me), Sara Beery, Rio Akbar

👍 Jitendra Hushare

Henrik Cox (Sentinel) (henrik@conservationxlabs.org)

2022-08-02 13:47:52

Hi Jitendra, this sounds fantastic! Also just sent an email to see if there's a good fit with the work we're doing

👍 Jitendra Hushare

Nicholas Osner (nicholasosner@gmail.com)

2022-08-03 03:59:10

Hi Jitendra, I believe we at WildEye could help you out with TrapTagger - our free-to-use and open source platform. I have sent you an email with more details.

Ben Weinstein (benweinstein2010@gmail.com)

2022-08-08 12:00:54

Are there any students looking for a research project? I had a field researcher from french guiana write me about ant identification. Lots of interesting questions on taxonomic classification, multiple backgrounds. In general I get alot of these requests and don't know how they connect to the wider community. I can imagine starting from iNaturalist and then bringing in local domain knowledge? I think it really speaks to our need to go beyond the individual projects for each researcher.

👍 Sara Beery, Kakani Katija, Rowan Converse, Andy Viet Huynh

Ritwik (rittyun@yahoo.com)

2022-08-10 11:02:01

Hi, general question. When training object recognition models how do you handle the background class? when there are no objects of interest in the image. e.g. in pytorch, the label 0 is reserved for the background label but it's not clear what to do with the bounding boxes. they can't be 0 area as it leads the fractions to be invalid, and any other box (even if as large as whole image) can be a misleading label. may work if the background is more or less constant like fixed camera traps, but in general it's not clear to me how to deal with the background.

Chinmay Talegaonkar (ctalegaonkar@ucsd.edu)

2022-08-10 11:12:08

In terms of the loss function, you can try to use focal loss for single shot object detectors. For 2 stage, you don't need to handle that.
In terms of data processing, bounding boxes are usually regressed as anchor box offsets, and anchor box locations are fixed. Which is independent of the class label. One thing you can try, is use an indicator variable and don't penalize the loss on the bounding box prediction if the pixel is labeled as background.

Benjamin Kellenberger (benjamin.kellenberger@wur.nl)

2022-08-10 11:27:26

Hi!
What typical object detectors do is to simply ignore all the bounding boxes that are predicted for background. The way this is done depends on the model type:

Two-stage detectors (e.g., Faster R-CNN) check the intersection-over-union of each detected bbox with the ground truth; if it falls below a threshold it gets ignored.

One-stage models (esp. YOLO) predict bboxes in a grid over the image; here, the bbox that is predicted in the same grid cell as the ground truth (and has best overlap) is used for training; the rest gets ignored.

What is always trained is the class score (and objectness, if available), because these are deﬁned for every bbox (same principle as above).

By the way, 0 is not automatically reserved for background. This depends on the implementation—for example, Detectron2 uses 255 for the background. Also, some models have no background class at all: RetinaNet has sigmoid-activated class outputs and thus does not need to predict an extra background class.

👍 Justin Kay, Sara Beery, Ritwik

Daniel Davila (daniel.davila@kitware.com)

2022-08-10 12:17:21

You can also look at some of the newer anchorless approaches (e.g. FCOS, Centernet2, etc...) which densely classify each pixel, and then apply the bbox loss to just the positive pixels.

Ritwik (rittyun@yahoo.com)

2022-08-10 12:58:48

thanks a lot everyone. makes much more sense now. i was following the torchvision FRCNN tutorial to implement this and they mention to set the background class=0 (as reserved and not to be used for other classes) but don't say anything more on it.

so basically if the RPN network in a 2 stage model has no "objectness" boxes passing the criteria then the class-box prediction network has nothing to predict and would return nulls

Ben Weinstein (benweinstein2010@gmail.com)

2022-08-10 18:53:04

has anyone used https://labelstud.io/? I'm starting to looking into open source tools (or with cheap 'enterprise' add ons) where we can start creating active learning and model verification environments. We have been using zooniverse (both public and private) or locally annotating in QGIS for a few years and I think we are maturing beyond those tools where we want direct model integration and some control. Or atleast a toy model we use to guide annotations, which we than download and develop more novel models offline (more likely). I could build something from scratch (or piggyback on awesome tools like AIDE @Benjamin Kellenberger). Just a hypothetical for the moment.

labelstud.io

Label Studio – Open Source Data Labeling

Most flexible data labeling tool that supports all your data types. Prepare training data for computer vision, natural language processing, speech, voice, and video models.

Original URL: https://labelstud.io/

👏 nyakundi lamech

Sara Beery (sara.beery@gmail.com)

2022-08-10 19:05:10

We've been working on something like this with @Tom Bernardin and @Caleb Robinson for remote sensed building detection

😎 Jon Van Oast, nyakundi lamech

Ben Weinstein (benweinstein2010@gmail.com)

2022-08-10 19:22:29

sounds good. I think my general question is whether we should be building something or trying to rally behind some reasonable open source starting point.

👍 Matt Weldy

Tom Bernardin (tbernard@umass.edu)

2022-08-10 21:20:57

Hi Ben, I could connect you with my team if you want a demo of Caleb's tool. It is open source, as are (or will be) the modifications we made for our use case.

Devis Tuia (devis.tuia@epfl.ch)

2022-08-11 03:49:17

Ben, we can definitely talk for AIDE, to see if it can align to your needs. If you want we can setup a meeting together with @Benjamin Kellenberger.

Frederic (frederic@apic.ai)

2022-08-11 03:59:36

I am a big fan of CVAT and can help you if you would need some support or insights. They have a lot of integrations for automatic labeling functions using different frameworks or completly custom code. Good api, … https://github.com/openvinotoolkit/cvat

openvinotoolkit/cvat

Powerful and efficient Computer Vision Annotation Tool (CVAT)

Website

<https://cvat.org>

Stars

7710

👍 Timm Haucke, Howard L Frederick

Ritwik (rittyun@yahoo.com)

2022-08-11 04:50:39

i've just started using label-studio and really like it..
for the specific question, i would tend to lean towards using existing opensource tools.. simply because, like in case of label-studio, it does all the heavy-lifting allowing you to spend time and resources on your main goal rather than re-inventing the wheel.. it was even so easy to deploy it on the cloud and serve it to annotators remotely..

Matt Weldy (matthewjweldy@gmail.com)

2022-08-11 12:33:58

I keep checking to see if label studio has added a spectrogram option to their audio annotation features. There has been a ticket on their github for quite awhile.

Amrita Gupta (agupta375@gatech.edu)

2022-08-11 12:50:59

I've been using the free version of Label Studio for both text annotation and image pair annotation, albeit without any model integration to speed up/guide labeling. Overall it's not a bad starting point, but I do think a lot of useful functionality is only available in the enterprise version, which is a little frustrating.

Ben Weinstein (benweinstein2010@gmail.com)

2022-08-11 14:34:10

@Amrita Gupta do you get the sense that the model integration is possible without the enterprise version? It was unclear from the docs. In general i'm leaning towards label studio, but I just got word today that a potential next project has been using CVAT and i'll probably be asked not to re-invent too much, @Frederic I can imagine three avenues for model integration, 1) deciding which images are shown next, 2) pre-label images with existing markers, either from a live model, or a batch upload, 3) decide what portion of the image is to be shown, perhaps integrating with cloud geotifs. These are going to be large airborne tiles with mostly background. Does that sound about right?

Amrita Gupta (agupta375@gatech.edu)

2022-08-11 14:46:36

Yes model integration is possible with the free community edition (but you have to wrap your model code into a LabelStudio ML backend server class). You can then train/fine-tune models and get preannotations, which is enough to "hack" batch AL functionality. Online AL with continuously training models seems to be only available through the Enterprise version.

Ben Weinstein (benweinstein2010@gmail.com)

2022-08-11 16:46:27

Thanks @Devis Tuia, @Benjamin Kellenberger had talked about this sometime in past. I see three avenues for active learning (copied from above) 1) deciding which images are shown next, 2) pre-label images with existing markers, either from a live model, or a batch upload, 3) decide what portion of the image is to be shown, perhaps integrating with cloud geotifs. I wasn't sure whether we should go the truly open source route (AIDE, CVAT) or try to hack into a more commercial application. I haven't even gotten the final funding for this yet, so i'm just in the information phase.

Frederic (frederic@apic.ai)

2022-08-12 04:24:19

@Ben Weinstein
Regarding your questions, cvat was not designed for this active learning approach, anyhow using its api or db access you could integrate something that might work. Anyhow the speed might not be great.

1) deciding which images are shown next
While having a loop fetching labels from the api, you could add images to a job based to the finished annotations.
Possible issues: the browser of the client might have a cashed version of the task and need to retrieve the changes.

2) pre-label images with existing markers, either from a live model, or a batch upload,
Pretty easy and you can use different approaches:

use the api to create tasks, upload images and annotations.

If its a commen model and task you can just use the serverless functions and replace the example models with your models.

2.1 Since https://nuclio.io/ is used for serverless function, you can build your own nuclio container for labeling.

3) decide what portion of the image is to be shown, perhaps integrating with cloud geotifs.

use the api to upload cropped geotifs

You might be able to set Zoom block via the api. but not on a individual image basis. 😕

-> GeoTIFS and image tiling is not supported right now see, but you can implemented and get help from the community https://github.com/openvinotoolkit/cvat/issues/531

Overall CVAT might not be perfect but allows you to hack it, to make it work 🙂

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2022-08-18 18:57:52

Bit late to the party, but for what it's worth...

I've used both CVAT and Label Studio, and we've been using a commercial platform at work. We're also looking into active/unsupervised learning. In the past I've written labelling tools out of frustration with some of the limitations of open source stuff. My gut feeling is that it's sensible to split out functionality to the tools that do it best (e.g. storage, labelling, ML annotation+inference and active learning/sampling). The risk is that a lot of tools are almost there, but they cater to common use-cases that don't have a huge amount of complexity and none of them do everything perfectly.

Label studio is pretty good. I've used it in a pinch to do some semantic segmentation, though there were some funny GUI issues if you had overlapping polygons. I think for most part all of these tools are pretty good at tagging and bounding box annotation. The differentiators from a labeller perspective are around how quickly you can segment objects. CVAT is also good and has a usable, if tersely documented, REST API. Error handling is... cautious? For example if you upload images into a project and then upload labels, you must have labels for all images (even if empty) or the importer breaks.

In terms of integration and on the "ops" side. I think there is a lot of smoke and mirrors from commercial platforms about exactly what is implemented and how well - especially when they're charging you $100+/seat/month. A lot of companies will try and sell you black-box ML integration which can work very well within their walled gardens, but have issues like:

Can we adjust anything about the model? Often no.

What model you use? Proprietary, but probably Detectron2

Does it work with some weird TIFF that we have? Maybe

Does the service offer image tiling for large inputs, as is common in remote sensing? Often not.

Can we take out the model for use elsewhere/can we self host? No, or pay us a lot of money.

Can we specify exactly which splits are used during training? Often no

The list goes on.

Most of these are non-negotiable unless you have a huge contract and pull on where development goes. At least with open source you can see how the sausage gets made. The downside with larger projects like Label Studio and CVAT is that even if you have the technical capacity to PR things, it can take a while to get integrated.

There are other toolkits like Lightly who focus more on the active learning side of things. A lot of their unsupervised learning code is open source. Some functionality may be available for free if you're a non-profit. They're based up the road from me in Zurich and I've met them - they're a nice competent team, growing fast. Also fiftyone which is supposed to be good for model assessment, but really struggled when I imported a moderately sized dataset into it so I can't say much there.

I think Frederic's last answer is sane. You can get away with a labelling tool that has a relatively complete API for file management (like CVAT) and offloading the sampling and annotation stages to your own code. Uploading annotations from a pre-trained model, before handing images to annotators, is what we do at the moment.

(3) is the difficult one to do well IMO. There are some companies who are selling GIS oriented labelling solutions (e.g. Picterra), but not cheap. I agree the simplest approach is to do that work offline and just crop/upload tiles. Though in principle there isn't any technical reason why you couldn't use something like rasterio to specify a window/windows into the image? At least some of this is offloaded in CVAT by datamuro, so you could look at adding an importer there that could take e.g. a single source image and a list of windows, and then have datamuro handle the tiling and forwarding to CVAT.

👍 Ritwik

💯 Howard L Frederick

Ben Weinstein (benweinstein2010@gmail.com)

2022-08-19 17:08:01

I think this is a really excellent summary. We want control for own models for the active learning side. I will give label studio a try because it feels just insane to have to develop own tools for such a common process. If it looks like we just can't get the control we need, its either CVAT or AIDE. I asked a similar question last week to @gvanhorn who was giving a talk on Merlin and it is clear that Cornell have spent quite a bit of time developing their own tools from an existing open source repo.

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2022-08-20 09:05:06

At some point I think it'd be worth the remote sensing community pulling together on this, because it's clear there's common functionality that lots of groups need that isn't really provided by the standard open source tools. Maybe we (as a field) need to sit down and figure out what the requirements are and make some recommendations, and then decide on how it gets implemented - or at least document some best practices. There's a huge group of people that would benefit - basically the entire EO sector for a start. My experience talking to labelling companies is that remotely sensed imagery is a bit of an unknown to them - unfamiliar formats, processing paradigms, etc. Though many have said that they have customers who are requesting more geo-oriented features and I think we'll start to see more support in the future. Same goes for bioacoustics I guess, my limited understanding is there aren't great open source audio annotation tools at the moment (unless you want to just tag/cluster spectrograms).

(edit: addendum to this is that virtually all the GIS tools and geo libraries that would need to do the complicated things - stuff like GDAL - are open source, like how every AV/media company in the world depends on ffmpeg... So in that sense I think a lot of the hard work has already been done and this is mostly a data serving problem?

One major pro for commercial tools is that the interfaces are often really well polished because they have a team of frontend devs working to make things look good, and the labelling "experience" for lack of a better word is pretty good on some of these platforms. My worry about freemium tools like label studio are that you end up with features locked behind subscriptions and there's little incentive to make those public.)

Valentin Lucet (valentin.lucet@gmail.com)

2022-09-24 17:02:09

I wanted to restart this thread to ask whether anyone had any advice on how to serve megadetector's results through something like label studio or fyftyone to multiple annotators for manual correction and further annotation of the results.
It seems that one obstacle is to convert from the COCO CT format to the classic COCO format?

Abhay (abhaykash12@gmail.com)

2022-08-11 17:49:05

Hi folks! Dan Morris just added me and its great to be here! Wanted to quickly introduce myself - I've been working with Felidae Conservation Fund, a nonprofit out in SF-Bay Area for about a year. Last fall, I migrated their data pipeline to the cloud and built a human-in-the-loop annotation platform using Megadetector & Annotorious (over Django, Dropbox & GCP). If you're curious about the approach, I've written a post about it here.

I'm looking for feedback & opportunities for collaboration! Please let me know if there is any piece that I can abstract out and open-source! We do have plans to open source some of the data after it gets cleaned up as well!

(Some broader context - My background is mainly in computing and AI/ML/DL but applied to NLP & RecSys and I also do full-stack web development. While I used to do more hands-on ML research, I've zoomed out to ML system design and in this domain, my focus is mostly on building usable plug-and-play open-source software layers on top of existing models)

abhaykashyap.com

Building a simple AI-powered, human-in-the-loop system to manage wildlife camera trap images & annotations - Abhay Kashyap

Building a simple AI-powered, human-in-the-loop system to manage wildlife camera trap images & annotations

Original URL: https://abhaykashyap.com/blog/ai-powered-camera-trap-image-annotation-system/

🎉 Dan Morris, Stephanie O'Donnell, Andy Viet Huynh, Talia Speaker, Jason Holmberg (Wild Me), Ed Miller, Carly Batist, Alexander Robillard, Timm Haucke, Alan Ma, Rita Pucci, Sara Beery, Jeff Reed

❤️ Suzanne Stathatos, Sinan Robillard, Zara McDonald, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Sara Beery

👍 Alexander Robillard, Olivier Gimenez

Dan Morris (agentmorris@gmail.com)

2022-08-11 18:15:17

Welcome!

To your point about public data release, when FCF has data that is ready to share publicly (minus location information, and minus images of people), we would be happy to help get that up on lila.science (I'll even volunteer to do whatever data munging is required to get it into the same format we use for other camera trap data sets on LILA).

Abhay (abhaykash12@gmail.com)

2022-08-11 18:24:56

Thanks Dan! 🙂 We'll keep you in the loop through that journey! Luckily, WildePod was set up with continuous model training in mind so it should be easy to export it out into COCO or other standardized formats for LILA!

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2022-08-11 19:16:52

WildePod is very cool!

🙌 Abhay

Ed Miller (ed@hypraptive.com)

2022-08-11 19:36:04

Great post, @Abhay! I am on a similar journey myself with the BearID Project. While BearID is primarily focused on camera trap data, I am currently working on a full-stack web app to annotate and identify wild bears on the Explore.org web cams in Katmai National Park. Even though it would be better to adapt something, I have succumbed to the urge to build something from scratch, mainly to build my web skills and take advantage of being an AWS Community Builder. I have been documenting my journey on this blog. I was considering the VGG Image Annotator, but I'll have a look at Annotorious as well.

explore.org

Bear Cam - live from Brooks Falls in Katmai National Park | Explore.org

Watch LIVE as hundreds of Alaskan Brown Bears descend on a mile-long stretch of the Brooks River to feast on the largest Sockeye Salmon run in the world.

Original URL: https://explore.org/livecams/brown-bears/brown-bear-salmon-cam-brooks-falls

DEV Community

BearCam Companion

In my first post as an AWS Community Builder, I laid out my goal for year 1 in the program. It is to...

Original URL: https://dev.to/aws-builders/bearcam-companion-585i

robots.ox.ac.uk

Visual Geometry Group - University of Oxford

Computer Vision group from the University of Oxford

Original URL: https://www.robots.ox.ac.uk/~vgg/software/via/

Ed Miller (ed@hypraptive.com)

2022-08-11 19:37:55

Based on your screenshots, I need to up my UX game! 😄

Ben Weinstein (benweinstein2010@gmail.com)

2022-08-11 19:46:20

@Ed Miller and @Abhay have a look at the thread I started yesterday on annotation (https://aiforconservation.slack.com/archives/CLWGQ4BJ6/p1660171984124629). It feels like we are in need of some community development here. I sense that as a community we are moving from "does computer vision work for ecology" which was the question a few years ago (and motivated this work https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/1365-2656.12780) to "how do we develop iterative pipelines that tangible change long-term monitoring programs". I was giving a talk to @Sara Beery summer school on this kind of shifts in the community this week. I'm going to both read of your materials and try to write some summaries for the entire group next week. There are alot of competing goals and platforms here. @Jon Van Oast @Jason Holmberg (Wild Me) and wildbook have something to contribute here as well. Might be the inspiration for a small paper. I'm writing a grant on active learning this week and welcome input on how/should/if we should make something that aims for a more universal approach (is this even a maintainable/worthwhile goal) or ecological workflows are just too varied and sharing knowledge and research aims into active learning is a better strategy than tool building. i'm 100% open to thoughts, @Benjamin Kellenberger and I had a brief conversation surrounding this earlier today.

Ben Weinstein (https://aiforconservation.slack.com/team/UMDLZLL1K)

has anyone used <a href="https://labelstud.io/">https://labelstud.io/</a>? I'm starting to looking into open source tools (or with cheap 'enterprise' add ons) where we can start creating active learning and model verification environments. We have been using zooniverse (both public and private) or locally annotating in QGIS for a few years and I think we are maturing beyond those tools where we want direct model integration and some control. Or atleast a toy model we use to guide annotations, which we than download and develop more novel models offline (more likely). I could build something from scratch (or piggyback on awesome tools like AIDE @Benjamin Kellenberger). Just a hypothetical for the moment.

Original URL: https://aiforconservation.slack.com/archives/CLWGQ4BJ6/p1660171984124629

🎉 Jon Van Oast, Sara Beery, Abhay

Abhay (abhaykash12@gmail.com)

2022-08-11 22:22:07

Thank you! Also, if y'all want access to the system to get a feel for it, please DM me your email ids and I can add you in!

@Ed Miller - That is pretty neat! Happy to share my Annotorious code if you want. It is extremely simple and I'd highly recommend it especially when the number of options aren't many. You can even customize the UI a fair bit. In the gif that you see, it's simply using Bootstrap5 classes. And thanks, my UI is Bootstrap5 so the entire scss is maybe 100 lines mostly for colors and fonts 🙂

@Ben Weinstein - That was a pretty interesting discussion because they were essentially the same questions I went through when I was building this! I've gone about it in depth in the post but I'll add a TL;DR with some more thoughts that is very much in line with the discussion in the thread.
So the broad context here is that most of the design & dev happened last Fall. I'm very strongly against building anything so at the time, Wildlife Insights was an option I seriously considered (other one was Zamba) but decided against it for two main reasons.

The big one was that the nonprofit needed a more generic org-management system for inventory, the camera traps out in the field, its status of use, when it needed to be checked, potentially emailing volunteers to go check it etc. For this, I'd have to build a web app anyway and Django came with all this out of the box so it was not really any work.

Next was access to models which is something you've touched on in the thread. I wanted an online-learning human-in-the-loop system. Since Megadetector was already mature, we wanted to use it and not even tune it since it was very likely they'd put out upgrades (as they did a month ago). So our goal was species detection where it seemed like the models were still not very mature. So the intent here was to create a basic model and then, as annotators did their job, have the model update at a regular cadence. Since other volunteers fell through and I got busy, we didn't get to it but do intend to do it this fall if there isn't already an existing solution.

I'm not sure when Wild Me came about but I did learn about it early this year but WildePod was already up. The core goal for me was data standardization, ideally with object-level annotations from their existing images & camerabase annotations. After that, the hope was (and is) to keep useful pieces of WildePod and migrate elsewhere!

I strongly agree with the opinions in that thread about tool fragmentation. When I was first reading through Dan's blog post, I was surprised at how many custom solutions existed but it made sense given how nascent this is. Usually the tech matures first before the products and from my perspective, it looks like the models have matured with Megadetector/CVPR workshops but the software layer around it has some ways to go.

When I built WildePod, my goal was a "Bring your own storage & compute" approach to keep the expense around the website low. So the app is simply a layer around Dropbox. This way, any open source solution won't have to bear storage costs and people can link their own dropbox accounts. Another thing here is compute for inference. The Django app simply hits an endpoint so technically you can host your own model. Only downside here is that you might have to deploy it to the cloud yourself. I don't have an answer to the latter since it is a trade-off between cost of compute vs labor-cost for end user in terms of deployment.

Stepping outside the build scenario, since this is also a common CV task, there is tooling aimed at enterprises that have matured (labelbox, labelstudio, roboflow, scaleai, google's vertex AI etc.). Here I'd skew towards something like that when the dataset & the org is larger and has capital or if the company sponsors research/nonprofits. I had a conversation with Joseph Nelson, who runs Roboflow (invited him to this group) who reached out and wanted to see how they can help. They've had success with it in the past as well. In an ideal world, if there are small set of paid & free options that folks rally around, it'll greatly help standardize the data underneath. I think the key is data standardization so researchers can seamlessly jump vendors.

I'd love to chat and learn more about this. I've mostly done this in isolation while consuming content from the web (like wildlabs summer seminar series which was great!) so I probably have a lot of blind spots. I'm curious to know if folks here sync occasionally over zoom? If not, would there be any interest in doing so at some cadence?

Ed Miller (ed@hypraptive.com)

2022-08-13 21:42:30

@Ben Weinstein the thread and discussion are definitely of interest for BearID. I definitely think the time is right for convergence, at least on no-cost/low-cost labelling. For the Bearcam Companion application I am working on, a big part of it is for me to become more experienced with some of the technologies and tools for web development. I am very much building much of it from scratch. Longer term, and for the BearID Project more generally, the more standard tools we can utilize the better.

@Abhay I am interested in your code. I am especially interested in how you are saving data for different users (labelers). Can different users label the same image? In my case, for identifying the individual bear, I am allowing each user to select a label when editing, then provided a tabulation of "votes" in the main view, e. g. "480 Otis (4/6 = 66%)." I guess I could leave my current method for the top view and only use Annotorious for the editing. I will need to connect this to AWS DynamoDB through the Amplify DataStore API. It looks like I can use the Firebase as a reference.

Abhay (abhaykash12@gmail.com)

2022-08-19 01:24:06

Sorry just saw this.!

Yup. Multiple users can label the same image. That piece of logic is fairly primitive at the moment. As of now, the system will throw images with the least number of votes at users. The users can then add/update/delete boxes. This actually goes into the system as a "vote" per box.

So when another user sees the same image, they will see a union of boxes created by machines and humans.

We have a hardcoded threshold for now (~2) as the vote difference. When that threshold is breached, boxes will not be shown anymore.

Right now the weights are uniform. As things mature, we'll very likely add some weights to annotators depending on tenure/success rate etc. to soften the hardcoded threshold.

Abhay (abhaykash12@gmail.com)

2022-08-19 01:26:06

For the data part, since everything is in Django, it is simply writing it to and from the Django models. Underneath this, the database is actually Postgres hosted on CloudSQL. All this is fully abstracted through Django's ORM.

Depending on the infrastructure you're using you might find language specific ORMs to make write-read easier.

The biggest overhead here is usually CRUD on these tables. If the service makes it easy, then great. If you are maintaining a single row of data along the lines of with mostly appends & no related models, I'd recommend skipping relational DBs altogether and write directly to a beefy spreadsheet like Airtable (it has a simple python API)

(In my case, I had to manage user accounts, inventory, camera traps etc etc which required a basic relational DB)

Ed Miller (ed@hypraptive.com)

2022-08-21 21:16:16

@Abhay I integrated Annotorious to create/edit/delete the bounding boxes on my images. My website is build on AWS Amplify and React. It took me less than an hour to integrate with my UI and data store. I was able to add a label vocabulary, but the user has to start typing. How did you set yours up to only have the selection options you want?

For now, editing objects is an "admin" feature. I'm not dealing with multiple users yet. I'll need to expand my Objects tables to enable per users labeling. I already support that for the bear identification though.

Abhay (abhaykash12@gmail.com)

2022-08-22 12:26:13

That looks great! Since MegaDetector only has three classes, I render them as buttons. If it is a long list, you're likely be left with the option list. You could even try a mix of the two by rendering the top-K options as buttons with the option to choose more from.
You can do all this by customizing the widget [instructions here]. The example has the template to add any arbitrary html element you want (and also style them the way you want).
(I also DMed you my widget to see what modification might entail)

recogito.github.io

Writing Editor Widgets

This guide explains how you can write your own widgets and plug them into the annotation editor.

Original URL: https://recogito.github.io/guides/editor-widgets/

Valentin Lucet (valentin.lucet@gmail.com)

2022-09-25 09:01:09

Hi @Abhay! I was wondering whether the source code for this project is available somewhere? Your app demonstrates how to serve MD results to multiple users and I haven't seen any other examples like that.

👍 Sara Beery

Abhay (abhaykash12@gmail.com)

2022-09-26 17:18:56

Hi Valentin. Just saw this.
I have two repos up

https://github.com/hayabhay/megadetector-fastapi - This is a fastapi wrapper with a dockerfile that is ready-to-deploy on Google Cloud Run. It can essentially run any MD models (you can choose which one to bundle in docker) and can arbitrarily scale to users with GCR. (this is already running for us). Only downside is that GCR doesn't support GPUs so you'll be doing inference on CPUs (~15ish seconds per image). It also comes with a Streamlit UI to quickly test things

I also have an older version for MD v4 that was deployed as a Google Cloud Function. Again, its very similar to GCR but it doesn't require an api wrapper.

Feel free to let me know if that works for you and if you have any custom requirements. I'll be more than happy to work with you to get MD up on the cloud (my knowledge is limited to Google Cloud Platform however)

🙌 Josh Seltzer, Valentin Lucet, Jason Holmberg (Wild Me)

Abhay (abhaykash12@gmail.com)

2022-09-26 17:56:09

Also, in case I misread your question and If you're asking about the Django app that does human-in-the-loop annotations, it is right now in a private repo since it has some coupling with Felidae's branding. If you're interested, I can put out a ready-to-use open source version for it as well or if you're just interested to see the code and take pieces of it, I can add you to the org's Github. Let me know what works best. If you'd prefer to have a quick chat, let me know and I'll be happy to get on a zoom call with you!
(also adding this to the wider channel in case there are more use cases)

❤️ Valentin Lucet, Sara Beery, Stephanie O'Donnell

Valentin Lucet (valentin.lucet@gmail.com)

2022-09-26 18:27:19

Hi @Abhay! Thanks for the links, I've seen them a while back and already starred them 🙂 I am indeed asking about the django app side of things. I understand that a certain IP is attached with your app and that you may not be able to share it all.
One of the options we are considering for our project is to run MD on a research cluster (on GPUs) for speed once, and then serve the results (the images with the bounding boxes) for review by multiple users in our research team. That would require building an app similar to yours that would just serve the images with the boxes, offer the user the ability to vote for a species ID (I like your voting system) and possibly other tags (sex, age class) and also edit/add bounding boxes if any are missing (hence my interest in your integration with annotorious). I don't know if that is clear, but basically my webdev skills are limited and I'm exploring the feasibility of that idea on my end. Seeing the ORM model structure of the django app would be very helpful for me for example.

Valentin Lucet (valentin.lucet@gmail.com)

2022-09-26 18:28:22

So your offer of sharing ready to use version or to add me to the repo, as well as chatting, all or any of that would be of great help for me 🙂

Abhay (abhaykash12@gmail.com)

2022-09-26 18:34:41

Oops! Sorry for the initial misread! And yes, that sounds great. We can do any or all of it depending on how you want to proceed. The larger codebase is a bit gnarly since it does have a fair number of ad-hoc product related patches. Regardless, I can walk you through the code as needed and/or extract specific bits of it to fit any existing apps you have.

To start, I can add you to the repo so you get a sense for it. Then, we can quickly jump on a call so I can get a better sense of what you're planning to do and from there we can figure out how best to move forward. Also, we have a new volunteer who is working on the Django side of things as well and there might even be room to build reusable components along the way.

Valentin Lucet (valentin.lucet@gmail.com)

2022-09-26 18:36:39

Awesome, I will send you a DM!

Abhay (abhaykash12@gmail.com)

2022-09-26 18:36:59

🚀

🎉 Sara Beery

Swayam Thakkar (swayamt1302@gmail.com)

2022-08-22 23:55:15

Hello Everyone,
I am a ML enthusiast currently pursuing bachelor's in computer science from MIT World Peace University, Pune, India and also working at the Wildlife Institute of India on camera trap data.

I am glad to be a part of this community !!
Thank you @Sara Beery for adding me in !!

😎 Jason Holmberg (Wild Me), Dhruv Sheth, Jon Van Oast, Sara Beery

👋 Ed Miller, Dhruv Sheth, Abhay, Omiros Pantazis, Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Ritwik, Toryn Schafer, Dan Morris, nyakundi lamech, Alexander Robillard

👋:skin_tone_5: Ando Shah

Tjomme Dooper (tjomme@fruitpunch.ai)

2022-08-25 03:44:31

Hi everyone, I have been working on the AI for Wildlife Lab for a while now, but recently @Victor Anton was so kind as to point me here 🙏 Excited to hear from all of the projects that are going on and to get to know the people working on them! Any recommended channels?

Maybe the FruitPunch AI community can help out here and there by crowdsourcing some ML work. If you need extra hands for data wrangling, engineering, or analysis, if you'd like to explore many more ML models than you have time for, hit me up 🐘🍉

🙌 Stephanie O'Donnell, Jason Holmberg (Wild Me), Carly Batist, Alexander Robillard, Abhay, Sara Beery, Ed Miller, Fadel Seydou, Sinan Robillard, Victor Anton

Paul Allin (allinpaul@gmail.com)

2022-08-26 07:24:46

Hi,

Paul Allin (allinpaul@gmail.com)

2022-08-26 07:26:24

I'm Paul from South Africa, working on my PhD in automating aerial animal censuses by applying ML to remote sensed imagery. Keen to get in touch with others and thank you @Devis Tuia for adding me!

🙌 Stephanie O'Donnell, Dan Morris, Caleb Robinson

👋 Omiros Pantazis, Carly Batist, Declan, Fadel Seydou, Abhay, Caleb Robinson, nyakundi lamech, Benjamin Kellenberger, Alexander Robillard

Devis Tuia (devis.tuia@epfl.ch)

2022-08-26 09:10:50

welcome!

Ben Weinstein (benweinstein2010@gmail.com)

2022-08-26 09:56:49

hi paul! what taxa are you working on.

Caleb Robinson (calebrob6@gmail.com)

2022-08-26 17:08:32

Hi Paul, welcome to the community!

Paul Allin (allinpaul@gmail.com)

2022-08-27 05:35:22

Thanks! I'm looking at large (>50kg) mammals

Ben Weinstein (benweinstein2010@gmail.com)

2022-08-27 10:00:07

Out of curiosity, what happens when you apply our 'bird' detector to such images. https://deepforest.readthedocs.io/en/latest/bird_detector.html We'd like to formalize this towards an 'animal' detector in the next year or so.

Fadel Seydou (fadel.seydou@gmail.com)

2022-10-19 08:59:05

Hello @Paul Allin,
I'm Fadel from Switzerland (EPFL) currently working on my Master thesis on "automated aerial census of large herbivores". I would love to connect and discussion this topic with you and learn more about your work.

Paul Allin (allinpaul@gmail.com)

2022-10-22 03:17:55

Hi Fadel, great to hear about your masters. Maybe easiest to set up a call. Can you WhatsApp me? +27712287116

💯 Fadel Seydou

Devis Tuia (devis.tuia@epfl.ch)

2022-10-23 14:01:11

Hello @Fadel Seydou good to hear from an EPFL colleague 😄. In which lab are you?

Luke Sheneman (sheneman@uidaho.edu)

2022-08-26 11:41:40

Hi all - my name is Luke Sheneman and I work at the University of Idaho as Director of Research Computing. I am a co-investigator on an NSF DISES project using a camera trap grid deployed in Eastern Oregon that will be used to forecast interactions between predators and and ungulates/livestock, specifically in drought conditions. I am handling the AI side of things, including developing species classifiers and developing/deploying satellite-enabled edge AI. @Dan Morris pointed me here. Super excited to be a part of this community!

🎉 Dan Morris, Abhay, Toryn Schafer, Ethan Shafron, Alexander Robillard, Jason Holmberg (Wild Me)

👋 Carly Batist, Mark Goldwater, Omiros Pantazis, Benjamin Kellenberger, Carl Boettiger, Ed Miller

😎 Eddie Zhang, Sara Beery, Jason Holmberg (Wild Me)

Nicholas Osner (nicholasosner@gmail.com)

2022-08-30 10:26:36

Hi Luke. I work on an open-source camera-trap-processing web application called TrapTagger . At present, we only offer an in-house southern-African-species classifier, with MegaDetector as an empty-image remover for our users outside of Africa. However, we are looking to add more species classifiers and are in the process of collaborating with a few organisations to host their classifiers for both their own use, and the use of others. Perhaps we can assist you by providing you with a GUI and annotation workflow to wrap around the species classifiers you are working on, or you could even just provide your classifiers for others to use free of charge if you like. Send me an email at nic@innoventix.co.za if you would like to set up a discussion around this.

Michael Bunsen (notbot@gmail.com)

2022-09-14 17:29:55

Hi Luke, that sounds like an awesome project. I am based in Oregon but working remotely with several automated insect monitoring projects based in Montreal and the UK. I am handling the software and infrastructure side of things, but also hoping to meet some folks and setup some edge devices with partners in the Pacific Northwest.

Michael Bunsen (notbot@gmail.com)

2022-09-14 17:30:51

I'll take a look at TrapTagger as well @Nicholas Osner! I am working on a desktop application with some similar functionality, but focused on moth and insect classification.

Nicholas Osner (nicholasosner@gmail.com)

2022-10-03 03:27:46

Hi Michael, I'm sorry - I managed to miss your message until now. Let me know if you would like to have a chat about TrapTagger. Maybe you can make use of some our open source code. Worst case scenario, we have a bit of a chat about our respective projects. Let me know.

Abhay (abhaykash12@gmail.com)

2022-08-30 12:51:59

Hi folks! I was migrating to MegaDetector v5 over the past couple of days and the model looks great! Faster computation would mean lower cloud costs 🙂 As a thank you, I packaged MegaDetector with FastAPI and also added a Streamlit UI to quickly visualize & compare MegaDetector models. Please let me know if folks find this useful and also if you have any comments, feedback, suggestions!
Github repo -> https://github.com/hayabhay/megadetector-fastapi
In the upcoming weeks, I'll create some ready to use containers that can directly be deployed to Google Cloud Run for our purposes and share it in the repo as well. The migration itself was pretty smooth except for one gotcha - details in the thread.
(cc: @Dan Morris - please let me know if it makes sense for parts of this to move to the MegaDetector repo)

👍 Mitch Fennell, Sara Beery, Yves Bas

Abhay (abhaykash12@gmail.com)

2022-08-30 12:57:44

The biggest gotcha while migrating was using YOLOV5.

Since imports within the repo isn't relative, the repo root requires being in PATH/PYTHONPATH .

As a result, yolov5 packages aren't namespaced and having a top level module like utils will lead to non-obvious errors. So any referencing code must have it's own non-conflicting namespace at the root level.

Abhay (abhaykash12@gmail.com)

2022-08-30 12:59:39

Also, there was a fair bit of setup in terms of cloning the repo, checking out a particular commit and setting up the env. I've replaced it by automatically downloading the archive zip file and setting the necessary env from within the MegaDetector code itself.

Dan Morris (agentmorris@gmail.com)

2022-08-30 14:10:52

FWIW all of that stuff with checking out a particular commit and a particular version of PyTorch was working around this issue:

https://github.com/ultralytics/yolov5/issues/6948

...an incompatibility between certain versions of YOLOv5 and certain versions of PyTorch. It appears (in bold because I'm not 100% sure yet) that this has been resolved, i.e. that new versions of PyTorch and new versions of YOLOv5 now get along nicely and can run MDv5. YMMV. We'll update instructions if we're sure about that at some point.

👀 Abhay, Mitch Fennell

Abhay (abhaykash12@gmail.com)

2022-08-30 14:13:40

Ah! Interesting! I did forget to mention that the version of PyTorch was also a gotcha and that was more to do with PyTorch's compatibility with yolov5 (even with the specific commit).
I had to use these versions for it to work

torch==1.9.1
torchvision==0.10.1

Dan Morris (agentmorris@gmail.com)

2022-08-30 14:14:20

Also see this discussion:

https://github.com/ultralytics/yolov5/discussions/9138

I verified this in one (non-GPU) environment, and everything went fine. Need to verify in GPU environments.

👀 Abhay

Abhay (abhaykash12@gmail.com)

2022-08-30 14:25:23

Ah! I think the latest torch version that I used (torch==1.12.1 ) triggered this error AttributeError: 'Upsample' object has no attribute 'recomputescalefactor' (for the specific commit as well). Looks like it is still an open issue.

Also, I see the No module named 'utils' was mentioned here (likely from not being in the path). For me, this was because i had a utils.py in my own API that prevented it from being imported (since I had yolov5 in sys.path before the my project root).

<https://github.com/WongKinYiu/yolov7/issues/520|#520 'Upsample' object has no attribute 'recompute_scale_factor'>

Namrata Deka (dnamrata@cs.ubc.ca)

2022-08-31 04:11:52

Hi everyone! I am Namrata Deka and I'm a MSc student at UBC. I'm working on a project to learn image representations that are invariant of non-causal/spurious distractors (eg. background, lighting, camera angle etc.). I would love to apply this on a conservation dataset and was wondering if anyone here is aware of a dataset with continuous (or high-dimensional) target labels like bounding boxes or segmentation masks that also has continuous (or high-dimensional) labels for a distractor that is correlated with the target in the train set and less correlated or independent in the test(OOD) set?

🐘 Tjomme Dooper, Sara Beery, Jason Holmberg (Wild Me)

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-08-31 08:37:42

check out LILA BC - they have tons of animal camera trap, audio, etc. datasets that are labeled and ML-ready. https://lila.science/

LILA BC

LILA BC (Labeled Image Library of Alexandria: Biology and Conservation)

Est. reading time

1 minute

Original URL: https://lila.science/

Sara Beery (sara.beery@gmail.com)

2022-08-31 10:58:29

Explicitly labeled distractors via bounding boxes or segmentation masks isn't something that exists already for any of the LILA datasets as far as I know, but camera trap data is static so labeling a fixed background object in a given camera would be quite lightweight

👀 Namrata Deka, Juan Sebastián Cañas Silva

👍 Namrata Deka, Tiziana Gelmi Candusso

Namrata Deka (dnamrata@cs.ubc.ca)

2022-08-31 11:09:11

Thanks a lot for the suggestion. Good idea to take advantage of static images to add more labels. 🙂

Ed Miller (ed@hypraptive.com)

2022-09-04 15:01:51

I have published the Bearcam Companion web application along with the latest blog in the series. In this post I discuss how to use AWS Amplify Hosting and GitHub to create an auto-deployed web application to a custom URL. The blog post is here. The Bearcam Companion website is here.

DEV Community 👩💻👨💻

Bearcam Companion: Hosting with Amplify and GitHub

The Bearcam Companion application was pretty much ready to go after my last post on Lambdas. The...

Original URL: https://dev.to/aws-builders/bearcam-companion-hosting-with-amplify-and-github-5bpm

app.bearid.org

Bearcam Companion

Companion application for the Bearcam by BearID Project

Original URL: https://app.bearid.org/

🎉 Jon Van Oast, Jason Holmberg (Wild Me), Carly Batist, Eddie Zhang, Stephanie O'Donnell, Marconi Campos, Abhay, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Antón Álvarez

:bearid: Dan Morris, Jason Holmberg (Wild Me), Stephanie O'Donnell, Abhay, Crystal Huang

Chris Yeh (chrisyeh96@gmail.com)

2022-09-05 22:31:59

Several questions for the ML community here:

1. Are there any "state-of-the-art" techniques for transfer learning on a new set of labels? For example, training an animal species classifier for one set of species, then transfer to a new (but possibly overlappping) set of species. (By "state of the art", I mean any technique that isn't just fine-tuning the last layer of a neural net.)

2. What is the state-of-the-art approach for handling hierarchical labels in ML models? MegaClassifier provides some post-hoc utilities to remap labels based on the biological taxonomy. But what about methods for leveraging hierarchical labels during training?

3. What about transfer learning with hierarchical labels? Say your source and target label sets have overlapping hierarchies - is there a way to leverage that info without completely throwing away the last layer of your neural net?

This hierarchical transfer learning problem increasingly common in the projects I'm working on, but I don't have a good way to solve it. If anyone has worked on this problem before, please let me know what you've tried!

😊 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Chris Yeh (chrisyeh96@gmail.com)

2022-09-05 22:34:20

Tagging some friends, in case y'all have any ideas and/or have read good papers on these topics: @Sara Beery @Lily Xu - OLD ACCOUNT, DO NOT MESSAGE @Suzanne Stathatos

👀 Suzanne Stathatos

Mark Goldwater (mgoldwater@whoi.edu)

2022-09-05 23:06:21

Out of curiosity re: 1., have you experienced any problems with simply fine-tuning? Are you trying to achieve better results than that approach provides? More efficient training that exploits overlap?

Beckett Sterner (bsterne1@asu.edu)

2022-09-05 23:39:22

@Atriya Sen 👆

Chris Yeh (chrisyeh96@gmail.com)

2022-09-06 01:56:49

@Mark Goldwater: correct, I'm asking about ways to achieve better results than simple fine-tuning

Devis Tuia (devis.tuia@epfl.ch)

2022-09-06 02:11:17

I think Alexander Mathis has some solutions for transfer learning. They are learning global models (they call them “supermodels”) for pose estimation of different species (e.g. a model for canines, one for birds) that can then be finetuned . All based on deeplabcut. For efficient finetuning, we generally use our software AIDE, with pretrained models that can be finetuned using some active learning scheduling to minimize the tuning cost. Check with @Benjamin Kellenberger if you want more details.

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-06 11:11:07

not exactly the answer, but related: I am working on a fine-grained tree species classification paper that uses a set of nested hierarchical models, we found that pretraining on the flat model, then using that as a starting point for each hierarchical model was useful. Similarly, our 'bird' detection model has shown to be useful as a starting point for bird species classification.

👍 Sara Beery

Suzanne Stathatos (suzanne.stathatos@gmail.com)

2022-09-06 12:16:56

I second @Benjamin Kellenberger as the right person to ask on this. Generally what I've seen is that fine tuning works just as good as if not better than complex methods for transfer learning - though I could be in the same boat as you.

@Elijah Cole (Deactivated) has worked a lot on understanding the costs/benefits on label granularity I.e his paper here. He probably has good insight on (3) the hierarchical labels aspect.

arXiv.org

On Label Granularity and Object Localization

Weakly supervised object localization (WSOL) aims to learn representations that encode object location using only image-level category labels. However, many objects can be labeled at different...

Original URL: https://arxiv.org/abs/2207.10225

👍 Chris Yeh, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Chris Yeh (chrisyeh96@gmail.com)

2022-09-06 13:35:05

Thank you all for your responses!

@Ben Weinstein: how are you training the "set of nested hierarchical models"? And are those hierarchical labels different from the flat pretraining?

@Devis Tuia: sounds interesting - do you have any paper/publication you can point me to?

@Benjamin Kellenberger: everyone seems to be tagging you - would love to hear your thoughts on this!

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-06 14:43:23

I don't know if this figure will be useful yet, but here is the conceptual figure from the draft. They are the same labels in both architectures, that's why I said it wasn't exactly what you were looking for. So training the flat model for all classes and then using that as the starting point for each of the models on the left.

❤️ Suzanne Stathatos

Chris Yeh (chrisyeh96@gmail.com)

2022-09-06 15:01:30

@Ben Weinstein: Is this understanding correct? You first train a classifier to predict the flat labels. Then you fine tune multiple instances of the classifier to focus of subsets of the flat labels. Finally, you combine these fine-tuned models in a hierarchical way. (But the individual models themselves are not hierarchical?)

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-06 15:07:05

yes I think thats reasonable. When you say the individual models are not hierarchical, that interests me, if you mean that they have separate optimizers then yes. I've actually been really interested to see if anyone can show me a mixture-of-experts type hierarchy that is co-trained with a single combined loss function. Just for implementation, it feels confusing how to feed batches into such a network. We found it much easier to train separately the pieces and create separate dataloaders for each. Its definitely less efficient.

👍 Valentin Gabeff

Devis Tuia (devis.tuia@epfl.ch)

2022-09-06 15:12:14

@Chris Yeh, for the reference to AIDE: https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/2041-210X.13489. For a reference to the supermodels, maybe @Valentin Gabeff can you provide one from Alex’s team (Alex is not in the slack channel)?

👍 Valentin Gabeff

Valentin Gabeff (valentin.gabeff@protonmail.ch)

2022-09-06 18:29:35

Alexander pointed me to this pre-print:
https://arxiv.org/pdf/2203.07436.pdf

The approach is a bit different as applied to animal pose estimation but the general design shares some ideas with what has been mentionned:
Pre-train a 'SuperAnimal' model that performs well on different species, and then fine-tune it for a specific dataset (or in this case it would be fine-grained species) to improve performance.
(see details about gradient masking and pseudo-labeling in the pre-print)

And no, Alexander is not here (yet?) ;)

👍 Ben Weinstein

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-06 20:44:02

this is a really nice paper and something we would like to move towards for UAV-based object detection.

👍 Devis Tuia

Atriya Sen (atriya@atriyasen.com)

2022-09-07 12:21:59

You may be interested in our paper here: https://ojs.aaai.org/index.php/AAAI/article/view/17750, where we combine learning with hierarchical labels and automated reasoning over taxonomies.

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2022-09-19 18:06:50

Hi Chris! Very late to respond here and I think others on this thread would be much more helpful on supervised learning tasks (I've focused much more on planning)

But in the planning literature, there may be some interesting ideas to explore:

learning diverse skills with RL (very related to transfer learning): this approach uses maximum entropy to train a base model, then use that pretrained model for specific downstream tasks. hierarchical RL to help solve more complex tasks https://arxiv.org/pdf/1802.06070.pdf

multi-task RL: uses "soft modularization" to train policy with shared parameters on multiple different tasks https://arxiv.org/pdf/2003.13661.pdf

And one idea in the supervised learning world I really like is MIMO, multi-input multi-output. The idea is to use a single NN to act as an ensemble with N multiple inputs and N multiple outputs. Then you average the outputs as your final prediction. It's nice because it helps with stability (benefits from ensemble) and requires minimal extra storage or computation. There may be variations of this NN architecture that you could adapt to transfer learning skills. https://arxiv.org/pdf/2010.06610.pdf

😍 Suzanne Stathatos

Holger Klinck (hk829@cornell.edu)

2022-09-09 09:29:27

Hi everyone,

We are looking for new team members for our BirdNET team (birdnet.cornell.edu)! These are full-time, 3-year positions. Specifically, we are looking for:

An Ecologist:
https://www.tu-chemnitz.de/verwaltung/personal/stellen/257080_4_EPu.php

A Data Scientist:
https://www.tu-chemnitz.de/verwaltung/personal/stellen/257080_5_EPu.php

An Embedded Systems Engineer:
https://www.tu-chemnitz.de/verwaltung/personal/stellen/257080_6_EPu.php

The job descriptions are provided in German and English (scroll down to the bottom of the page for the English version). These positions will be hired through Chemnitz University in Germany. Remote work is possible; however, successful international applicants will be required to obtain a German work permit.

If you have questions about these positions, please get in touch with Stefan Kahl @ sk2487@cornell.edu.

Cheers,

Holger

👍 Oisin Mac Aodha, Stephanie O'Donnell, Frederic, Andy Viet Huynh, Jaanak, Josh Veitch-Michaelis, Rita Pucci, Sara Beery, Yuanqi Du, Amandine Gasc

👍:skin_tone_2: Swayam Thakkar

💯 Carly Batist

Sara Beery (sara.beery@gmail.com)

2022-09-12 18:44:39

Hi to all the new faces on here!! Feel free to introduce yourselves 😄

👍 Jaanak, Yuanqi Du, Jason Holmberg (Wild Me), Jason Parham, Dhruv Sheth, Francisco Carrillo Pérez

💯 Jon Van Oast, Jason Holmberg (Wild Me), Heather, Francisco Carrillo Pérez

Francisco Carrillo Pérez (carrilloperezfrancisco@gmail.com)

2022-09-13 03:01:52

Hi all! 👋 my name is Francisco Carrillo Perez, and I am a last-year PhD student at the University of Granada, Spain, and a Visiting Researcher at Stanford University, USA! I work in multimodal classification and multimodal generative models applied to bioinformatics problems (mainly cancer data!) but I am really interested in conservation, and I want to learn more about how AI can help. Looking forward to it!

👋 Devis Tuia, Riccardo de Lutio, Benjamin Kellenberger, Declan, Sara Beery, Suzanne Stathatos, Dan Morris, Lucia Gordon, Andy Viet Huynh, Jason Holmberg (Wild Me), Antón Álvarez, Lauren E. Gillespie, Rita Pucci, Chris Yeh

Thiên-Anh Nguyen (thien-anh.nguyen@proton.me)

2022-09-13 04:02:43

Hi everyone! I'm Thiên-Anh, a PhD student at @Devis Tuia's lab at EPFL (Switzerland). I am working on monitoring and understanding treeline dynamics in the Swiss Alps using remote sensing and deep learning 🌲. Methods-wise, I'm interested in explainable deep learning, semantic segmentation, time series, noisy/multi-sensor data, domain adaptation and many more (basically all the challenges we face when using data from the real world)... I'm not working exactly in conservation, but I'm interested in all the methods and projects discussed here, and hope I can bring something to the discussions! 👋😊

🎉 Robin Zbinden, Benjamin Kellenberger, Andrés C Rodríguez, Sara Beery, Dan Morris, Jason Holmberg (Wild Me)

👋 Diego Marcos, Riccardo de Lutio, Benjamin Kellenberger, Sean Nachtrab, Robin Zbinden, Declan, Sara Beery, Suzanne Stathatos, Andy Viet Huynh, Lucia Gordon, Jason Holmberg (Wild Me), Lauren E. Gillespie, Rita Pucci

🌳 Burak Ekim

Sean Nachtrab (sean.nachtrab@gmail.com)

2022-09-13 08:17:57

Hello all, I'm a machine learning engineer. I've been following this slack for a while and am very interested in what's being worked on!
I work in the rail sector on autonomy at otiv.ai in Belgium but am American. I've been interested in climate, ecology and environment since my BSc at King's but the pandemic encouraged me to be pragmatic with my employment choice upon finishing uni. I have a few friends in ecology/environmental PhDs who do really interesting stuff and want to help!

I've spent the last two years in industry working on sensors (camera, lidar, ...) and computer vision (segmentation, detection, classification, ...)
I've grown quite disillusioned of full time startup work, especially in autonomous vehicles and I'm considering going back to uni for a PhD or changing my line of work to have more free time and a more substantial positive impact on the world/environment.

I'm looking to offer help in computer vision, remote sensing, dataset creation and do volunteer work as well as discuss PhDs and applied ml/computer vision, feel free to message if you want to chat or have opportunities!

👋 Thiên-Anh Nguyen, Declan, Tjomme Dooper, Omiros Pantazis, Sara Beery, Suzanne Stathatos, Dan Morris, Andy Viet Huynh, Kakani Katija, Lucia Gordon, Jason Holmberg (Wild Me), Rita Pucci

Nasrin Montazeri (n_seemorgh@yahoo.com)

2022-09-13 11:25:12

Hello everyone. I'm Nasrin Montazeri from Iran & I'm an electrical engineer in a corporation in the HVAC industry in my hometown. and looking forward to a Ph.D. position to pursue my education. Besides, I'm a semi-professional rock climber and mountain addict. Also, I'm studying french at B1 level. À bientôt :)

👋 Sara Beery, Omiros Pantazis, Declan, Suzanne Stathatos, Dan Morris, Andy Viet Huynh, Lucia Gordon, Jason Holmberg (Wild Me), Rita Pucci

Ditiro Rampate (ditirorampate@gmail.com)

2022-09-13 11:42:44

Hello everyone👋, I am Ditiro from Botswana. I am a volunteer ML engineer with OmdenaAI where we solve global challenges using AI, from climate change to wildlife conservation. I am currently looking for job and graduate(Masters in AI or computer vision) opportunities. SO great great to be here😉

👋 Sara Beery, Declan, Suzanne Stathatos, Dan Morris, Andy Viet Huynh, Lucia Gordon, Jason Holmberg (Wild Me), Jon Van Oast, Omiros Pantazis, Malte Pedersen, Rita Pucci

Viktor Domazetoski (viktor.domazetoski@hotmail.com)

2022-09-13 13:49:11

Hello everyone! I am Viktor and I am from Macedonia, though currently I am studying in Goettingen, Germany. Although I finished my Bachelors in Computer Science and continued a Masters in Statistics I decided that I want to use my skills in AI and Data to help save the environment. This led me to start a second master in Ecosystem Modelling of which I just finished my first year. Currently I am looking at PhD opportunities in the interface between these two fields utilizing fields such as Network Science, Natural Language Processing and Computer Vision. I am very excited to find this community and pleased to meet you all and feel free to reach to chat about anything :D

👋 Declan, Lucia Gordon, Jason Holmberg (Wild Me), Sara Beery, Toryn Schafer, Andy Viet Huynh, Jon Van Oast, Dan Morris, Devis Tuia, Eddie Zhang, Omiros Pantazis, Suzanne Stathatos, Lauren E. Gillespie, Rita Pucci

G. LeBuhn (lebuhn@sfsu.edu)

2022-09-13 14:59:00

Hello everyone. I'm a plant and insect biologist in San francisco (SFSU). I am interested in using sensors to monitor insects (bees and herbivores) and to track ecosystem services - particularly pollination service. I am hoping that computer vision will help manage and process the image data.

👍 Sara Beery, Declan, Jon Van Oast, Dan Morris, Omiros Pantazis, Michael Bunsen, Jason Holmberg (Wild Me)

👋 Cameron Trotter, Sara Beery, Suzanne Stathatos, Lauren E. Gillespie, Rita Pucci

🐝 Ando Shah

Michael Bunsen (notbot@gmail.com)

2022-09-14 17:12:25

Hello G! I am working on an autonomous insect monitoring project in a research lab at Mila (Quebec). We are focused on lepidoptera in the moment.

Here is a recent paper that describes some of our current methods and partners:
https://www.cell.com/trends/ecology-evolution/fulltext/S0169-5347(22)00134-3

And here is webinar that showcases some of the hardware projects in development: https://youtu.be/2Z8aG7qYAa0

I am currently working remotely from Portland, Oregon and I would love to connect with more folks interested in this topic on the west coast!

Trends in Ecology & Evolution

Emerging technologies revolutionise insect ecology and monitoring

Insects are the most diverse group of animals on Earth, but their small size and high diversity have always made them challenging to study. Recent technological advances have the potential to revolutionise insect ecology and monitoring. We describe the state of the art of four technologies (computer vision, acoustic monitoring, radar, and molecular methods), and assess their advantages, current limitations, and future potential. We discuss how these technologies can adhere to modern standards of data curation and transparency, their implications for citizen science, and their potential for integration among different monitoring programmes and technologies.

Original URL: https://www.cell.com/trends/ecology-evolution/fulltext/S0169-5347(22)00134-3

YouTube

UK Centre for Ecology & Hydrology (https://www.youtube.com/c/UKCentreforEcologyHydrology)

Workshop 1: Hardware design for camera traps for moth monitoring

Original URL: https://youtu.be/2Z8aG7qYAa0

Valentin Gabeff (valentin.gabeff@protonmail.ch)

2022-09-14 03:23:45

Hello everyone, I'm Valentin from Switzerland. I have just started a PhD in the same EPFL lab as @Thiên-Anh Nguyen, co-supervised by @Devis Tuia's and Alexander Mathis.
I will be working on the interaction between wildlife and the environment using CV & DL with a focus on the Swiss Alps (project description).

Looking forward to share thoughts here and hopefully to meet when we have the occasion.

👋 Benjamin Kellenberger, Robin Zbinden, Thiên-Anh Nguyen, Omiros Pantazis, Sara Beery, Declan, Dan Morris, Suzanne Stathatos, Jason Holmberg (Wild Me), Lauren E. Gillespie

🎉 Robin Zbinden, Sara Beery, Rita Pucci

serge sarkis (sergesarkis7@gmail.com)

2022-09-14 04:03:00

Hello everyone, I'm Serge from Lebanon. I have an MS in mechanical engineering with emphasis on smart materials and soft robotics. I'm currently a research assistant at the American University of Beirut. We are building a reforestation robot and planning to study the biodiversity of local wildlife sanctuaries and national forest using CV and DL.
Very excited to join the community!

👋 Omiros Pantazis, Sara Beery, Declan, Dan Morris, Suzanne Stathatos, Lauren E. Gillespie, Rita Pucci

Graeme Phillipson (graeme.phillipson@bbc.co.uk)

2022-09-14 11:33:08

Hello! I’m Graeme from BBC Research & Development. We’re interested in camera traps and remote cameras for producing natural history television programmes.

👏 Lucia Gordon, Oisin Mac Aodha, Stephanie O'Donnell, Carly Batist, Felipe Parodi, Sara Beery, Dan Morris, Suzanne Stathatos, Jason Holmberg (Wild Me), Cathy Atkinson, Viktor Domazetoski, Lauren E. Gillespie, Omiros Pantazis, Ștefan Istrate

👋 Cameron Trotter, Valentin Gabeff, Sara Beery, Jason Holmberg (Wild Me), Georgia Atkinson, Adam Noach, Rita Pucci

😎 Jon Van Oast, Sara Beery

🦁 Roni Choudhury

⚡ Roni Choudhury

Graeme Phillipson (graeme.phillipson@bbc.co.uk)

2022-09-14 11:36:01

https://www.bbc.co.uk/rd/blog/2021-04-winterwatch-artificial-intelligence-automated-monitoring

BBC R&D

AI on Winterwatch: An automated eye for watching the wildlife

We have developed our system for automated remote camera monitoring - now identifying fish and potentially 'interesting' moments.

Original URL: https://www.bbc.co.uk/rd/blog/2021-04-winterwatch-artificial-intelligence-automated-monitoring

🙌 Stephanie O'Donnell, Sara Beery, Justin Kay, Emily Lines, Ștefan Istrate

Robert Dawes (robert.dawes@bbc.co.uk)

2022-09-20 06:10:08

Hi, I'm Robert, also from BBC Research & Development and in the same team as Graeme.

🙌 Stephanie O'Donnell, Adam Noach, Cathy Atkinson, Andrew Schulz, Oisin Mac Aodha, Omiros Pantazis, Sara Beery, Gedeon, Ștefan Istrate, Viktor Domazetoski

👋 Georgia Atkinson, Sara Beery, Rita Pucci, Cameron Trotter, Declan

Matthew Judge (matthew.judge@bbc.co.uk)

2022-09-21 06:15:24

Hi everyone! Also joining from BBC R&D, and also in Rob's and Graeme's team 🙂

👋 Georgia Atkinson, Oisin Mac Aodha, Stephanie O'Donnell, Cameron Trotter, Rita Pucci, Andrew Schulz, Sean Nachtrab, Omiros Pantazis, Sara Beery, Catherine Villeneuve, Dan Morris, Jon Van Oast, Eddie Zhang, Jason Holmberg (Wild Me), Declan, Ștefan Istrate

Chinmay Talegaonkar (ctalegaonkar@ucsd.edu)

2022-09-14 11:51:26

Hello everyone, I am Chinmay from India. I am just starting my PhD at UC San Diego in the ECE department. I will be focusing on 3D vision problems, and real world applications of computational imaging methods. I am quite interested in finding applications of my research for Wildlife conversation! Before starting my Ph.D. I worked as a deep learning engineer at a startup in Bay area, where I developed object detection models for industry use cases. I attended @Sara Beery’s talk at ICCP 2022 at Caltech, which brought me to this amazing group! Outside of work, I like hiking and visiting national parks.

👋 Valentin Gabeff, Stephanie O'Donnell, Toryn Schafer, Sara Beery, Dan Morris, Lucia Gordon, Graeme Phillipson, Suzanne Stathatos, Jason Holmberg (Wild Me), Lauren E. Gillespie, Rita Pucci

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-14 12:47:32

I'm working this week on making/contributing to an open-source python package for aligning raw airborne imagery from UAV with the stitched georeferenced mosaic so that users can run machine learning models on high quality raw imagery and then try to place those detections/classifications into world coordinates. Many of you will have experience in these areas and I'm looking for any opinions/data on workflows. Given that the orthomosaic exists in a 3d or 2.5d space, every pixel in the raw imagery does not have a corresponding world coordinate. The larger the z dimension displacement, the larger the shift in the transformed location (orange box). This is in collaboration with https://github.com/UTokyo-FieldPhenomics-Lab/EasyIDP, which is a package that performs the reverse action, finds the location of raw imagery given a point in the orthophotos. I think this is a really underserved part of the workflow and something you can find many posts about in the assorted photogrammetry tools (pix4d, agisoft, etc), but no detailed workflow and example. All thoughts welcome. See this thread for technical details/code: https://github.com/UTokyo-FieldPhenomics-Lab/EasyIDP/discussions/44#discussioncomment-3579788

UTokyo-FieldPhenomics-Lab/EasyIDP

Stars

Language

Python

<https://github.com/UTokyo-FieldPhenomics-Lab/EasyIDP/discussions/44#discussioncomment-3579788|Comment on #44 Object detection in raw imagery>

👍 Rowan Converse

😍 Suzanne Stathatos, Sara Beery, Rita Pucci

🌳 Sara Beery

👀 Sean Nachtrab

Daniel Davila (daniel.davila@kitware.com)

2022-09-14 15:33:14

One of my colleagues specializes in UAV based 3d vision, and they're working through a lot of these problems on our noaa adapt program. I bet he'd love to talk your ear off about what they've done, if you want me to connect yall. Just let me know! This is outside my wheelhouse though lol

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-14 15:48:39

happy to talk to anyone, it feels like the kind of task that has been solved (its more annoying than it is hard) but just does not exists in a repeatable way. I keep waiting for someone to just appear and be like, oh that's the 'export shape annotations' button, etc. But I've talked to developers from all the software companies and I haven't seen anything that suffices.

Tjomme Dooper (tjomme@fruitpunch.ai)

2022-09-15 05:43:24

Hey Ben, this sounds super relevant to a project we're running at FruitPunch at the moment.

Tjomme Dooper (tjomme@fruitpunch.ai)

2022-09-15 05:44:09

https://www.fruitpunch.ai/challenges/ai-for-trees

Ethan Shafron (ethan.shafron@gmail.com)

2022-09-16 09:37:19

I've worked on similar tasks where we used template matching with openCV to align overlapping geo-referenced DiMAC images - it's definitely not the fastest or sleekest option, but I think it's relevant. I'm not sure exactly how well it would work for trying to match ortho vs raw images, but it might be worth testing as a relatively simple baseline.

👍 Ben Weinstein

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2022-09-18 13:38:23

When we did this sort of thing for Mars (matching rover ground imagery to orbit, different satellites, etc). Typically the process was to generate key points (SIFT) in both domains and then run some of robust matching and warping process. Which is probably the "draw the rest of the -- owl" bit... and I'm not sure how flexible the pipeline was for non Martian data...

Probably depends somewhat on expected localisation accuracy? But also if you're doing detection on imagery that was already used to generate the orthomosaic then don't you already have a warping function that's used to project points from each raw image into the ortho? (given the camera intrinsic/extrinsic info)

https://www.sciencedirect.com/science/article/pii/S0019103516303086?via%3Dihub

sciencedirect.com

Automated localisation of Mars rovers using co-registered HiRISE-CTX-HRSC orthorectified images and wide baseline Navcam orthorectified mosaics

We present a wide range of research results in the area of orbit-to-orbit and orbit-to-ground data fusion, achieved within the EU-FP7 PRoVisG project …

Original URL: https://www.sciencedirect.com/science/article/pii/S0019103516303086?via%3Dihub

Ben Koger (benkoger@gmail.com)

2022-09-27 11:23:15

We have a preprint (also in review) that describes our method for doing this (or at least similar) with drone video: https://www.biorxiv.org/content/10.1101/2022.06.30.498251v1 Basically use structure from motion to generate 3D georeferenced landscape models from some of the raw images and then use the resulting calculated camera matrices from those frames to project pixel locations into the georefernced 3D landscape space. We use local features in the raw frames to estimate the correct camera matrix even for frames not used to build the landscape model. The overal code we use is here: https://github.com/benkoger/overhead-video-worked-examples

bioRxiv

Multi-animal behavioral tracking and environmental reconstruction using drones and computer vision in the wild

1. Methods for collecting animal behavior data in natural environments, such as direct observation and bio-logging, are typically limited in spatiotemporal resolution, the number of animals that can be observed, and information about animals’ social and physical environments. 2. Video imagery can capture rich information about animals and their environments, but image-based approaches are often impractical due to the challenges of processing large and complex multi-image datasets and transforming resulting data, such as animals’ locations, into geographic coordinates. 3. We demonstrate a new system for studying behavior in the wild that uses drone-recorded videos and computer vision approaches to automatically track the location and body posture of free-roaming animals in georeferenced coordinates with high spatiotemporal resolution embedded in contemporaneous 3D landscape models of the surrounding area. 4. We provide two worked examples in which we apply this approach to videos of gelada monkeys and multiple species of group-living African ungulates. We demonstrate how to track multiple animals simultaneously, classify individuals by species and age-sex class, estimate individuals’ body postures (poses), and extract environmental features, including topography of the landscape and game trails. 5. By quantifying animal movement and posture, while simultaneously reconstructing a detailed 3D model of the landscape, our approach opens the door to studying the sensory ecology and decision-making of animals within their natural physical and social environments. ### Competing Interest Statement The authors have declared no competing interest.

Original URL: https://www.biorxiv.org/content/10.1101/2022.06.30.498251v1

benkoger/overhead-video-worked-examples

Worked examples of behavioral data extraction from overhead video.

Language

Jupyter Notebook

Last updated

6 months ago

👍 Ben Weinstein

Sara Beery (sara.beery@gmail.com)

2022-09-15 13:45:42

NeurIPS workshop that explicitly requests submissions from real-world applications including conservation!

https://twitter.com/yoonholeee/status/1570461511797338113

twitter

Yoonho Lee (https://twitter.com/yoonholeee/status/1570461511797338113)

We're organizing the second Workshop on Distribution Shifts (DistShift) at #NeurIPS2022, which will bring together researchers and practitioners. Submission deadline: Oct 3 (AoE) Workshop date: Dec 3 Website: <https://sites.google.com/view/distshift2022>

Original URL: https://twitter.com/yoonholeee/status/1570461511797338113

❤️ Justin Kay, Oisin Mac Aodha, Mark Goldwater, Suzanne Stathatos, Alan Papalia, Lucia Gordon, Lukas Picek, Eddie Zhang, Jaanak, Dhruv Sheth, Robin Zbinden, Gedeon, Omiros Pantazis, Agnethe Seim Olsen, Fadel Seydou, Ando Shah, Rita Pucci

😎 Jon Van Oast, Dhruv Sheth

Sara Beery (sara.beery@gmail.com)

2022-09-15 13:45:58

"This workshop aims to convene a diverse set of domain experts and methods-oriented researchers working on distribution shifts. We are broadly interested in methods, evaluations and benchmarks, and theory for distribution shifts, and we are especially interested in work on distribution shifts that arise naturally in real-world application contexts. Examples of relevant topics include, but are not limited to:

Examples of real-world distribution shifts in various application areas. We especially welcome applications that are not widely discussed in the ML research community, e.g., education, sustainable development, and conservation. We encourage submissions that characterize distribution shifts and their effects in real-world applications; it is not at all necessary to propose a solution that is algorithmically novel.

Sara Beery (sara.beery@gmail.com)

2022-09-15 13:46:33

"Benchmarks and evaluations. We especially welcome contributions for subpopulation shifts, as they are underrepresented in current ML benchmarks. We are also interested in evaluation protocols that move beyond the standard assumption of fixed training and test splits -- for which applications would we need to consider other forms of shifts, such as streams of continually-changing data or feedback loops between models and data?

Kamran Zolfonoon (kzolfonoon@umass.edu)

2022-09-15 16:31:14

Hi everyone! I’m Kamran a MS-CS student at the University of Massachusetts, Amherst. I’ll be working with the US Fish & Wildlife Service this semester to build a computer vision pipeline that collects data from Bald Eagle nest cameras. The hope is to use this data to quantify breeding adult Bald Eagle nest attendance patterns and provisioning behavior.

Currently working on turning unlabeled images into a dataset. If anyone has experience using cloud labeling services with trap camera (or similar) images I’d love to connect and learn from your experience!

👋 Suzanne Stathatos, Sara Beery, Roni Choudhury, Declan, Dan Morris, Omiros Pantazis, Jaanak, Rita Pucci

Mel Guo (melguo236@gmail.com)

2022-09-21 15:37:36

Hey Kamran! Thanks for your intro, your work with USFWS sounds so interesting! Is there a webpage for your research project on data collection from bald eagle nest cameras? Super interested in learning more as a fellow MS-CS student interested in birds!

Vinicius Amaral (amaralvin7@gmail.com)

2022-09-16 00:51:25

Hi all! I'm a PhD student/oceanographer at the University of California Santa Cruz and I use CV to classify images of particles in the ocean. I'm especially interested in topics such as class imbalance and domain shift. Happy to be here and learn from you all.

👋 Suzanne Stathatos, Jon Van Oast, Robin Zbinden, Sara Beery, Déva Sou, Lucia Gordon, Roni Choudhury, Declan, Dan Morris, Omiros Pantazis, Jaanak, Rita Pucci

Rebecca (rebeccayap92@uchicago.edu)

2022-09-16 08:44:00

Hi all! I thought I would introduce myself. I’m a graduate student at the University of Chicago Harris School of Public Policy. I’m born and bred in Singapore. I am highly interested in conservation and have a running idea that I think will run a series of projects. Very excited to meet and interact with all of you at some point!

👋 Sara Beery, Roni Choudhury, Declan, Dan Morris, Suzanne Stathatos, Omiros Pantazis, Jaanak, Rita Pucci

Roni Choudhury (roni.choudhury@kitware.com)

2022-09-16 10:03:30

fellow u of c graduate here 👋

👋 Rebecca

Ethan Shafron (ethan.shafron@gmail.com)

2022-09-16 10:05:07

Hi everyone, I figured I would introduce myself as well - I'm a graduate student and staff research specialist at University of Montana, where I work for the Spatial Analysis Lab and am a student in the Global Climate and Ecology Lab. My background is in imaging spectroscopy, computational ecology, and remote sensing, and my graduate work is focused on understanding where, why, and when Carbon allocation towards growth in trees becomes decoupled from primary productivity. This is a bit outside the realm of Computer Vision, but I think there are a lot of leverage points that computer vision could help with - there's still quite a bit of mechanistic uncertainty in the terrestrial carbon cycle, and a ton of that is driven by bottlenecks in data acquisition and processing of organism-scale data (think soils, wood, leaves, roots). How can we close that gap?! Maybe CV is part of it? Maybe not?

I'd be curious to touch base with any other folks working more in the vegetation/biogeochemistry/paleoclimate/remote sensing space - I think the use cases of CV in this space are a bit more narrow, but could go a long way in improving our understanding of biogeochemical processes as the climate continues to change.

👋 Roni Choudhury, Sara Beery, Justin Kay, Dan Morris, Carly Batist, Rowan Converse, Suzanne Stathatos, Benjamin Kellenberger, Declan, Jason Holmberg (Wild Me), Omiros Pantazis, Jaanak, Adam Noach, Heather, Lauren E. Gillespie, Rita Pucci, Antón Álvarez

👍 Roni Choudhury, Sara Beery, Carly Batist, Jason Holmberg (Wild Me)

Declan (declan.pizzino@consbio.org)

2022-09-16 15:59:59

While CV seems to be dominant in this slack, there is plenty of space for using other AI solutions for conservation applications! I'm part of a small, primarily RS-focused, team of geospatial analysts at a sm/med conservation non-profit. Happy to connect with you to chat about the work we do at CBI

🎉 Carl Boettiger, Sara Beery

Rebecca (rebeccayap92@uchicago.edu)

2022-09-16 16:43:58

I’m also beginning to see a lot of applications that ML (and I guess CV) can do in ecology, and hoping to learn more - about both ecology and ML in tandem (I consider myself a newbie in both).

Heather (h_peacock@ducks.ca)

2022-09-19 13:43:44

Hi @Ethan Shafron, I am also interested in using ML and RS for carbon related questions (and the ecological and climate change implications), mostly storage atm, would be interested to hear more about your research and if you have any thoughts on modelling (quantifying) carbon storage in terrestrial ecosystems. Thanks!

Ethan Shafron (ethan.shafron@gmail.com)

2022-09-23 13:02:02

Hi @Declan - I've actually heard a fair bit about CBI! My partner works for the California Native Plant Society, and specifically works with EEMS - would love to hear about what you all are doing on the RS side of things though. Ofc there are plenty of non-CV ML solutions, I'm just off the heels of Sara's CV for ecology workshop, so that's what's been occupying my mind these days :)

@Rebecca - welcome! It's always cool to see "newbies" in spaces like these - if you're here then you're probably curious about a lot of different things, which is basically what drives this whole sub-field!

@Heather - There's a lot of stuff on Carbon modelling these days and it's hard to sift through it all to find what's relevant at varying spatial/temporal scales. Happy to chat about C fluxes and storage and hear more about what you're working on!

😍 Declan

Rebecca (rebeccayap92@uchicago.edu)

2022-09-30 11:19:59

Hi Ethan! I am very curious about this side of things. Do you think I would be able to PM to chat more with you? I have a few questions about how to grow and position myself to be on the conservation side of things, and would love to chat about an idea that has been brewing.

Ethan Shafron (ethan.shafron@gmail.com)

2022-09-30 11:35:39

absolutely!

Erik Peterson (omahaesp@gmail.com)

2022-09-16 22:17:53

Hello all — bit different here from a lot of the academic perspectives. I'm a product manager, former software engineer, in big enterprise B2B, but I do a lot of day to day data analysis and increasingly ML. Looking to help with any open source positioning or some entrepreneurial approaches to private sector stuff

👍 Gedeon, Dan Morris, Omiros Pantazis, Sara Beery

👋 Suzanne Stathatos, Jaanak, Adam Noach, Rita Pucci

Gedeon (gedeonmuhawenayo@gmail.com)

2022-09-17 06:49:18

Hi all, I am a Machine Learning Research Engineer at Rwanda Space Agency. I use CV & Geospatial data for conservation. Happy to be here.

👋 Gabriel Tseng, Dan Morris, Omiros Pantazis, Lucia Gordon, Sara Beery, Jason Holmberg (Wild Me), Declan, Suzanne Stathatos, Jaanak, nyakundi lamech, Adam Noach, Gedeon, Rita Pucci

Josh Seltzer (jyseltz@gmail.com)

2022-09-17 12:46:02

Hi everyone! It's nice to find this community and to see a few familiar people here 👋

To quickly introduce myself -- I'm passionate about providing technical solutions to help empower conservation projects with local and indigenous knowledge, and I am super interested in combining exciting advances in AI (especially NLP / computer vision / machine listening) with a variety of hardware systems that can be used for ecological research and biodiversity conservation. Besides biodiversity, I also see a lot of opportunities for AI for language conservation (but have been struggling to find other people interested in that space).

I also run a tech company (nexxt.in) and am living in Panama at the moment (if anyone is nearby feel free to reach out 😉), and might be slightly obsessed with all things monkey related 🐒 if anyone wants to chat about that!

👋 Sara Beery, Jason Holmberg (Wild Me), Declan, Suzanne Stathatos, Jaagat P., Jaanak, Adam Noach, Dan Morris, Heather, Rita Pucci

🙌 Gedeon

👋:skin_tone_3: Pen-Yuan Hsing

Heather (h_peacock@ducks.ca)

2022-09-19 13:40:04

Hi Josh! Fellow monkey enthusiast here! I'm finishing my PhD on global primate biogeography and conservation. I suppose the first question is, what is your favourite monkey?

Josh Seltzer (jyseltz@gmail.com)

2022-09-20 08:53:23

Hey @Heather, fellow Canadian as well i suppose (guessing from your email)? And wow that sounds really cool!! I would love to hear more about your studies 🙂 and I love squirrel monkeys because of how mischievous they are for sure haha, but I am fascinated by most of the platyrrhines

Ashley Chang (she/her) (ash.chang0921@gmail.com)

2022-09-17 22:54:28

Hello! I am Ashley, and I am a senior at California High School (San Ramon, CA). I love anything Data Science related, especially ML and AI! I am very passionate about many environmental and social justice topics, in which I utilize ML and AI to help identify and potentially find solutions for them. (I am a huge advocate for problem-solving)! Furthermore, I joined this Slack due to its diverse community in AI! :)

Although I am not in college yet, I plan to pursue research once I get there! I love learning and growing, and I love to teach others. Aside from academics and any STEM-related passions, I love to write, go on hikes, and watch any cyberpunk-related shows!

I currently hold an internship with NASA, in which I utilize CVAT (Computer Vision Annotation Tool) to identify biofilm in caves. Then, the identifications (i.e. “annotations”) are integrated into a ML and AI algorithm to ultimately be used on the NeBula robot. Hopefully in the distant future, the NeBula robot will be sent to Mars to potentially identify life in Martian volcanic caves!

I am also the president of the Data Science club at my school, in which I currently introduce AI and ML to my peers. I would love to expand my knowledge in AI onto projects, so if anyone is interested in talking about AI, ML, or something completely unrelated, I would be more than happy to chat about it! Super excited to be a part of this community. I hope you all have a great day. :)

👋 Sara Beery, Suzanne Stathatos, Eddie Zhang, Adam Noach, nyakundi lamech, Josh Seltzer, Dan Morris, Jason Holmberg (Wild Me), Heather, Rita Pucci

😎 Jon Van Oast, Jason Holmberg (Wild Me)

👋:skin_tone_3: Pen-Yuan Hsing

Josh Seltzer (jyseltz@gmail.com)

2022-09-18 09:56:08

Super cool that you've done so much while still in high school, especially the NASA project that sounds amazing!

❤️ Ashley Chang (she/her)

Valentin Lucet (valentin.lucet@gmail.com)

2022-09-18 17:11:22

Yes it is extremely impressive!

❤️ Ashley Chang (she/her)

Adam Noach (amn55@cam.ac.uk)

2022-09-18 06:02:46

Hi everyone! In two weeks I’ll be starting a PhD in computational ecology at the University of Cambridge. The project is focused on peat-forming wet woodlands - will start by pitching new plots, estimating C stocks using some combination of drone ALS+photogrammetry, TLS and other ground measurements, and then looking at modeling the C cycle/ecosystem dynamics more generally + potential for carbon storage. There’s also scope for looking at how these ecosystems respond to sea-level rise and their flood mitigating potential using existing satellite data.

I’m largely here thanks to having come across the ID Trees project and @Ben Weinstein ’s work on tree crown segmentation (thank you!) - which I still find very interesting - and which inspired my master’s project last year. I’m delighted to be part of this community and especially keen on finding out what any other forest ecologists in here are up to!

🙌 Josh Seltzer, Ashley Chang (she/her), Dan Morris, Sara Beery, Eddie Zhang, Suzanne Stathatos, Rita Pucci

Rebecca (rebeccayap92@uchicago.edu)

2022-09-18 09:01:30

Very interesting research! I also looked at the ID Trees project - super intriguing to me since I want to look into trees and forests!

🌳 Adam Noach

Rebecca (rebeccayap92@uchicago.edu)

2022-09-18 09:02:37

I would love to read your master’s project if you are willing to share, @Adam Noach!

Adam Noach (amn55@cam.ac.uk)

2022-09-18 12:13:02

I'm in that awkward middle ground of extending it into a paper - but when it's cooked up ( & if it's appropriate and welcome) I'd be happy to share it here. I can see it's already been mentioned but DeepForest will be up your alley if ID Trees was!

Rebecca (rebeccayap92@uchicago.edu)

2022-09-18 20:57:59

Got it! i will look at that.

👍 Adam Noach

Rebecca (rebeccayap92@uchicago.edu)

2022-09-18 20:58:24

Hmm when I have mastered Python in entirety, I will play around with this package.

Cathy Atkinson (cathy.atkinson@highlandsrewilding.co.uk)

2022-09-19 05:24:02

Hi all. I'm Cathy from the UK. I work for Highlands Rewilding where we are using all sorts of remote sensing, camera traps, acoustic sensors etc to monitor biodiversity and carbon storage (forestry and peat restoration) on our rewilding sites in Scotland. 🌳🌲🦌🐦🦊🐁🐗🌱

🌳 Tjomme Dooper, Rebecca, Dan Morris, Lucia Gordon, Marconi Campos, Sara Beery, Ashley Chang (she/her), Adam Noach

👋 Suzanne Stathatos, Rita Pucci, Declan

👋:skin_tone_5: Ando Shah

Heather (h_peacock@ducks.ca)

2022-09-19 13:37:38

Hi! I'm Heather from Canada, I am finishing my PhD in Geography, using GIS to map and quantify global primate habitat loss and fragmentation. I also work at DUC as a GIS specialist. I am very interested in ML models for conservation, and models for regional scale landscape ecological research. Hope to learn a lot from this community!

👏 Dan Morris, Sara Beery, Suzanne Stathatos, Declan, Ashley Chang (she/her), Lauren E. Gillespie, Olivier Gimenez, Adam Noach, Rita Pucci

Declan (declan.pizzino@consbio.org)

2022-09-19 14:08:42

Hi Heather! Glad to see you here 😄

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-19 14:09:31

Hi everyone. Can we do a big round up on current drone's that teams are using. We are running out of time to update from our DJI's to get up to date with the federal drone list: https://www.diu.mil/blue-uas-cleared-list. A general desire is

Quadcopter for heavy wind conditions landing on a boat (We have fun videos of staff diving into the alligator heavy everglades to recover vertical take off drones)

Under 30k

PPK accuracy for geotagging each image

On the US federal blue list

We own several DJI Inspires and a Wingtra One (Gen I).

diu.mil

DIU

A quick reference list of all Blue UAS

Original URL: https://www.diu.mil/blue-uas-cleared-list

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2022-09-20 04:33:28

Hey ben - this might be a question to also drop in to the WILDLABS dicusssions and tag to our drones group so you can get feedback from the wider conservation tech community. https://wildlabs.net/

wildlabs.net

Buy CORE™ S-4K No Glow Trail Camera and More | Bushnell

WILDLABS is home to the global conservation technology community of 6,000 people in 120 countries discussing 1,200 topics like biologging, camera traps, and machine learning. With engaging spaces to ask questions and collaborate together, share your own work, and discover new ideas and innovations, WILDLABS is your platform to connect with #Tech4Wildlife experts and projects from around the world.

Original URL: https://wildlabs.net/

Pietro Perona (perona@caltech.edu)

2022-09-20 11:09:07

thanks!

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2022-09-20 12:59:22

We have great nightime pictures of quick animals with bushnell trophy no glow. Seems the new version is even better than the ones we bought in 2020 https://www.bushnell.com/trail-cameras/standard-trail-cameras/core-s-4k-no-glow-trail-camera/PB-119949C.html

Bushnell

Shopping for the CORE™ S-4K No Glow Trail Camera - Learn more about the latest Standard Trail Cameras and other hunting or shooting gear at Bushnell.

Original URL: https://www.bushnell.com/trail-cameras/standard-trail-cameras/core-s-4k-no-glow-trail-camera/PB-119949C.html

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2022-09-20 12:59:36

Looking forward to those NE italian night shots!

Pietro Perona (perona@caltech.edu)

2022-09-21 10:48:45

Thank you Tiziana!

Majid Mirmehdi (m.mirmehdi@bristol.ac.uk)

2022-09-21 15:06:48

Hi
It might be helpful if there was a dedicated channel for announcing datasets?

✔️ Jon Van Oast, Sara Beery, Carl Boettiger

Majid Mirmehdi (m.mirmehdi@bristol.ac.uk)

2022-09-21 18:48:38

It could be under new papers…it’s just that not all new papers will introduce new datasets. Would be much easier to have a channel with only links to datasets old and new.

Sara Beery (sara.beery@gmail.com)

2022-09-22 15:10:50

You're free to make any channels you think are useful!

👍 Majid Mirmehdi

Olof Mogren (olof.mogren@ri.se)

2022-09-22 11:05:37

Hi everyone! I'm a researcher at RISE, the Swedish research institute. I have a PhD in computer science, and am now heading the deep learning research part of our center for applied AI. We have some exciting projects related to biodiversity and modelling of earth and water. I was just invited to this Slack by @Devis Tuia, whom I just had the pleasure to have invited to our seminar series, RISE Learning Machines, and who just gave a great talk there. We are looking for master's students for a couple of projects this spring, including Deep learning for detection of coffee berry disease and Aerial View Goal Localization with Reinforcement Learning. Great to be here!

RISE

Learning Machines Seminars – Upcoming Seminars | RISE

Here you find all the upcoming Learning Machines Seminars. Update your calendar with upcoming seminar dates of your interest!

Original URL: https://ri.se/lm-sem

RISE

Master thesis proposal: Deep learning for detection of coffee berry disease | RISE

Climate change brings along problems for farmers around the world. Assisting agriculture using AI to make farming more resilient may become an important step towards adapting to the new normal. Limiting the uncertainties and risks in food production can substantially improve the situation for many farmers, and increases the availability and robustness of food supplies. Deep learning is one of the most important tools in the AI toolbox, and has shown impressive results on many tasks on data of different modalities, not the least for computer vision.

Original URL: https://www.ri.se/en/work-with-us/open-job-positions/master-thesis-proposal-deep-learning-for-detection-of-coffee-berry

RISE

Master thesis: Aerial View Goal Localization with Reinforcement Learning | RISE

Master thesis at RISE Research Institutes of Sweden (spring 2023): Aerial View Goal Localization with Reinforcement Learning – Towards Emulating Search-and-Rescue

Original URL: https://www.ri.se/en/work-with-us/open-job-positions/master-thesis-aerial-view-goal-localization-with-reinforcement

👀 Stephanie O'Donnell, Aleksis Pirinen

🙌 Stephanie O'Donnell, Devis Tuia, Adam Noach, Edvin Listo Zec, Dan Morris, Robin Zbinden, Josh Seltzer, Suzanne Stathatos, Sara Beery, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Aleksis Pirinen, Malte Pedersen, Jeff Reed

😎 Jon Van Oast, Sara Beery, Edvin Listo Zec

🙏 Michael Bunsen, Sara Beery

Abhinav Sharma (abhinav.sharma@iiitg.ac.in)

2022-09-22 13:37:32

Dear All, I am a senior-year undergraduate student in the Computer Science and Engineering Department at the Indian Institute of Information Technology,Guwahati. In the past I have research experience of working at Georgia Institute of Technology on Graph Neural Networks and Approximate Computing and recently I have worked on Reinforcement Learning and Causal Inference for design space exploration in Facebook's Augmented Reality System Investigator. Looking forward to contributing in the interdisciplinary fields. Feel free to connect with me on LinkedIn.

👋 Suzanne Stathatos, Sara Beery, Carly Batist, Chinmay Talegaonkar, Jason Holmberg (Wild Me), Stephanie O'Donnell, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Adam Noach, Jeff Reed

Dan Morris (agentmorris@gmail.com)

2022-09-22 17:09:35

Really neat to see Esri releasing a bunch of pretrained models relevant to this Slack, including models based on data that folks on this Slack have helped make available:

https://www.esri.com/arcgis-blog/products/arcgis/imagery/new-pretrained-deep-learning-models-sept-2022/

That includes a tree detection model (based on @Ben Weinstein's NEON tree dataset), a bird detection model (based on @Benjamin Kellenberger's Aerial Seabirds West Africa dataset), an elephant detection model (based on the Aerial Elephant Dataset), and a bunch of other interesting ones that aren't quite so on-the-nose conservation-focused, but are relevant and interesting.

ArcGIS Blog

New Pretrained Deep Learning Models (September 2022)

Learn about our recently released pretrained deep learning models available in ArcGIS Living Atlas of the World.

Original URL: https://www.esri.com/arcgis-blog/products/arcgis/imagery/new-pretrained-deep-learning-models-sept-2022/

🙌 Stephanie O'Donnell, Riccardo de Lutio, Stefan Schneider, Catherine Villeneuve, Justin Kay, Sara Beery, Carly Batist, Jason Holmberg (Wild Me), Suzanne Stathatos, Jake Wall, Kasirat, Josh Seltzer, Fadel Seydou, Jeff Reed

😎 Jon Van Oast, Rowan Converse, Sara Beery, Jason Holmberg (Wild Me)

👍 Monty Ammar

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-22 22:41:41

this is cool, but also kind of terrifying, there is alot of improvement to be made on deepforest and a company with ESRI's capability could really make it better. I get nervous when it gets just passed around as a 'tree detector' without all the docs associated with deepforest about all the time it doesn't work/could be better.

🤔 Josh Seltzer

💯 Declan, Emily Lines

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-22 23:00:13

Does anyone know the current url (@Sara Beery, @Pietro Perona) for the pasadena tree dataset at visipedia? Pasadena Urban Trees
"This includes dense aerial and street view imagery for 30,000 trees labeled with geo-location and tree species from Pasadena, California."
[URL] [Paper] https://visipedia.github.io/datasets.html

Visipedia

Datasets

Visipedia Webpage

Original URL: https://visipedia.github.io/datasets.html

Sara Beery (sara.beery@gmail.com)

2022-09-22 23:23:37

I think @Pietro Perona is the best bet here

Pietro Perona (perona@caltech.edu)

2022-10-17 04:33:28

We are looking for the data - usual problem w switching servers. Please also contact Jan Wegner at ETH

👍 Ben Weinstein

Brad Pickens (bradley_pickens@fws.gov)

2022-09-23 09:57:44

I'm curious if there is interest in this ESA call, as it could be bring together both fish and wildlife experts in deep learning: Ecological Society of America invites proposals for Symposia, Organized Oral Sessions to be hosted in Portland, OR, Aug 6-11, 2023, https://www.esa.org/portland2023/session-types/submit-a-proposal-for-an-invited-pa[…]e=Informz&utm_medium=Email&utm_campaign=ESA&_zs=NJIml&_zl=QGNJ2

😎 Jason Holmberg (Wild Me), Sara Beery

Dan Morris (agentmorris@gmail.com)

2022-09-23 11:18:30

Oh I'm super-interested. I had no idea ESA was in Portland next year. My dream is to get all the benefits of in-person conferences without actually going anywhere, which I'll roughly define as "without leaving Seahawks territory", and you're telling me I can have TWS in Spokane followed by ESA in Portland, and meet lots of conservation folks, and never really have to go anywhere? Sign me up!

👍 Kakani Katija, Michael Bunsen, Sara Beery

❤️ Michael Procko, Sara Beery, Carl Boettiger

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-23 13:43:09

I live in Portland and will submitting a proposal, I welcome everyone to join me. I'll post it here. I'll probably host a dinner too.

👍 Sara Beery, Casey Youngflesh, Carl Boettiger

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-23 13:44:18

Something like, "Automated ecological monitoring using Computer Vision: Where are we going? How do we get there?"

Brad Pickens (bradley_pickens@fws.gov)

2022-09-23 15:17:36

Thanks Dan and Ben! I wonder if we could get a couple of sessions proposed?! In addition to Ben's broad topic area, maybe something on Application of Deep Learning in Wildlife and Fisheries? Organized oral sessions are 6 talks/Symposia are 4 talks (longer; broad topics)/Inspire Sessions are 6-10 talks. Ben- what category are you thinking? ...Conference theme is "for all ecologists" - meaning both academics and us applied folks 😀

👍 Yseult Hb

Michael Bunsen (notbot@gmail.com)

2022-09-23 15:28:34

I am also in Portland and would love to attend if not present as well. I could contribute some slides on the state of automated insect monitoring, or perhaps do a shorter "Inspire Session" on that topic. I would love to rally support for getting a network of insect monitoring stations setup in the USA, or at least up and down the Pacific Northwest

Emily Lines (erl27@cam.ac.uk)

2022-09-23 17:24:35

@Ben Weinstein that sounds like a fantastic proposal - I'd be interested!

Sara Beery (sara.beery@gmail.com)

2022-09-24 08:37:34

I'm also definitely interested! I've never been to an in-person ESA, and I share @Dan Morris's enthusiasm for meeting lots of ecologists and conservationists without needing to go too far!

Casey Youngflesh (caseyyoungflesh@gmail.com)

2022-09-26 11:17:46

Yes to all of this! @Ben Weinstein I’d be game

Atul Ingle (ingle@uwalumni.com)

2022-09-26 12:29:10

For those planning to attend, I'd also like to invite you to visit my new computational imaging lab at Portland State University in downtown Portland!

(One of my research thrusts is in high speed/low light image sensing under resource constrained scenarios. I believe this can have applications in ecological monitoring esp situations where conventional vision sensors+algorithms fail.)

👀 Emily Lines, Ethan Shafron

❤️ Sara Beery, Carl Boettiger, Barbie D

🙌 Michael Bunsen

Brad Pickens (bradley_pickens@fws.gov)

2022-09-30 10:42:14

For background, it is the largest ecology meeting in the US -- precovid- upwards of 4,000 people regularly attend. ESA is also one of the more high-tech ecology conferences and a good place to see the next big ideas in ecology as a whole (overarching plants, insects, wildlife, fisheries, landscape ecology, land cover mapping). Some of these topics can be rare at other conferences. Looks like we should have interest for at least 1-2 sessions...

👍 Michael Bunsen

Ben Weinstein (benweinstein2010@gmail.com)

2023-01-23 20:14:38

Our session (@Dan Morris, @Sara Beery, @Emily Lines, @Tessa Rhinehart, @Kakani Katija), "The future of ecological monitoring is collaboration with Artificial Intelligence", was accepted.

🙌 Sara Beery, Michael Bunsen, Kakani Katija, Emily Lines, Tessa Rhinehart

Michael Bunsen (notbot@gmail.com)

2023-01-23 20:21:03

Oh fantastic! I am part of another session proposal focused on insect monitoring. Still waiting to hear back, but I look forward to the event regardless. Perhaps an AI for Conservation happy hour is in order?

❤️ Tessa Rhinehart

👍 Casey Youngflesh, Kakani Katija

Eric Colson (ecolson@gmail.com)

2022-09-26 11:45:12

Hi All, so happy to be joining this group. I am a long-time leader of data science teams in Industry (Netflix, Stitch Fix, etc). Looking to see where i can help wrt to environment/conservation.

👋 Josh Seltzer, Stephanie O'Donnell, Declan, Dan Morris, Felipe Parodi, Carly Batist, Lucia Gordon, Ben Weinstein, Jason Holmberg (Wild Me), Marconi Campos, Jaanak, Abhay, Peter Bull, Casey Youngflesh, Andrew Schulz, Adam Noach, Sara Beery, Carl Boettiger, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

🎉 Jon Van Oast, Jason Holmberg (Wild Me), Olivier Gimenez, Marconi Campos, Peter Bull, Sara Beery

Abhay (abhaykash12@gmail.com)

2022-09-26 17:07:21

Hi Eric! Nice to see you here! (I was a DS at Sfix)

👋 Eric Colson

Carl Boettiger (cboettig@berkeley.edu)

2022-09-27 22:34:47

Hey Eric, great to see you here!

👋 Eric Colson

Abhay (abhaykash12@gmail.com)

2022-09-26 17:21:42

And since Eric is here, I must say that a lot of my thinking around making life easy for downstream users by building horizontals is shaped by what he and his team had set up at Sfix 🙂

🙌 Eric Colson

Kakani Katija (kakani@mbari.org)

2022-09-27 15:58:08

Excited to share with the community our first peer-reviewed FathomNet publication in Scientific Reports: https://t.co/G1gLUdNDde . FathomNet is a multi-institutional multi-individual effort to build a global labeled image database for underwater life (and other objects). Seeded with data from MBARI, NOAA, and National Geographic Society, we look forward to data contributions and expertise from a broad user community. Check it out, share it with your networks. FathomNet can be accessed at www.FathomNet.org, follow us on Twitter @fathomnet, and we have a python API at www.GitHub.com/FathomNet . Looking forward to receiving your feedback!

Nature

FathomNet: A global image database for enabling artificial intelligence in the ocean

Scientific Reports - FathomNet: A global image database for enabling artificial intelligence in the ocean

Original URL: https://www.nature.com/articles/s41598-022-19939-2

🎉 Jon Van Oast, Peter Bull, Ben Weinstein, Michael Procko, Carly Batist, Oisin Mac Aodha, Abhay, Dan Morris, Avi Sundaresan, Subhransu Maji, Marconi Campos, Josh Seltzer, Andrew Schulz, Carl Boettiger, Devis Tuia, John Martinsson, Riccardo de Lutio, Lukas Picek, Viktor Domazetoski, Timm Haucke, Yseult Hb, Toryn Schafer, Eddie Zhang, Rose Hendrix, Justine Boulent

👀 Michael Procko, Sara Beery, Nico Lang, Aleksis Pirinen, Rita Pucci

🦀 Peter Bull, Sara Beery, Carl Boettiger, Aleksis Pirinen, Lukas Picek, Levi Cai

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-27 16:07:42

does fathom net come with a detector in the python client. Like if I have an image can I go

import fathomnet
m = fathomnet.model()
predictions = m.predict_image("my_image")
predictions["class"]

Kakani Katija (kakani@mbari.org)

2022-09-27 16:16:10

No it doesn't. You can check out our model zoo at https://github.com/fathomnet/models and more details on the python api are at https://github.com/fathomnet/fathomnet-py

Daniel Davila (daniel.davila@kitware.com)

2022-09-27 18:01:31

Congrats! Super exciting

🎉 Kakani Katija

Dan Morris (agentmorris@gmail.com)

2022-09-27 20:07:34

The model zoo is very cool! I have no particular reason to want to detect benthic organisms, but who am I to turn away a well-documented model zoo? So FWIW I was able to download and run the "MBARI Monterey Bay Benthic" model, it was very straightforward.

👍 Kakani Katija

Kakani Katija (kakani@mbari.org)

2022-09-27 22:02:09

Awesome. Would love to get your input @Dan Morris on how we can make it better (e.g, standardization, etc).

Dan Morris (agentmorris@gmail.com)

2022-09-27 22:06:55

My only off-the-cuff recommendation would be to include images on the model zoo README that conveys "this is the gestalt of the training data for each model"... that makes it a lot easier for a new user to judge whether the angle/color/murkiness of their images is close enough to the domain of the model to expect reasonable results, and IMO no amount of verbal description can capture that. But once a user decides "yes, this model is reasonably close to my data", the models look well-documented.

👍 Kakani Katija, Carl Boettiger, Justin Kay

Kakani Katija (kakani@mbari.org)

2022-09-27 22:19:34

Yeah, we’ll have to think carefully on that. Totally agree.

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-27 23:03:32

how close are we to higher order taxonomy detection for these type of images. Fish/Jellyfish/Octopus/Crab. How deep the taxonomy can we go?

Ben Weinstein (benweinstein2010@gmail.com)

2022-09-27 23:04:59

I am imagining that if there was some live underwater feed and we wanted to partition the images to the relevant annotators/experts. When you see a seastar send it to seastardude@gmail.com, etc.

Devis Tuia (devis.tuia@epfl.ch)

2022-09-28 03:03:31

This is really cool! We are quite busy with underwater imagery here at ECEO, I will definitely share that with my students!

👍 Kakani Katija

Kakani Katija (kakani@mbari.org)

2022-09-28 06:10:20

@Ben Weinstein, check out the model zoo and the MBARI benthic object detector. We're already pretty close for high-level morphological classes (of course could be better); doing the same thing for midwater animals now too. The FathomNet taxonomy goes to species. Yep totally! The community mods (eg notifications, etc) are currently underway for FathomNet. Should be released by Spring. 🤞

Levi Cai (levi@levicai.com)

2022-09-28 13:57:25

I've been building some higher order marine organism detectors that we're using on AUVs in the field that are "working" to some degree, partially using some of the fathomnet dataset. Happy to share/chat.

❤️ Justin Kay

Kakani Katija (kakani@mbari.org)

2022-09-29 10:39:38

Great! @Levi Cai, you should contribute your models to the FathomNet Model Zoo!!

fathomnet/models

A collection of machine learning models trained on FathomNet

Stars

Last updated

2 months ago

❤️ Justin Kay

Kangyu Zheng (zkysfls@gmail.com)

2022-09-28 21:23:19

Hi, I want to know are there any recordings for these talks?

Sara Beery (sara.beery@gmail.com)

2022-09-29 08:40:04

@Tilo Burghardt?

Tilo Burghardt (tb2935@bristol.ac.uk)

2022-09-29 14:16:45

Dear @Kangyu Zheng, all workshop slides from all speakers will be up at the weekend on the website. We did not record this 1st workshop to avoid any GDPR issues, keep it very low key and allow for just a single public link for people to join. The next edition of the workshop will provide recordings and hopefully full proceedings. Best, Tilo

👍 Gedeon

Kangyu Zheng (zkysfls@gmail.com)

2022-09-29 14:47:49

Thanks for the reply!

Luke Sheneman (sheneman@uidaho.edu)

2022-09-29 10:56:36

For anybody interested, I've developed a working edge-AI prototype that can detect and classify species (25 classes) and transmit encoded results via an Iridium satellite modem to our servers at the University of Idaho. It's based on an NVIDIA Jetson Nano and as part of its workflow runs both MegaDetector v5 and a bespoke model I trained (using YOLOv5m) for doing species-level classification for our taxa of interest. Results so far are super encouraging. Will be sharing code, design, and pre-trained model(s) soon. Works off Li-Ion battery and solar, so hopefully can be deployed in remote areas indefinitely. Field tests are about to start. Many thanks to @Dan Morris and Idaho Fish and Game for pointers and training data!

😍 Suzanne Stathatos, Subhransu Maji, Josh Seltzer, Sara Beery, Luke Sheneman, Carly Batist, Jason Parham, Valentin Lucet, Alan Papalia, Jason Holmberg (Wild Me), Eric Colson, Timm Haucke, Talia Speaker, Jaanak, Elizabeth Bondi, Mitch Fennell, Eddie Zhang, Jeff Reed, Emilio Luz-Ricca, Ando Shah, Nick Giampietro, Carl Boettiger, Oscar Schafer

💯 Eelke, Jason Parham, Swayam Thakkar, Aleksis Pirinen, Timm Haucke

👍 Justin Kay, Atul Ingle, Jason Parham, Rowan Converse, Dan Morris, Leonardo Viotti, Aleksis Pirinen, Timm Haucke, Gedeon, Howard Windsor, Prabath Gunawardane

😎 Jon Van Oast, Timm Haucke, Carl Boettiger

🙌 Valentin Gabeff, Timm Haucke, Ștefan Istrate, Yseult Hb, Kai Waddington, Jacob Kamminga, Nick Giampietro

👀 Sean P. Rogers

Sara Beery (sara.beery@gmail.com)

2022-09-29 11:04:48

This is so cool 😎

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-09-29 11:17:46

Out of curiosity, what species does the model cover? Also, do you know how quick the on-board processing is? Like how many seconds/image type thing

Luke Sheneman (sheneman@uidaho.edu)

2022-09-29 11:33:52

@Carly Batist we are training the classifier on deer, elk, moose, domestic livestock (cattle, sheep), wolves, coyotes, mountain lion, bear, rabbits, humans, vehicles, and a smattering of other things (domestic dogs, badgers, bobcat, wild turkey, etc). For inference, once the models are loaded and initialized it takes just under 1 second per image on the Jetson Nano to run through MegaDetector and a little less than that to run through the classifier model. So maybe inference at 1.5 seconds per image. We are seeing a full cycle of wake up, pull images off trail cam, detect/classify, transmit via satellite, sleep typically takes about 5 mins.

👍 Carly Batist, Jason Parham, Rowan Converse, Valentin Lucet, Timm Haucke

Michael Bunsen (notbot@gmail.com)

2022-09-29 12:32:31

This is fantastic. Congratulations on making an end-to-end system, and with a satellite connection! I look forward to looking at the design and code when you are ready, and I would be happy to test them. What did you use for the camera?

Luke Sheneman (sheneman@uidaho.edu)

2022-09-29 13:00:48

Thanks @Michael Bunsen. It currently works with a Moultrie M-50i camera. As built, it could work easily with other cameras that support automated "USB Mode" where it presents as a USB drive when it detects power on its underside USB port.

Dan Morris (agentmorris@gmail.com)

2022-09-29 13:17:22

Very cool! There are a number of related efforts (e.g. the Conservation X Labs work announced on another thread), but AFAIK all the other work like this is targeting a very general-purpose device, which will definitely be necessary to shift the field in this direction, but I think we'll also need some successes where projects like this make something work end-to-end for the team that's building it, starting in a small (ideally n=1) number of ecosystems, where you can really kick the tires on your own end-to-end process, and inform all of the work on general-purpose devices. Excited to hear how field tests go!

👍 Luke Sheneman, Jason Holmberg (Wild Me), Valentin Lucet, Carly Batist, Sara Beery

➕ Sara Beery

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2022-09-29 14:00:58

Awesome work!

Timm Haucke (timm@haucke.xyz)

2022-09-29 16:40:47

@Luke Sheneman looks super interesting! We are also using the Jetson Nano for a recent camera trap prototype and have some experience with optimizing power consumption & MegaDetector models, so let me know if I can be of any help!

❤️ Sara Beery, Luke Sheneman, Carly Batist

Luke Sheneman (sheneman@uidaho.edu)

2022-09-29 18:19:14

Thanks @Timm Haucke! I took a glance at your paper and it looks really interesting. We'll assess real-world performance and battery life in the field soon, and I am sure I will have some questions!

👍 Timm Haucke

Frederic (frederic@apic.ai)

2022-10-03 09:42:46

I have about 20 jetson Nano (4GB, A02) dev kit left and could sell them, since we switched to the NX. If any one is interested DM me.
Currently they are just collecting dust.

Since availability of the dev kit is super bad, this might help some of you. - (Hopefully i am not spamming this thread.)

👍 Carly Batist, Timm Haucke, Sara Beery, Michael Bunsen

🙌 Michael Bunsen, Luke Sheneman

🎉 Michael Bunsen

Jeff Reed (jeff@reedfly.com)

2022-09-29 23:36:43

What a great group and great presentations! It's humbling to be around such smart people...and inspiring to be around people who care about wild places so much. I study animal communication (PhD in computational linguistics) in the Greater Yellowstone Ecosystem and am active in local conservation efforts (mostly policy driven...from wolf conservation to mining restrictions).

Are there any studies on the effectiveness of conservation strategies using game trail cameras: e.g. decrease poaching (there are similar studies of CCTV usage for crime)?

I'm also interested in any efforts to formulate an ethic for the use of camera traps in ecology research or conservation. For example, the strict standards of Yellowstone Park require us to prove a reduced ecological footprint, so we have focused on building very low-power AI cameras for based on PIR signals that don't require much on the edge in terms of compute (we use https://greenwaves-technologies.com/low-power-processor/) and do a REALLY good job of filtering out the majority of false positives (e.g. grass, snow, reflectance)...positives (i.e. non-plant life) are sent to the cloud where they can be efficiently analyzed against very big and diverse models...i.e. we don't see many real-time camera trap use cases that require classification on the edge IF false positives of non-animals are removed from the data flow (see sample architecture image). In other words, if you're going to have a connected camera trap (which is the premise if you're doing AI on the edge) and you want to save battery life, then filter out "non-animals" and send the rest to the cloud for AI analysis, human in the loop, and alerting. Since PIR is the lowest power sensor we have to date, it seems like a great place to focus AI research for low-footprint camera traps; clearly, Google and Ring haven't made it a focus yet with their outdoor products (because they mostly assume power is plentiful).

I'm very interested to see the actual specs of the Conservation X camera that was announced at Edge Impulse yesterday...it looks promising, but it isn't clear if an "IoT platform" comes with it, which is somewhat necessary if you're doing to deploy these at scale in conservation research (e.g. firm updates, model updates). I'm working on a cougar population estimation using camera traps in Yellowstone Park, and was wondering if anyone had a contact at Conservation X that would be willing to talk over the phone to see if the timing of our project and the release of their cameras would make it feasible to use.

Thanks again for putting on such a great event 🙂 And if any of you are working on PIR AI models, I would love to collaborate.

GreenWaves Technologies

Ultra low power GAP processors

We design highly efficient and easy to program ultra-low-power RISC-V processors to interpret and transform rich data sources such as images, sounds and radar signals using AI and signal processing

Est. reading time

4 minutes

Original URL: https://greenwaves-technologies.com/low-power-processor/

😎 Jon Van Oast, Sara Beery, Timm Haucke, Olivier Gimenez

Giana Cirolia (giana@berkeley.edu)

2022-10-01 16:22:29

Hi!

I am very new to this group and my background is not AI or environmental science (I am a PhD in human microbiome research)

But…I am helping a native community member of the Dakota people plan environmentally based youth development activities on a plot of land she has recently purchased for the sole purpose of ecological rehabilitation and community centered environmental stewardship.

One thing they would love to do is center youth led research projects (with long term benefit) which focus on environmental preservation, and or, tracking the outcomes and benefits of traditional/holistic restoration initiatives in grasslands and natural free range pastures.

I was wondering if anyone had ideas of ways in which the tools, processes and initiatives you work on could integrate into a youth/native people centered land restoration and community empowerment project on a long term protected site.

Would love anyones thoughts about how the youth could be agents of change in this process and also how they could get more connected to stem learning and resources from perspectives they care about.

Looking forward to hearing from you all!

👍 Jaanak, Sara Beery, Dan Morris, Abhay, Jon Van Oast, Aleksis Pirinen, Anjali Ravunniarath, Carly Batist, Elizabeth Bondi, Toryn Schafer, Eddie Zhang, Emerson de Lemmus, Ashley Chang (she/her), Gedeon, Adam Noach, Rita Pucci, nyakundi lamech

👍:skin_tone_3: Pen-Yuan Hsing

Dan Morris (agentmorris@gmail.com)

2022-10-01 17:12:59

I don't exactly have an answer for you, but I know some colleagues at the Point No Point Treaty Council:

https://pnptc.org/

...who are using technology (i.e., the kind of technology we talk about on this Slack) for wildlife surveys in the area governed by the Treaty of Point No Point:

https://goia.wa.gov/resources/treaties/treaty-point-no-point-1855

...which protects fishing and hunting rights for several tribes in Western Washington. If you email me offline (agentmorris@gmail.com), I'm happy to make an introduction if that would be helpful.

❤️ Sara Beery, Jon Van Oast, Tiziana Gelmi Candusso, nyakundi lamech

Giana Cirolia (giana@berkeley.edu)

2022-10-05 18:45:30

Thank you so much!

Giana Cirolia (giana@berkeley.edu)

2022-10-05 18:50:58

I know the young folks are already involved in learning the names of and identifying their wild and natural species.

Would curated data collection on the ground or cameras set up to survey native species be of use to anyone in your network?

Perhaps passive collection as the young people survey their land is helpful? If so, I would be very happy to see how their “species learning” could serve this conservation community group as well.

Fridah Nyakundi (nyakundi@stanford.edu)

2022-10-14 17:19:25

@nyakundi lamech was involved in a youth project that does exactly this. But I don't know how valuable comments from a project outside of the US would be to you?.

Giana Cirolia (giana@berkeley.edu)

2022-10-14 17:50:27

Always valuable thank you! Please message me I would love to learn from this expertise and experience

Irené Tema (irenetema2014@gmail.com)

2022-10-09 16:16:33

Hi everyone!

My name is Irené Tematelewo (I go by Tema). I am a PhD student in computer science at Oregon State University, working with Prof. Thomas Dietterich. My research is on anticipatory distribution shift correction and, I'm currently exploring the Snapshot Serengeti CameraTrap dataset in an effort to forecast changes in species population over time based on captured images. Any pointer to other interesting image based classification dataset with temporal information is welcome.

I'm here thank to @Bistra Dilkina and I'm delighted to join the community.

👋 Suzanne Stathatos, Declan, Eddie Zhang, Ashley Chang (she/her), Lucia Gordon, Aleksis Pirinen, Rita Pucci, Valentin Gabeff, Sara Beery

👍 Bistra Dilkina

Suzanne Stathatos (suzanne.stathatos@gmail.com)

2022-10-09 16:20:47

LILA BC is a good place to look for camera trap datasets.

Are you looking for temporal shifts at a multi-year-long level, though, if you’re looking for population changes?

Irené Tema (irenetema2014@gmail.com)

2022-10-09 16:32:08

Thanks @Suzanne Stathatos! I'm mostly looking for classification datasets with trend and/or seasonal (or both) change in the class probabilities over time. For example, when studying the Snapshot Serengeti dataset, I noticed that the estimated population of Wildebeest is close to zero during dry months because they migrate out of the Serengeti for locations where there is grass and it when starts raining in the Serengeti, they migrate back because the grass is more rewarding.
So my study of the animal population over time is to find those species with predictable changes.

Sara Beery (sara.beery@gmail.com)

2022-10-11 04:28:07

Very cool!! I think Snapshot Serengeti is the longest-running accessible, public dataset with images from camera traps, but iNaturalist could be another interesting modality to investigate? There you also have the added complexity of opportunistic sampling, but in aggregate changes are captured over time

👍 Irené Tema

Kakani Katija (kakani@mbari.org)

2022-10-11 09:26:20

I would also check out ecotaxa. They focus largely on plankton classification with multi-year observations using a suite of underwater and in-lab bright field imaging tools.

👍 Irené Tema

Irené Tema (irenetema2014@gmail.com)

2022-10-11 14:51:19

Thanks @Sara Beery and @Kakani Katija. I'll check them out

Rita Pucci (rita.pucci85@gmail.com)

2022-10-10 06:40:08

Hi to all of you!

👍 Aleksis Pirinen

Rita Pucci (rita.pucci85@gmail.com)

2022-10-10 06:41:11

I just want to say that from the 1st of November I will be part of the Naturalis and I am super excited about that! I will be working on AI & Biodiversity with a great team of experts in both AI and Biodiversity!

👏 Silvia Zuffi, Diego Marcos, gvanhorn, Stephanie O'Donnell, Adam Noach, Lucia Gordon, Rita Pucci, Ashley Chang (she/her), Sara Beery

🙌 vishva shah

Thijs (thijs@q42.nl)

2022-10-10 13:16:42

Naturalis Biodiversity Center in Leiden?

Thijs (thijs@q42.nl)

2022-10-10 13:16:52

@Rita Pucci?

Rita Pucci (rita.pucci85@gmail.com)

2022-10-10 14:08:54

Yes!! Exactly

Thijs (thijs@q42.nl)

2022-10-10 14:41:47

Nice, I live in Leiderdorp, do you live in Leiden area? Might be nice to catch up?

Rita Pucci (rita.pucci85@gmail.com)

2022-10-10 14:49:33

Yes!!

Rita Pucci (rita.pucci85@gmail.com)

2022-10-10 14:49:40

Super cool!

Rita Pucci (rita.pucci85@gmail.com)

2022-10-10 06:41:49

🎉💃🎉💃🎉💃🎉

Pietro Perona (perona@caltech.edu)

2022-10-11 01:14:18

How did they count the frogs?
https://www.bbc.co.uk/news/science-environment-63206140

BBC News

Conservation: Explosion in frog numbers after mass pond digging

Scientists brought endangered species back from the brink by simply building more ponds.

Original URL: https://www.bbc.co.uk/news/science-environment-63206140

👍 Jorrit van Gils, Carl Boettiger

Carl Boettiger (cboettig@berkeley.edu)

2022-10-25 13:08:02

Yeah that's a great a question. As they describe it, it sounds like these authors base their estimates on what we call dynamic occupancy models, ( https://www.pnas.org/doi/10.1073/pnas.2123070119) Such models essentially seek to model the probability of detection for some latent occupancy, (see methods section and appendix https://www.pnas.org/doi/suppl/10.1073/pnas.2123070119/suppl_file/pnas.2123070119.sapp.pdf). such models are typically bayesian hierarchical models estimated by MCMC -- they provide their code at the end of the supplement).

I think this area is ripe for advances from the ML community but doesn't get as much attention as it could! Numerical constraints often mean that these models must ignore lots of factors that probably also influence probability of detection, as well as generally ignoring ecological factors like whether like to group together or establish separate territory (i.e. so that observations are not properly independent) etc etc. Because there is no perfect census method to compare to, it's rather difficult to say how well current methods of counting actually work.

Titus (titus@colossal.com)

2022-10-11 07:58:08

Hey Everyone, I’m new to this community. My name is Titus and I’m the VP of Strategy and Head of Computational Sciences at Colossal Biosciences. My background is in AI+Genomics but my team is working across genomic, audio, and imaging methods for conservation and de-extinction. This community is great and looking forward to chatting with people as the opportunity comes up! If anyone ever needs anything or wants to chat, shoot me a note.

colossal.com

Reawakening the Lost Wilds of Earth

Extinction is a colossal problem facing the world. See how COLOSSAL, the de-extinction company, is developing the science that will save us, our planet and the species that inhabit it.

Original URL: https://colossal.com/

😊 Aleksis Pirinen, Alexander Robillard, Carly Batist, Otto Brookes, Sara Beery

💡 Ștefan Istrate, Alexander Robillard

👏 Stephanie O'Donnell, Alexander Robillard, Jon Van Oast, Irené Tema, Adam Noach, Declan

Giana Cirolia (giana@berkeley.edu)

2022-10-12 20:04:15

Hi!

Maybe it is far afield but one area where we are really lacking in preservation efforts in the biodiversity created by diverse diets, food practice and local ecologies.

I’d be thrilled to discuss further if your team has been thinking at all about how to preserve and maintain vaninshing human/health associated microbes and plants :)

I am a human microbiome PhD student at UCB but also part of their nsf digital transformation of development cohort which helps PhD students center their questions in translational efforts that serve humanity, particularly vulnerable groups (people, species etc).

I am looking for better ways to expand the questions we ask around the human biome to push preservation of diverse cultural food practice and diverse edible species.

If any of that helps I would be very happy to talk more (at least on the bio side of things).

I am still teaching myself the computational side!

Feel free to email
Giana@berkeley.edu

Kieran (kag25@sussex.ac.uk)

2022-10-12 10:55:15

Hi all, I'm new to this space. I'm doing a PhD at Sussex University looking at building interpretable machine learned representations of soundscape audio recordings collected from ecosystems - i.e. ecoacoustics. I'm a data scientist and FOSS (free & open source) software engineer by trade, my ecological knowledge is not learned within an academic environment but from trudging around various landscapes on foot! Looking forward to chatting about the use of sound (and more) as a tool for tracking biodiversity, changes in habitat quality and more.

👍 Oisin Mac Aodha, Dan Morris, Adam Noach, Declan, Ivor Simpson, Justin Kay, Ali Johnston, Barbie D, Julia Marisa Sekula, Sara Beery, Aleksis Pirinen, Jinsu Elhance

👍:skin_tone_3: Pen-Yuan Hsing

Julia Marisa Sekula (jmsekula@stanford.edu)

2022-10-12 14:41:31

Hi All! I’m a Stanford MBA |Msc and currently doing an independent study project looking to understand the dynamics of the 🧬 genomic data market and how industry is thinking about potential new sources of this data - especially for conservation/biodiversity purposes 🌿. For this, we are speaking to Academics, Pharma, SynBio, to Insurers. If anyone has any leads or would like to chat - we would love an introduction!

🙌:skin_tone_3: Pen-Yuan Hsing

🙌 Jon Van Oast, Suzanne Stathatos, Dhruv Sheth, Sara Beery, Fridah Nyakundi

Giana Cirolia (giana@berkeley.edu)

2022-10-12 19:58:58

Hi! I am not involved in the purely plant side of this, but I know many efforts are under way to think about conservation of human internal ecological biodiversity.

It’s a really difficult topic because it’s not easy to fund collections in non US populations and also there are limited frameworks for ethical and human/culturally centered approaches to storing and using that data.

Happy to talk more if any of my information relative to the human biome can help!

Giana Cirolia (giana@berkeley.edu)

2022-10-12 19:57:13

Hi All,

Thank you again for your initial connections relative to native land rehabilitation.

I was wondering if I could ask a further question.

Many tribes seek to restore their lands from previous monoculture farmland or pesticide exposure into natural habitats which support small sustainable farming efforts, biodiversity restoration and rejuvenation of native species.

Are there ways that current satellite and land survey data are being leveraged give remote help to develop recommendations for targeting/ organizing soil/species recovery processes?

For example, from such data has anyone developed processes for identifying the needs of the soil (without physical testing) or the locations most primed to support self-sustained regrowth of native places and recovery of native species?

Is this very very domain specific or can generalized insights for soil management suggestions be gleaned from such data without feet on the ground testing? Or in combination with ground testing?

😯 Aleksis Pirinen, Pen-Yuan Hsing

❤️ Adam Noach, Aleksis Pirinen, Alexander Robillard, Sara Beery, Yseult Hb

Alexander Robillard (RobillardA@SI.EDU)

2022-10-13 10:04:35

Hi Giana, this sounds very similar to some of the work being done by Propogate- much of their tools do this with a focus on agroforestry as a method for sustainable economic development, conservation and carbon sequestration, you might find their analytics tools to be exactly what you're looking for. Their tools utilize multiple input layers to assess soil quality, hydrology etc to outline which species might be best to plant. Keep in mind theyre not doing this directly with AI but with pre-generated spatial layers. Their resolution is pretty decent as well. Happy to put you in touch with one of my friends who works there if its of interest. https://propagateag.com/

Best of luck!

propagateag.com

Propagate

Agroforestry project development and investments platform.

Original URL: https://propagateag.com/

Giana Cirolia (giana@berkeley.edu)

2022-10-13 11:26:49

My goodness that would be such a gift! Thank you so much I would be so grateful.

My email is giana@berkeley.edu :)

👍 Alexander Robillard

Howard Windsor (Wildbook@hwindsor.me.uk)

2022-10-13 09:23:47

Hi all,
I'm a UK based experienced software engineer (20+ years) who now has some time and a desire to use my skills and experience to aid conservation and help combat climate change rather than grow shareholder profit.
I've worked extensively on embedded systems, everything from writing direct to HW registers in device drivers on bare metal systems through RTOS and embedded linux.
I have also worked on flask based REST API Web servers (github.com/hwindsor).
I have seen and written well designed software systems that are simple to extend. I have also experienced the troubles with badly designed systems so can help a lot on problems to avoid.
If you have any software projects that you would like some help on, design as well as development, please drop me a DM and let's see what we can do together.

👋 Josh Seltzer, Adam Noach, Dan Morris, Jon Van Oast, Eelke, Ted Schmitt, Barbie D

👍 Aleksis Pirinen, Sam Kelly, Sara Beery, Luke Sheneman

❤️ Giana Cirolia, Mark Fisher, Pen-Yuan Hsing

👍:skin_tone_3: Pen-Yuan Hsing

😎 Jon Van Oast

Yuanqi Du (yd392@cornell.edu)

2022-10-14 17:11:04

Hey guys, my name is Yuanqi Du (https://yuanqidu.github.io/), a first year CS PhD student at Cornell working with Prof. Carla Gomes. I have been working on molecular simulation and drug discovery for about three years, and now I am new to materials science and computational sustainability! I hope to learn a lot from you! Also, we are leading an AI4Science101 initiative https://ai4science101.readthedocs.io/en/devel/index.html where we aim to provide a series of motivational and overview blogs to introduce both AI tools and Science problems to people and motivate people to join the exciting field, if anyone is interested in contributing interesting things related to conservation, please DM me!

Yuanqi Du's Personal Website

Yuanqi Du

About me

Original URL: https://yuanqidu.github.io/

👋 gvanhorn, Justin Kay, Declan, Avi Sundaresan, Dan Morris, Carly Batist, Kangyu Zheng, Sara Beery, Jaanak, Emilio Luz-Ricca, Gedeon, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Chris Yeh

Kasirat (kasirat_turfi@hotmail.com)

2022-10-16 01:46:30

Hi everyone, anybody working with point clouds? Want to connect with people who are working with forest point clouds!🌲either ALS or TLS!

👀 Ethan Shafron

👍 Chinmay Talegaonkar, Aleksis Pirinen, Gedeon

Eddie Zhang (ete@ucsb.edu)

2022-10-17 10:53:27

Just stumbled on this competition, super super cool! www.ai4climatecoop.org

🎯 Josh Seltzer, Sara Beery

Eddie Zhang (ete@ucsb.edu)

2022-10-17 10:55:07

I'm thinking about getting involved right now, dm me if it looks interesting to you

Josh Seltzer (jyseltz@gmail.com)

2022-10-17 11:28:19

I just signed up the other day! I haven't gotten around to watching the introduction videos yet, as there is a lot to take in and some of it is honestly beyond my comprehension at this point. I'm definitely interested though and would love to team up with someone : )

Anselm Bradford (ans@anselmbradford.com)

2022-10-18 13:02:34

https://openopps.usajobs.gov/tasks/4347?fromSearch 4 month plus detail with NOAA “AI-Ready Data Lead for the NOAA Center for Artificial Intelligence - Open to All”

👍 Justin Kay, Leonardo Viotti

Alessandra Sellini (sellini.alessandra@gmail.com)

2022-10-19 06:38:24

Hello everyone! Since I’m new in this community, I’d thought to introduce myself 😄

I completed an MSc in Marine Biology and Ecology at James Cook University, Australia, then moved to the Philippines to work as a research assistant and research fellow at a local NGO 🌿 There I first approached the world of tech and conservation while working on the creation of 3D reef models based on DO-SVS footage. At the beginning of the pandemic, I moved back to Italy, my home country, and worked for two years at the WWF Italy Conservation Office as an Associate 🐼 Besides stakeholder engagement and drafting different documents, I had the opportunity to become a drone pilot. Working on a project for cetacean conservation, we planned to deploy a drone to gather footage and genetic material from whale blows 🐋

I grew fascinated by the possibilities offered by the intersection between tech and conservation and decided to pursue this path. I entered the world of AI/ML after attending a 9-week Data Science bootcamp in Brussels. As a final project, my team and I developed a model to target new potential Marine Protected Areas using current environmental, biodiversity and fishing data. Having only two weeks, we chose to focus on European waters and the Arab Sea 🌊

I’m now looking for new adventures 🚀 If anyone is interested in talking about AI, ML, or marine biology, I would love to hear from you!

👍 Valentin Lucet, Rebecca, Alexander Robillard, Dan Morris

👋 Viktor Domazetoski, Rebecca, Adam Noach, Alexander Robillard, Déva Sou, Lucia Gordon, Anjali Ravunniarath, Taiki Sakai - NOAA Affiliate, Suzanne Stathatos, Jon Van Oast, Howard Windsor, Yseult Hb, Eddie Zhang, Malika Nisal Ratnayake, Kieran, Jaanak, Moira Shooter, Omiros Pantazis

Jorrit van Gils (vangilsjorrit@gmail.com)

2022-10-21 08:57:00

Hi all,

Programming skills are very useful as Ai for conservation researcher. Therefore, I am looking for a par-time programming job where I can increase Python skills, preferably related to computer vision/AI. Last year I worked for an AI website where I learned working with datasets, object oriented mapping, SQL, GIT and Docker. Despite the steep learning curve, I'm not an experienced programmer yet. So, I’m looking for an opportunity where I can contribute while improving these skills. It would be very much appreciated if someone can point me in the right direction.

For more info please check my webpage.

Thanks a lot!

Jorrit

👍 Jon Van Oast, Kai Waddington, Helena Russello, Peter van Lunteren

🙌 Abhay

Sicily Fiennes (sicilyfiennes@gmail.com)

2022-10-23 07:11:08

Hi everyone,

Just a couple of queries from me! I’m working on an application for the CV4Ecology summer school and trying to make a good labeling plan, which was one of the reasons why I was unsuccessful the first time applying. Part of my PhD project is using machine learning to improve visual identification for birds in the wildlife trade (focussing on the case study of . My existing dataset consists of labeled images of over 100 species, with between 70-300 image per species. I’m also interested in attempting multi species identification in images and object detection for images in marketplaces where there are many individuals. If any successful applicants from last year want to get in touch please do!

Secondly, I’m looking for a little extra funding (£1,000-2,000) for some fieldwork. If anyone knows of any small pots of money closing soon, that don’t have long decision time, feel free to comment below or PM me, I’d be so grateful!

Rowan Converse (rowanconverse@unm.edu)

2022-10-23 22:28:06

Hi Sicily, I was in the 2022 cohort and would be happy to chat! Send me a DM 🙂

Taiki Sakai - NOAA Affiliate (taiki.sakai@noaa.gov)

2022-10-25 13:04:54

Hi Sicily! I was also in the 2022 cohort and happy to chat if you want (also hi Rowan!)

👋 Rowan Converse

Sicily Fiennes (sicilyfiennes@gmail.com)

2022-10-25 17:17:29

Hi both! Thanks for the message this sounds great. Will PM you!

Phuc Le (phuc.le@ug.fuv.edu.vn)

2022-10-23 12:14:09

Hi everyone, my name is Phuc, I’m a Vietnam-based CS undergrad student. I am currently working on my capstone thesis which develops an AI-intergrated mobile app that is able to recognize 26+ species of native and non-native freshwater turtles and tortoises in Vietnam to combat illegal trutle trades both online and offline. Primary users will be law enforcers, then nature enthusiasts, educators, etc. The model should be able to classify turtle species in images from the field or social media. I am in data collection phase and doing that by taking photos of turtle species being protected in a rescue center. They have >2000 individuals of 22 species, which is enough for me to take pictures of. However, I have some questions that dont know how to start with

What device should I use to take pictures? mobile phone?, a Sony camera?, or a mixture of both? Is resolution a matter in self-collected data?

What characteristics of the data should I taken into accounts (for example, quantity, diversity in background, lighting, camera angles, turtle individuals) to ensure generalizability?

Could anyone have experiences in projects that use self-collected data - not dependent on Internet sources - to train ML/AI model share some thoughts on these questions? Are there any guidelines out there that I can apply to my project? Or if you know any papers or projects that also intend to create animal identification apps for similar purpose, please share.
This is my very first big AI project so please forgive if I ask silly questions :)
Thanks everyone in advance!

Valentin Ștefan (valentin.stefan.vst@gmail.com)

2022-10-24 10:33:57

You can have a look at the work done by iNaturalist - the seek app, Flora Incognita app, or PlantNet app. The more sensors you use to capture the images, the better would be for generalizability, but from my experience with CNN for insect localization & classification, if you develop a model that will be used on a phone, then capture your training images with the phone, exactly how you would expect a user to do it. My main worry is that if you aim for too much generalizability, then it might not work at some point.
See also the guidelines of YOLOv5 at https://github.com/ultralytics/yolov5/wiki/Tips-for-Best-Training-Results

play.google.com

Flora Incognita - Apps on Google Play

Plant identification with automatic image recognition

Original URL: https://play.google.com/store/apps/details?id=com.floraincognita.app.floraincognita&hl=en&gl=US

Alexander Robillard (RobillardA@SI.EDU)

2022-10-24 14:45:21

Hi Phuc, feel free to send me an DM/email, happy to make some time to talk if you're interested! Key features for turtles are mostly carapace and plastron shots. Side photos of the head being the second most important feature. I also strongly recommend trying to develop a mask/segmentation model to extract the shell

Phuc Le (phuc.le@ug.fuv.edu.vn)

2022-10-26 05:51:39

Thanks @Valentin Ștefan for the insights! I have used Pl@ntNet before and was impressed by its ability to recognize in Vietnam. I totally forgot about it until you mentioned. I will take a look at their technology. I also used YOLOv5 to localize the position of the turtle in an image and it generalized quite well. But I think I will use different networks for the classification task because I want the model to return many predictions so that users can verify them. But still the guide from YOLOv5 team is very informative and helpful!

Phuc Le (phuc.le@ug.fuv.edu.vn)

2022-10-26 05:52:07

Hi @Alexander Robillard, thanks for your note, I have DM-ed you for further discussion 🙂

Valentin Ștefan (valentin.stefan.vst@gmail.com)

2022-10-26 05:57:08

Note that, with the new release of YOLOv5 - v6.2, you can also do image classification not only localization. I think you can tell the model to return several predictions (possibly with the --conf-thres and --iou-thresin the detect.py script. If you explore further with YOLOv5, let me know as I am also interested in how it performs and how can the weights be deployed as an app on a phone. At the moment, I am unsure about how fast will it run on a phone and I do not have the coding skills yet to build an app. I presume the nano weights version would be a better option

Phuc Le (phuc.le@ug.fuv.edu.vn)

2022-10-26 06:29:04

I didnt know that they have classification versions. Thanks a lot. I will test these models out.

Anh Quoc Nguyen (nguye2aq@mail.uc.edu)

2022-10-26 20:33:19

Imo, u can just take the photos with the best quality, then lower its quality during training as augmentation. Also, i dont see why u should stick to yolov5, u can just do yolov7, any other new SOTA ones. And yes, u should take into accounts all those data characteristic. Hey dm me, I'd love to discuss and help out

Phuc Le (phuc.le@ug.fuv.edu.vn)

2022-10-26 21:21:05

@Anh Quoc Nguyen I've DM-ed you!

Alayna Van Dervort (av@thebigwild.com)

2022-10-26 12:48:45

Hello All,

Alayna Van Dervort (av@thebigwild.com)

2022-10-26 12:49:13

Can I gain some advice on the best way to get an animal to look directly at a camera trap?

Michael Procko (xprockox@gmail.com)

2022-10-26 12:50:11

Is this for a specific animal or all species that might be present in the study area?

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-10-26 12:50:27

Do you mean by some sort of light/sound stimulus?

Alayna Van Dervort (av@thebigwild.com)

2022-10-26 12:51:01

This one in particular is for Large Cats

Alayna Van Dervort (av@thebigwild.com)

2022-10-26 12:51:38

I have mega tons of data that is from the side but very limited data on direct facial images

Alayna Van Dervort (av@thebigwild.com)

2022-10-26 12:52:02

(especially because much of the data is at night)

Alayna Van Dervort (av@thebigwild.com)

2022-10-26 12:53:14

I see much of the Snow Leopard cats take a min to look directly at the camera.. I would love to know how.why

Michael Procko (xprockox@gmail.com)

2022-10-26 12:54:05

Not sure if you can use images for this (may need video because it happens so fast), but mountain lions often look directly at cameras before fleeing when the cameras are paired with audio playback of human voices... https://www.washingtonpost.com/news/speaking-of-science/wp/2017/06/21/mountain-lions-are-terrified-by-the-voices-of-rush-limbaugh-and-rachel-maddow/

Washington Post

Mountain lions are terrified by the voices of Rush Limbaugh and Rachel Maddow

Scientists spent several days scaring the big cats with their chatter in an effort to understand how fear of humans shapes their ecosystem.

Original URL: https://www.washingtonpost.com/news/speaking-of-science/wp/2017/06/21/mountain-lions-are-terrified-by-the-voices-of-rush-limbaugh-and-rachel-maddow/

Michael Procko (xprockox@gmail.com)

2022-10-26 12:55:39

If you're restricted to using just images, and no video, I would imagine you'd need a scent lure or something they are more likely to investigate for a longer period of time.

Alayna Van Dervort (av@thebigwild.com)

2022-10-26 12:57:38

Interesting, thank you. Rush L. WOuld scare the hell out of me too ;) I am looking for noninvasive methods. We are also collecting video and I was wondering about the use of scent. DO you or anyone have experience using it?

Michael Procko (xprockox@gmail.com)

2022-10-26 12:59:56

Yes, but I would have to dig it up from archives of messages, and am running to a meeting now. I will make a note to come back and respond!

Alayna Van Dervort (av@thebigwild.com)

2022-10-26 13:13:42

Thank you!

Dan Morris (agentmorris@gmail.com)

2022-10-26 17:09:35

I've only owned cheap consumer camera traps, but I'm curious whether anyone knows about the bandwidth of the IR flash on research-grade vs. consumer-grade cameras. I wouldn't say it's "giant red light" on my cheap consumer camera trap, but it's close: you can definitely see visible red during the flash, and every coyote walking by the camera looks right at it. I would guess the flash is much narrower in bandwidth on serious cameras, but I have no data to back that up. But if that's accurate, maybe a solution is: use a terrible camera?

👍 Mitch Fennell

Dan Morris (agentmorris@gmail.com)

2022-10-26 17:09:46

[doesn't help you during the day]

Dan Morris (agentmorris@gmail.com)

2022-10-26 17:11:21

Of course, sound will be more effective in attracting attention (I assume), but it sounds like you're trying to walk the fine line between attraction attention and deterring presence, and maybe "unintentional red flash" hits that sweet spot.

Dante Wasmuht (dante@conservationxlabs.org)

2022-10-27 03:53:45

https://onlinelibrary.wiley.com/doi/full/10.1002/ece3.8536

Dante Wasmuht (dante@conservationxlabs.org)

2022-10-27 04:14:57

sounds like they played a recording of a juvenile puma call plus light flash to get a mugshot

Paul Allin (allinpaul@gmail.com)

2022-11-01 03:44:55

baited camera traps can work but then your observations are beginning to interfere with their natural behaviour so depends a bit on the overall objective

Scott Hosking (jshosking@gmail.com)

2022-10-27 10:49:58

New Special Interest Group: Biodiversity monitoring and forecasting
How can we best catalyse and champion the development of new AI and data-driven methods for monitoring and forecasting biodiversity change?https://www.turing.ac.uk/research/interest-groups/biodiversity-monitoring-and-forecasting

🎉 Oisin Mac Aodha, Aleksis Pirinen, Sara Beery, Dan Morris, Carly Batist, Tiziana Gelmi Candusso, Ali Johnston, Emily Charry Tissier, Cathy Atkinson, Ando Shah

Sara Beery (sara.beery@gmail.com)

2022-11-01 13:36:37

We're hosting an InfoSession for those interested in the CV4Ecology Workshop in 2023 this Thursday Nov. 3 from 10-11PT! More details here: https://twitter.com/cv4ecology/status/1587217345180622849

👍 Oisin Mac Aodha, Subhransu Maji, Josh Veitch-Michaelis, Dan Morris, Valentin Ștefan, Mitch Fennell, Taiki Sakai - NOAA Affiliate, Kalyan Nadimpalli, Catherine, Lukas Picek, Stephanie O'Donnell, Riccardo de Lutio, Déva Sou, Ben Williams

👍:skin_tone_3: Pen-Yuan Hsing

😎 Jon Van Oast, Emerson de Lemmus

Ben Weinstein (benweinstein2010@gmail.com)

2022-11-02 12:50:22

Its grant writing season. Quick poll! What does the AI for ecology community need in the next five years?

🐵 New data like benchmarks, hardware and image databases
🎈 New open source models that perform reasonable predictions on common ecological tasks (e.g. megadetector -> camera traps, deepforest -> trees, merlin -> bird sound id)
🐘 Better tools that allow ecologists to create machine learning models with GUI interfaces/relatively little code (e.g. AIDE)
🦕 Better algorithms that allow existing data to applied more easily across space and time (class imbalance, few shot learning, geographic/taxonomic generalization).
You are welcome to add comments, but please vote for only one.

🦕 Ben Weinstein, Dante Wasmuht, Toryn Schafer, Matt Weldy, Yseult Hb, Alba Solsona, Lukas Picek, Omiros Pantazis, Tiziana Gelmi Candusso, Taiki Sakai - NOAA Affiliate, Déva Sou, Josh Veitch-Michaelis, Emerson de Lemmus, Daniel Grzenda, Devis Tuia, Sara Beery, Ethan Shafron, Claudia Haas, Kakani Katija, nyakundi lamech, Jason Holmberg (Wild Me), Sicily Fiennes

🐵 Georgia Atkinson, Lukas Picek, Valentin Lucet, Heather Lynch, Valentin Ștefan, Emilio Luz-Ricca, Yves Bas, Sara Beery, Ethan Shafron, Ando Shah, Kakani Katija

🐘 Carly Batist, Mitch Fennell, Felipe Parodi, Tiziana Gelmi Candusso, Blair Costelloe, Peter van Lunteren, Rowan Converse, Antón Álvarez

🎈 Mitch Fennell, Felipe Parodi, Dan Morris, Devis Tuia, Stefan Schneider, Sara Beery, Kelly Easterday, Peter van Lunteren

Valentin Lucet (valentin.lucet@gmail.com)

2022-11-02 14:49:32

I've only been around this space for a short while, but I've seen the discussions that you led about standards and how "not to re-invent the wheel", especially when it comes to processing images. I think that's another one to add to the list: we have many tools but nothing that has been widely embraced (except maybe for megadetector?).

Peter van Lunteren (peter@addaxdatascience.com)

2022-11-03 17:37:36

For those of you who are interested: I’ve made a GUI for Windows, Mac and Linux which uses the MegaDetector model to analyse images. The idea behind this is to make the power of MegaDetector available for non-coding ecologists too. Opening and installing the GUI is done by bash/batch files, so the user doesn’t have to be bothered with installing anaconda, git for windows, labelImg and all the other prerequisites.

Besides running MDv5, it can:

Place empty images, people, vehicles or animals in subfolders

Export .xml label files in Pascal VOC format for further model training

Create an input file for further processing in Timelapse

Manipulate data by drawing boxes or cropping detections

Review and edit annotations using the open-source annotation software labelImg

If you know any colleagues or friends who can use this GUI, spread the word! -> https://github.com/PetervanLunteren/EcoAssist

PetervanLunteren/EcoAssist

An application for detecting animals in camera trap images using the MegaDetector model.

Stars

Language

Python

👍 gvanhorn, Dan Morris, Timm Haucke, Mitch Fennell, Tiziana Gelmi Candusso, nyakundi lamech, Yuerou Tang, Evan Hallein, Jason Holmberg (Wild Me)

🙌 Suzanne Stathatos, Lucia Gordon, Felipe Parodi, Taiki Sakai - NOAA Affiliate, Carly Batist, Cameron Trotter, Jacob Kamminga, Timm Haucke, Talia Speaker, Yseult Hb, Ethan Shafron, Tiziana Gelmi Candusso, David Will, Luke Sheneman, Kakani Katija, Jaanak, nyakundi lamech, Jason Holmberg (Wild Me)

🎉 Jon Van Oast, Fadel Seydou, Valentin Gabeff, Gedeon, Timm Haucke, Sofía Miñano, Yseult Hb, Tiziana Gelmi Candusso, Sara Beery, David Will, nyakundi lamech, Jason Holmberg (Wild Me), Dan Morris

💯 Eelke, Sam Kelly, Dante Wasmuht, Timm Haucke, Tiziana Gelmi Candusso

👍:skin_tone_3: Pen-Yuan Hsing

😍 Sara Beery, Antón Álvarez

Jon Van Oast (jon@wildme.org)

2022-11-03 17:48:42

this is great! thank you.

Dan Morris (agentmorris@gmail.com)

2022-11-03 18:54:47

Wow, you've added tons of features here! Out of curiosity: it's rare for an ecologist to need to add/manipulate bounding boxes for analysis purposes, so I assume that you and/or users are using that functionality to either fine-tune MegaDetector or train a new detector... is that right? Can you share anything about what the edited boxes are being used for?

Peter van Lunteren (peter@addaxdatascience.com)

2022-11-04 01:06:07

A user wanted to use MD to kickstart his annotation process, so that he didn’t have to manually draw all the boxes. He would then only have to change the label to the species (and fine-tune the results, if needed). With this reviewed and annotated data he would train his own site-specific model. That’s why I added the feature of converting MD’s output.json to individual .xml files and getting rid of the labelImg installation hassle. I agree with you that most users won’t use this feature, but, nevertheless, it’s possible :)

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2022-11-04 08:47:29

Great work, thanks for sharing! If a user already has some of the dependencies on their system, can this tool use them instead of installing everything from scratch with the batch file?

Sofía Miñano (s.minano.glez@gmail.com)

2022-11-04 10:10:24

potentially of interest for future developments: over the summer I worked with others in the DeepLabCut team integrating MegaDetector and DeepLabCut. You can check the prototype at the following HuggingFace space

👀 Antón Álvarez

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2022-11-04 15:55:10

Ahhh this was exactly what i needed in July. I love this! Thank you so much!

Peter van Lunteren (peter@addaxdatascience.com)

2022-11-04 17:12:08

@Pen-Yuan Hsing The install files search for a working installation of anaconda and git for windows. If present, it will use them. The rest (gits and models), however, will be downloaded from scratch and stored in a hidden folder. This has a few reasons: I) I want the gits to be checked out at a specific time to avoid conflicts with newer commits, II) it will slow down the installation and opening of the app significantly, and III) if the user then adjusts a directory somewhere down the tree, it will error.

@Sofía Miñano very interesting! Might be worth incorporating in a future version of EcoAssist too 🙂

😀 Sofía Miñano

👀 Pen-Yuan Hsing

Luke Sheneman (sheneman@uidaho.edu)

2022-11-07 17:07:14

This is great! I've been contemplating doing this, and it looks like you've done a great job here.

Antón Álvarez (aalvarez@wwf.es)

2022-11-09 07:52:34

AWESOME! Thank you so much for your work! Some Iberian lynx field technicians are starting to use it, and they are fascinated!! They are very happy!! One question...about the use of custom yolov5 model, there are any consideration to take in account? (before to think about a train a new one?) Thanks so much! @Peter van Lunteren

🤩 Sofía Miñano

Peter van Lunteren (peter@addaxdatascience.com)

2022-11-09 15:18:23

Hi @Antón Álvarez, Thanks for your message. Always good to hear that people can use it for conservation 🙂 EcoAssist can run custom yolov5 models if they are retrained from the MegaDetector model using transfer learning. For example, if you find that MegaDetector is not great at recognising a certain species as “animal”, you can retrain the model and add some labelled data of the cases you want it to improve on. Additionally, you can expand on the three default classes MegaDetector uses (“animal”, “person” and “vehicle”) and retrain the model to detect custom classes (e.g. “Iberian Lynx”). In that case, if you add classes, you’ll need to adjust rundetector.py and separatedetectionsintofolders.py too. If you need any help with that, let me know!

😍 Antón Álvarez

Antón Álvarez (aalvarez@wwf.es)

2022-11-18 12:13:43

Thank you very much for clarifying everything, it was to have things clear before embarking on the training of the model, (i need to allocate my time or budget) if I have any questions I will contact you. And thank you very much (again) for the new EcoAssist v2.1 version!

😁 Peter van Lunteren

Dan Morris (agentmorris@gmail.com)

2023-01-06 15:07:13

I finally got a chance to try EcoAssist today, and not only am I super-impressed (it worked great on my Windows PC, even using the GPU correctly), the installer scripts you've written are absolute works of art. MegaDetector aside - in fact, conservation aside - you have a really neat approach to packaging and deploying Python code in general for users who have never heard of Python.

Batch file nerds all over the world will be in envy of what you've done in that batch file. I bet somewhere there's a Reddit forum called "things you didn't think you could pull off in a batch file", and if there is, you should totally post there.

The fact that you even did all the magic stuff to allow accelerated inference on M1 Macs - still with a one-click deploy! - is icing on the cake.

😁 Peter van Lunteren

👍:skin_tone_3: Pen-Yuan Hsing

💯 Tiziana Gelmi Candusso

Anna Boser (annaboser@ucsb.edu)

2022-11-04 16:16:56

Does anyone know what it's called/if there are papers on the phenomenon where overall R2 does not equal the mean of R2s calculated over different groups in your data? With the application of this being that a model's predictive ability in space (R2 calculated over temporal groups) versus in time (R2 calculated over spatial groups) is different?

Graeme Phillipson (graeme.phillipson@bbc.co.uk)

2022-11-08 06:35:45

Do you mean like Simpson’s paradox? ( https://en.wikipedia.org/wiki/Simpson%27s_paradox )?

Wikipedia (https://en.wikipedia.org/)

Simpson's paradox

Simpson's paradox is a phenomenon in probability and statistics in which a trend appears in several groups of data but disappears or reverses when the groups are combined. This result is often encountered in social-science and medical-science statistics, and is particularly problematic when frequency data are unduly given causal interpretations. The paradox can be resolved when confounding variables and causal relations are appropriately addressed in the statistical modeling. Simpson's paradox has been used to illustrate the kind of misleading results that the misuse of statistics can generate.Edward H. Simpson first described this phenomenon in a technical paper in 1951, but the statisticians Karl Pearson (in 1899) and Udny Yule (in 1903) had mentioned similar effects earlier. The name Simpson's paradox was introduced by Colin R. Blyth in 1972. It is also referred to as Simpson's reversal, the Yule–Simpson effect, the amalgamation paradox, or the reversal paradox.Mathematician Jordan Ellenberg argues that Simpson's paradox is misnamed as "there's no constriction involved, just two different ways to think about the same data" and suggests that its lesson "isn't really to tell us which viewpoint to take but to insist that we keep both the parts and the whole in mind at once."

Original URL: https://en.wikipedia.org/wiki/Simpson%27s_paradox

👍 Jose Ruiz-Munoz

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-11-07 09:42:55

WILDLABS’ annual State of Conservation Tech Survey is open! https://www.linkedin.com/posts/wildlabs-community_tech4wildlife-activity-6994417296681160705-t3mp

linkedin.com

WILDLABS Community on LinkedIn: #tech4wildlife

To conservation technology users and makers everywhere - how is #Tech4Wildlife working for you? What improvements would you prioritize in the tools you use day…

Original URL: https://www.linkedin.com/posts/wildlabs-community_tech4wildlife-activity-6994417296681160705-t3mp

👍 Oisin Mac Aodha, Sara Beery, Jaanak, Catherine, Ted Schmitt

🙌 Stephanie O'Donnell

❤️ Talia Speaker, Sara Beery

Leopoldo André Dutra Lusquino Filho (leopoldo.lusquino@unesp.br)

2022-11-09 13:28:19

Hello everyone! My name is Leopoldo and I'm a newly hired AI Assistant Professor at São Paulo State University. We are starting to develop here an ML project applied to the creation of a smart campus at our university.

We want to deal with the full scope of concerns typical of this domain (energy efficiency, transition to renewable energies, air quality, water consumption, environmental impact of university activities on the local community, etc) using ML models that are energy efficient themselves. We are also interested in discussing fair energy efficiency metrics for ML models, which also takes into account social factors.

Our team has professors from the areas of computing, electrical engineering and environmental engineering, as well as undergraduate, master's and doctoral students and post-docs. Our partnership network for this project also includes other Brazilian universities, such as the State University of Campinas, the Federal University of Rio de Janeiro and the Fluminense Federal University.

Our short-term goal (three next years) is to develop research in green AI and to create a smart campus model that can be replicated for institutions with decentralized campuses. In the long term, we want to create an inter-institutional research center on ML for climate change, with an emphasis on creating public environmental policies for Latin America, especially for the state of São Paulo (Brazil).

We are looking for mentoring from international labs with experience in the areas of ML applied to smart environments and climate change, so that through our partnership we can exchange professors and students between our institutions, create smart campus-based datasets located in developing countries and accelerate our research.

Anyone interested in talking to me about this topic, please send a direct message or an email to leopoldo.lusquino@unesp.br. Thanks everyone in advance!

🙌 Elijah Cole (Deactivated), Jon Van Oast, Stephanie O'Donnell, Emily Lines, Alexander Robillard, Jose Ruiz-Munoz, Aleksis Pirinen, nyakundi lamech, Yseult Hb, Eric Colson

❤️ Tiziana Gelmi Candusso, Sara Beery, Alexander Robillard, Julia Marisa Sekula

👋 Sara Beery, Alexander Robillard, Carly Batist, Vinicius Amaral

Alex Lascelles (alexlasc@mit.edu)

2022-11-11 11:06:20

Hi all!! 👋
My partner is job-hunting for a tech startup whose mission is climate related, one looking to hire full-stack engineers. We're both very climate-conscious but we're new to the climate tech job market -- I'm wondering if anyone here
a) knows of such an opportunity, or
b) is knowledgeable in this area & could spare 10mins for a chat to educate us a little.

Any help will be immensely appreciated!

p.s. Very excited to join this community! Lots of fascinating and important work you guys are involved with. For anyone interested, my background is UG in physics/astronomy --> MSc in how music and sound affects the brain --> currently work at MIT, CSAIL in the space between cognitive science, computer vision, and AI (http://olivalab.mit.edu/). Great to meet you all! 😊

👀 Josh Seltzer, Sara Beery

Alex Lascelles (alexlasc@mit.edu)

2022-11-11 11:06:40

(Extra info: Preference for an early-stage startup, but mature enough to have at least ~5-10 engineers already. Her background is 6+ yrs in Lead & Senior Software Engineering roles at an e-commerce company, working with React, TypeScript, Gatsby, Node.js, and AWS.)

Sara Beery (sara.beery@gmail.com)

2022-11-11 12:17:43

The Climate Change AI newsletter is also a good place to look for job opportunities!

Toryn Schafer (tschafer@tamu.edu)

2022-11-11 12:51:46

I know people who are or were at Jupiter Intelligence. Looks like they might have some openings:
https://jupiterintel.com/jobs/

Jupiter

Jobs

Est. reading time

2 minutes

Original URL: https://jupiterintel.com/jobs/

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-11-11 13:19:53

We’ve got some job boards on the Conservation Tech Directory! Including the previously mentioned Climate Change AI, which I second as a great resource!

conservationtech.directory

Vermont Tech Jam: October 22, 2022 at Hula Lakeside in Burlington

Visit the post for more.

Original URL: https://conservationtech.directory/

Suzanne Stathatos (suzanne.stathatos@gmail.com)

2022-11-11 13:25:27

There are also 2 slack channels that I know of dedicated to particularly climate-related things:

ClimateAction.Tech

WorkOnClimate

Alex Lascelles (alexlasc@mit.edu)

2022-11-16 00:26:24

Thank you so much for your suggestions @Sara Beery @Toryn Schafer @Carly Batist and @Suzanne Stathatos, we really appreciate it!

❤️ Suzanne Stathatos

slackbot

2022-11-16 18:28:28

This message was deleted.

Sachith Seneviratne (sachith.seneviratne@unimelb.edu.au)

2022-11-16 18:41:04

whoops let me see if I can move it, thanks.

Sachith Seneviratne (sachith.seneviratne@unimelb.edu.au)

2022-11-16 18:42:36

Deleted as wrong channel.

Devis Tuia (devis.tuia@epfl.ch)

2022-11-17 03:00:19

💥 MEGA PROJECT and JOB ALERT: Hello everyone! We (U. Schultz, @Devis Tuia, @Blair Costelloe @Tilo Burghardt, M. Wikelski, B. Risse and many more) are happy to share that we will start early next year the project WILDDRONE (wilddrone.eu)! It is a Marie Curie Network, meaning that we will build a network of 13 PhD students :femalestudent:🧑‍🎓 across Europe around themes of Drones for conservation in Africa :zebraface:, with PhD topics from computer vision, to robotics and ecology! This also means that we need your help 🆘, dear members, to recruit.. can you help us out by sharing this link among your peers/students/friends? https://www.sdu.dk/en/service/ledige_stillinger/1200187

💯 Aleksis Pirinen, Oisin Mac Aodha, Eelke, Riccardo de Lutio, Valentin Gabeff, Casey Youngflesh, Cameron Trotter, gvanhorn, Robin Zbinden, Suzanne Stathatos, Josh Seltzer, Andrew Schulz, Carly Batist, Gaspard Dussert, Antón Álvarez, Sara Beery, Ariane Roberge

🙌 Aleksis Pirinen, Justin Kay, gvanhorn, Robin Zbinden, Ben Weinstein

🦁 Blair Costelloe, Aleksis Pirinen, Elijah Cole (Deactivated), Robin Zbinden

👍 Holger Klinck, Robin Zbinden, Jorrit van Gils

😍 Jon Van Oast, Robin Zbinden, Namrata Deka, Maxime Vidal, Nicolas Arrieta Larraza

Eelke (eelke@lumax.ai)

2022-11-17 04:19:58

Excellent news @Devis Tuia. Congrats!

🙏 Devis Tuia

😎 Jon Van Oast

Devis Tuia (devis.tuia@epfl.ch)

2022-11-17 05:59:09

This brings you directly to the PhD projects description: https://wilddrone.eu/recruitment/

WildDrone - Drones for Nature Conservation

Recruitment - WildDrone

Click here to Apply in SDU's Job Portal Join the WildDrone Team Are you highly motivated to conduct impactful research? Would you like to collaborate with a diverse group of scientists to improve wildlife conservation practices? We are looking for 12 Doctoral Candidates to join the “WildDrone - Autonomous Drones for Nature Conservation Missions” Marie-Skłodowska-Curie

Original URL: https://wilddrone.eu/recruitment/

👍 Aleksis Pirinen

Nick Giampietro (giampiet@pdx.edu)

2022-11-17 14:12:59

Hi @Devis Tuia, I am not a PhD student but I am highly interested in this work for my master's thesis. Thanks for sharing. I will likely reach out to the supervisors to establish correspondence with them. Is there anything I should know before contacting the advisors for DC6, DC8, or DC10? Thanks again!

Devis Tuia (devis.tuia@epfl.ch)

2022-11-18 03:28:22

I would say, check the different project details on wilddrone.eu so that you can have a clearer idea of what they are about and then contact people. We are hiring at the PhD level, but maybe the single institutions have capacity for extra master projects!

Maxime Vidal (mvidal@student.ethz.ch)

2022-11-18 12:02:28

Congrats, love to see it !

Nick Giampietro (giampiet@pdx.edu)

2022-11-18 12:05:53

I appreciate that Devis. I'll reach out to the individuals to discuss. Personally I probably won't be able to move for these opportunities at this time, so I'm mostly interested in just establishing contact and sharing our respective projects, perhaps collaborating on a paper along the way 😄 . Thanks again for the info!

👍 Devis Tuia

Ariane Roberge (arianeroberge13@gmail.com)

2022-12-07 13:44:55

@Myriam Cloutier

👍 Myriam Cloutier

Nick Giampietro (giampiet@pdx.edu)

2022-11-17 14:22:37

Hi folks, I am working on my master's thesis at Portland State University and investigating drone-assisted afforestation. I am especially interested in restoring forest that has been hit by wildfire 🌲🔥. While I'm interested in all aspects of it, my primary focus as of now is surveying terrain pre/post seed-sowing 🌱🌱, such creating a general utility model for desirable places to plant trees, or measuring the success of a seed-sowing operation (e.g. changes to leaf-area index, normalized difference vegetation index, changes to soil moisture, or counting wildlife in the area). Composing satellite data with finer-grained drone/UAV/aerial sensor data seems likely to happen! 🤯

Currently I'm in the background research phase, so I am exploring options for a specific problem to tackle.

To that end I have a few questions:

Is anyone working on something similar? I would love to establish correspondence with some people interested in the same topics. Just looking to meet people through something besides cold-emailing paper authors 🙂

In addition to my own Google Scholar alerts and article database searches, if anyone has seen a recent conference talk, paper, or anything like that which feels related, can you share? Not trying to crowdsource my own work, just to enrich it a bit by lightly picking brains, while I get up to speed 🙂

Similar question as above, but any recommendations for existing datasets to work with? X‐band, C‐ band, and L‐band microwave data, aerial images of wildfire-burnt land, and post sowing seedling growth images are all useful. Also interested in thermal imaging and other multispectral data.

Thanks in advance for any leads folks can share.

👍 Jose Ruiz-Munoz, Jon Van Oast, Dan Morris, Aleksis Pirinen, Adam Noach, Sara Beery, Michael Bunsen, Elaf Almahmoud

Jacob Kamminga (j.w.kamminga@utwente.nl)

2022-11-18 10:02:35

Hi all, I have funding for a small pilot project. I am looking for a postdoc who has experience in computer vision topics: detection, classification, tracking, and counting. We are looking for a unified approach and overview of state of the art and open issues for megafauna monitoring using drones, automatic nest box monitoring, and wildlife camera trapping. The funding is for 3 months, and the position is at Wageningen University (Netherlands) in collaboration with University of Amsterdam and University of Twente. You may be seconded by Wageningen at your current institution. Let me know if you are interested or may know someone!

👍 Jon Van Oast, Sara Beery, Subhransu Maji, Helena Russello, Tiziana Gelmi Candusso

Devis Tuia (devis.tuia@epfl.ch)

2022-11-18 10:03:42

did you ask to Gert Koostra at WUR?

Devis Tuia (devis.tuia@epfl.ch)

2022-11-18 10:04:13

@Helena Russello works in his group in case you need more info

Jacob Kamminga (j.w.kamminga@utwente.nl)

2022-11-18 10:05:50

Thanks, I will ask!

Helena Russello (helena@russello.dev)

2022-11-18 18:03:54

Feel free to ping me!

✅ Jacob Kamminga

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2022-11-18 20:12:33

This postdoc sounds exciting, what megafauna are you looking into?

Jacob Kamminga (j.w.kamminga@utwente.nl)

2022-11-22 09:32:31

Hi Tiziana, sorry for the slow reply (slack didn't notify me of the updates in this thread). The WuR research group working on drones is working in African game parks so expect anything from antelope to elephants. In the Netherlands they are aiming to count geese and deer using drones. For this small pilot we want to map the synergies between 3 related topics and investigate a unified approach. We probably found a person to work on the pilot, not 100% sure yet. PM me if you want to connect 🙂

❤️ Tiziana Gelmi Candusso

Michael Bunsen (notbot@gmail.com)

2022-11-18 14:43:31

Hi all! We have a job opportunity for a frontend developer to help build interfaces for machine learning tools for an automated insect monitoring project. I would love to work with one of y'all! Or please pass on to anyone you know who may be interested.

😎 Jon Van Oast

Anselm Bradford (ans@anselmbradford.com)

2022-11-18 23:42:26

Not directly addressing your job posting, but I notice you’re in Montreal. You should check out VT Tech Jam next year! A short drive to your south in Burlington, VT. https://techjamvt.com

Vermont Tech Jam

Talk with job recruiters, connect with tech professionals and learn about training programs from industry experts at this rockin’ career and tech event.

Est. reading time

51 minutes

Original URL: https://techjamvt.com

Michael Bunsen (notbot@gmail.com)

2022-11-22 21:54:22

Hey awesome, thanks @Anselm Bradford! We are actually partially funded by the Vermont Center for Ecostudies so already have one foot in Vermont!

🙂 Anselm Bradford

Anselm Bradford (ans@anselmbradford.com)

2022-11-23 00:45:53

Oh neat! Check out https://www.caryinstitute.org/about too. Maybe relevant resources there also.

Cary Institute of Ecosystem Studies

About Cary Institute of Ecosystem Studies

Cary Institute is one of the world’s leading independent environmental research organizations. Our staff have a deep understanding of the key drivers of ecological change, among them climate change, invasive species, and human influence on land and water.

Original URL: https://www.caryinstitute.org/about

👍 Michael Bunsen

🙏 Michael Bunsen

Luke Sheneman (sheneman@uidaho.edu)

2022-11-18 17:46:16

Waterproof paint can be used to mark animals/insects in mark-recapture studies. Does anybody have experience using mark-recapture by application of Infrared-reactive paints and then later detecting with IR cameras? If so, I would be interested in knowing more about options for IR-absorbant/reflective paint. Many thanks!

Dan Morris (agentmorris@gmail.com)

2022-11-18 19:42:56

Out of curiosity, why the extra effort to use IR-reactive paint, instead of just visible paint? Is this a case where you expect that paint within the visible spectrum of the animal (or other animals) will have a behavioral/survival impact?

Nick Giampietro (giampiet@pdx.edu)

2022-11-20 20:52:10

Just guessing, but it might be more reliable to blob count in a spectrum where you're unlikely to have false positives. I'm interested in hearing the reason too though (along with any potential paint recommendations)

Frederic (frederic@apic.ai)

2022-11-21 10:40:05

We had some Ideas to use glitter as unique finger print for bees. Just spray them and the glitter pattern will be unique for every bee. The concept worked as a proof of concept. But in practice capturing the patterns on moving bees was not as easy, so we just glued markers on them. (MNIST for bees). Even though it was way way way more work, then just spray glitter on bees.

Side note:
I used UV reactive paint for another use case, trace the movement of bees pollinating and could share some details if you are interested.

👍 Valentin Ștefan

Luke Sheneman (sheneman@uidaho.edu)

2022-11-21 17:13:05

@Dan Morris, @Frederic , @Nick Giampietro - Yeah, it is for AI-controlled automated mark recapture in a controlled dark environment. We also don't want to impact normal predation by making the animal stand out more. The perfect "paint" would be IR reflective/absorbent but also otherwise invisible. Such a thing might not exist?

Although it would be an added bonus, we don't necessarily need to identify unique individuals, just recognize if we've seen the animal before. Frederic - I would be interested to know what paint you used for your bee work. Your glitter idea for bees is fascinating, but I can see how that might be challenging in practice!

Nick Giampietro (giampiet@pdx.edu)

2022-11-22 14:53:13

Yeah, it's probably hard to find a pigment that reflects IR without also reflecting more reds as well, since spectral radiance tends to be some number of smooth curves with a peak at a given frequency and, well, IR is close to red.

This article has some useful info, and even some pigments that look close to your requirements... Can't speak to whether they would be safe to use though

https://www.pcimag.com/articles/102920-ir-reflective-pigments-a-black-rainbow-of-options

pcimag.com

IR-Reflective Pigments: A Black Rainbow of Options

IR pigments allow paint formulators and coatings chemists to develop differentiated and functional coatings.

Original URL: https://www.pcimag.com/articles/102920-ir-reflective-pigments-a-black-rainbow-of-options

👍 Luke Sheneman

Michael Bunsen (notbot@gmail.com)

2022-11-23 12:43:34

If you find a way to mark the animal without it being visible to a non-IR camera, then you will be able to make ground-truth data for detecting individuals from normal RGB images as well.

Nicolas Arrieta Larraza (n.arrieta.larraza@gmail.com)

2022-11-27 15:09:38

Hi everyone!

I am brand new on the community thanks to @Devis Tuia. I recently watched his talk at AI for Good and I was amazed by his work. Moreover, he gave me hope and evidence that AI can definitely be used to support ecology and wildlife ❤️🌱

I recently graduated of a MSc in Data Science at the University of Twente and currently work as a ML engineer with audio data. My MSc thesis explored Deep Few-Shot Learning models for Acoustic Scene Classification.

I have always been looking for opportunities to apply AI for this specific topic. It was a few weeks ago when I found about AI for Good, and consequently about Devis' work, which has lead me here. I got to say I am unreasonably excited about the things that I have been reading on the Slack channels 😊

I am particularly interested in the use of audio data to support ecology and wildlife protection. I am currently researching on past projects, breakthroughs and resources. Do you have any recommendations? Any must that I should know/read about? Common use cases?✨

Lastly, I am actively looking for opportunities to contribute with my knowledge. Do you happen to know of any project that requires volunteers? (It does not have to be audio-related)

Looking forward to engage into interesting conversations and contribute in the field!

Cheers.

👋 Omiros Pantazis, Dan Morris, Sofía Miñano

Elijah Cole (Deactivated) (ecole@caltech.edu)

2022-11-27 18:22:29

@Carly Batist @Taiki Sakai - NOAA Affiliate Might have some pointers 🙂

❤️ Nicolas Arrieta Larraza, Carly Batist, Taiki Sakai - NOAA Affiliate

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-11-27 19:27:04

Hi! I do PAM with lemurs in Madagascar as well as some soundscape indices/analysis type stuff. I use ML models to detect lemur calls, with the huge help of my computer science expert collaborators @Emmanuel Dufourq and Lorene Jeantet! I have some materials on my website regarding PAM, including a list of good papers to look at for different use cases which might be helpful for you. You may also want to check out the Conservation Tech Directory to search for projects, tools, organizations, etc in that space (filter/search by audio, passive acoustic, etc). Happy to chat further over DM or hop on a zoom if you’d like to talk further!

conservationtech.directory

Computational bioacoustics with deep learning: a review and roadmap

Visit the post for more.

Original URL: https://conservationtech.directory/

❤️ Nicolas Arrieta Larraza

Devis Tuia (devis.tuia@epfl.ch)

2022-11-28 02:55:19

@Holger Klinck works a lot with audio as well!

❤️ Nicolas Arrieta Larraza, Carly Batist

Rita Pucci (rita.pucci85@gmail.com)

2022-11-28 11:11:40

also, Dan Stowell is working on audio and biodiversity

❤️ Nicolas Arrieta Larraza, Carly Batist

Taiki Sakai - NOAA Affiliate (taiki.sakai@noaa.gov)

2022-11-28 13:14:07

Hi Nicolas! Our lab uses passive acoustics to study marine mammals. Audio data is used a lot in our field since sound travels so far under water (among other reasons). AI/ML is definitely being used more and more, Dan Stowell's name has already been mentioned here but if you haven't read his review paper on bioacoustics and deep learning that's an excellent starting point.

Happy to chat if you have any marine mammal questions!

arXiv.org

Animal vocalisations and natural soundscapes are fascinating objects of study, and contain valuable evidence about animal behaviours, populations and ecosystems. They are studied in bioacoustics...

Original URL: https://arxiv.org/abs/2112.06725

❤️ Nicolas Arrieta Larraza, Carly Batist

Nicolas Arrieta Larraza (n.arrieta.larraza@gmail.com)

2022-11-28 13:30:57

Such a warm welcoming! Thanks a lot to all of you for the info, I am going to look into detail at the resources you shared and will keep in touch 😊

Antonio Ferraz (antonio.a.ferraz@jpl.nasa.gov)

2022-11-29 20:11:13

@Nicolas Arrieta Larraza, see my publication in the main channel posted today; we do have tons of audio analyses to do and potential future opportunities.

❤️ Nicolas Arrieta Larraza

Silvia Zuffi (silvia@mi.imati.cnr.it)

2022-11-30 12:35:08

https://twitter.com/mmbronstein/status/1597656746767822848

🐋 Taiki Sakai - NOAA Affiliate, Nicolas Arrieta Larraza

Holger Klinck (hk829@cornell.edu)

2022-11-30 16:44:46

Hi Nicolas, if you have any specific questions, please reach out. We are working on a bunch of marine and terrestrial projects. Here is one of them: https://birdnet.cornell.edu/ We also released a Python package for machine listening: https://shyamblast.github.io/Koogu/en/stable/ Cheers!

Jinsu Elhance (jelhance@gmail.com)

2022-11-29 13:18:47

I'm curious if anyone is working on determining the social and economic disadvantages of using AI models trained on publicly accessible data (potentially low resolution, scarce) when communities are assessing the value of protecting their land. I'm thinking particularly about pricing in Carbon Markets where higher accuracy models may allow markets to exploit land owners by withholding key information from them. If anyone has any case studies of conservation market exploitation through the digital divide I'd love to hear more.

👀 Declan, Josh Seltzer, Yseult Hb

Antonio Ferraz (antonio.a.ferraz@jpl.nasa.gov)

2022-11-29 20:02:56

Potential Project Collaboration: Hi, everyone. I’m looking for an ecologist/ornithologist with experience in bioacoustics to collaborate on a project integrating on-the-ground acoustics with synchronous NASA airborne measurements (spectroscopy and LiDAR). We developed a proof-of-concept showing significant relationships between acoustic diversity and habitat (structural and spectral) diversity. We are looking for ecologists to interpret our acoustic diversity results across habitat types and further discuss additional analyses and methods that might help to monitor (habitats and wildlife) biodiversity from ground, air and/or space.

😎 Jon Van Oast

🙌 Stephanie O'Donnell, Toryn Schafer

Antonio Ferraz (antonio.a.ferraz@jpl.nasa.gov)

2022-11-29 20:03:03

The focus is on Mediterranean ecosystems, particularly in California and the Cape Floristic Region (South Africa). We are interested in studying acoustic diversity across space, such as environmental and fire history gradients. There is also potential for temporal analysis. For example, we developed an acoustic network to collect data during springtime (March-July) in a Californian Natural Preserve. The acoustic campaign pairs with weekly NASA airborne spectral measurements provinding unprecedented spatial, spectral and temporal resolutions to measure habitat composition, phenology and plant trait variation over time. It is a unique dataset for studying how sounds and colors change during the growing season and whether they are tied.

Antonio Ferraz (antonio.a.ferraz@jpl.nasa.gov)

2022-11-29 20:03:12

We are primarily looking for faculty professors from the following universities (see below) due to the limitations of an upcoming NASA-JPL funding opportunity. Still, any discussion, insights or ideas for future collaborations are welcome; our research work goes beyond the referred funding opportunity.

Antonio Ferraz (antonio.a.ferraz@jpl.nasa.gov)

2022-11-29 20:03:15

Arizona State University
Carnegie Mellon University
Cornell University
Georgia Institute of Technology
Massachusetts Institute of Technology
Princeton University
Stanford University
Texas A&M University
The University of Texas at Austin
University of Arizona
University of California, Los Angeles
University of Colorado, Boulder
University of Michigan
University of Southern California

G. Andrew Fricker (africker@calpoly.edu)

2022-12-14 00:19:24

Hey Tony!! @Antonio Ferraz I know Cal Poly San Luis Obispo didn't make the NASA/JPL cut, but I know an ecologist here who might be able to do what you describe. I have a colleague at Cornell who might know some bird folks there. Let me know if you'd like to pursue any of those opportunities. Hope all is well.

Antonio Ferraz (antonio.a.ferraz@jpl.nasa.gov)

2022-12-15 16:03:50

hey, how are you? yes, let’s reconnect shortly

G. Andrew Fricker (africker@calpoly.edu)

2023-02-10 22:59:48

Hey man, doing good, busy busy. How about yourself?

G. Andrew Fricker (africker@calpoly.edu)

2023-02-10 23:00:04

Let me know if you want to connect. im sure you probably found your biologist by now.

Ben Weinstein (benweinstein2010@gmail.com)

2022-11-29 21:47:24

Is anyone in our community familiar with bird detection in NEXRAD data? We've been just testing out some proof of concepts for detecting large bird rookeries in the everglades.

Ben Weinstein (benweinstein2010@gmail.com)

2022-11-29 21:51:07

maybe, @Dan Sheldon

Dan Sheldon (sheldon@cs.umass.edu)

2022-11-30 08:27:11

Haha, yes, have been working on this for years. Also with @Subhransu Maji. What kind of birds? We have an ongoing project for detecting and tracking swallow roosts (AAAI, bioRxiv). Swallows (also bats) form the most distinctive patterns on radar, but many other species have related behaviors that also show up on radar.

arXiv.org

Detecting and Tracking Communal Bird Roosts in Weather Radar Data

The US weather radar archive holds detailed information about biological phenomena in the atmosphere over the last 20 years. Communally roosting birds congregate in large numbers at nighttime...

Original URL: https://arxiv.org/abs/2004.12819

👍 Subhransu Maji

Ben Weinstein (benweinstein2010@gmail.com)

2022-11-30 13:00:11

Thanks. This is all very preliminary, as part of a future grant proposal, so just trying to sketch out what is possible. Briefly, the everglades restoration act is spending about a billion dollars to reform large waterways and canals in south florida. One of the metrics of success is wading bird colony population numbers. These are large aggregations of nesting birds, anywhere from several hundred individuals to 25,000 megacolonies. We cover large areas (>1000 sq km, the WCA1,2,3 in the map) using piloted aircraft and drones, and have been looking for an early warning system to be able to locate colonies so we can send aircraft there. We have solid machine learning models for the optical data and 30 years of historic data that we could associate with locations of colonies that could serve as potential supervised labels. My question with nexrad is 1) What kind of spatial resolution can we expect in the 'legacy' (not dual-pol) stations. KAMX is right next to our sites. When we see dots in the nexrad, is that an individual bird? a flock? What kind of target size is it picking up? 2) i've seen this mostly applied to martins and other species with a single very large signature at dawn. The herons/egret/ibis are larger, but the pattern will be more chaotic as they come and go from the colony many times a day. Does this seem feasible? I'm imagining a workflow for detection colonies of high activity 1) getting nextrad data, 2) traditional background subtraction to remove noise (why is there always a ring of responses around the detector?) 3) look for patterns of presence in the biological reflectance range (i've read 20-30db with low cross-correlation ratios from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.1539). 4) temporal tracking of objects to perform optical flow, 5) flow patterns coming in an out of potential colonies using a recurrent neural network (not enough temporal coverage?). Happy to jump on zoom if its easier. I'm playing around here https://github.com/weecology/everglades_radar/blob/main/Download.ipynb

<https://github.com/weecology/everglades_radar/blob/main/Download.ipynb | Download.ipynb>

Ben Weinstein (benweinstein2010@gmail.com)

2022-11-30 13:01:40

Thread Reply:

Subhransu Maji (smaji@cs.umass.edu)

2022-11-30 13:28:53

Super cool. The scans in the notebook look somewhat different from some of the roosts we have been looking at (they look more like nocturnal migration). You could possibly use MistNet to separate rain from biology that works on non-dual pol data (e.g., cross-correlation is a dual-pol product). https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/2041-210X.13280

Ben Weinstein (benweinstein2010@gmail.com)

2022-11-30 13:31:51

interesting. I see the cross_correlation on this station, even though it is labeled legacy.

radar.fields.keys()
dict_keys(['reflectivity', 'velocity', 'differential_reflectivity', 'cross_correlation_ratio', 'differential_phase', 'clutter_filter_power_removed', 'spectrum_width'])

Subhransu Maji (smaji@cs.umass.edu)

2022-11-30 13:34:10

There were two updates to NEXRAD if I remember (one was resolution update, and another was dual-pol). This could the pre-resolution update@Dan Sheldon ?

Ben Weinstein (benweinstein2010@gmail.com)

2022-11-30 13:40:17

Are the detections directly around the sensor artifacts? I am often seeing that pattern of heavy detections in a ring then dropping of, what's going on there?

Subhransu Maji (smaji@cs.umass.edu)

2022-11-30 14:59:31

The radar beam points upwards at an angle — so most detections are near the radar as the migration is most intense at lower elevations.

Ben Weinstein (benweinstein2010@gmail.com)

2022-11-30 15:00:26

thanks, so they should be seen as genuine detections, and the detection probability drops off. Rather than artifacts that should be filtered.

Ben Weinstein (benweinstein2010@gmail.com)

2022-11-30 15:01:42

There isn't a MistNet for python? If i do get this grant, I'd be interested in porting it to pytorch. I maintain a few pytorch ecological related models -> https://deepforest.readthedocs.io/

Subhransu Maji (smaji@cs.umass.edu)

2022-11-30 15:24:32

There is a mistnet python (not sure it’s public facing yet)

👍 Ben Weinstein

Ben Weinstein (benweinstein2010@gmail.com)

2022-11-30 15:25:04

no rush, this is not at all pressing, just folding prelim data into grants.

Subhransu Maji (smaji@cs.umass.edu)

2022-11-30 15:27:27

Btw, Dan and I would be happy to chat sometime. Good luck with the proposal (end of semester here, so rather busy with all that)

Ben Weinstein (benweinstein2010@gmail.com)

2022-11-30 15:31:23

thanks, i'll leave a gif I just made here (wow matplotlib takes awhile to make animations). Thresholded reflectance above 0, cross correlation below 0.75. Definitely some action in there. Too early to say what happens, but some massive explosions around midday, that end quickly (unlikely to be weather?). Black cross are known historic bird colonies (not active every year).

Dan Sheldon (sheldon@cs.umass.edu)

2022-11-30 16:30:18

Exciting! Big questions are the extent to which the target species show up on radar, how they appear, and the extent to which they can be separated from other sources of reflectivity, including weather and also other species. This would depend a lot on both bird behavior and radar characteristics

Behavior: how concentrated are they while roosting, how synchronized/organized are their flights, especially how high do they fly.

Radar characteristics: mostly how far are the colonies from radar, which determines how high the beam is.

In your animation I see a big exodus from ~10:40–11:30 UTC (6:40–7:30 EDT), which is right around local sunrise (7:11 EDT). This looks very much like rather widespread activity of birds taking off from those areas right around sunrise. The areas are highly correlated with your colony markers but the reflectivity is widespread, i.e., coming pretty evenly from the whole surrounding area and not closely concentrated around the markers. I would guess it could include lots of different species, e.g., waterfowl, in addition to the waders. From this example it looks very encouraging that you could measure “morning bird activity in the Everglades” but depending on the mix of species and their relative abundance it’s unclear the extent to which you could measure a specific group of species. (This is only one example, of course!)

This is from 2021, so would have dual-pol data. The dual-pol upgrade was in 2012-2013.

Resolution is 250m x 0.5 degrees. The radar is sensitive enough to detect one large-bodied bird in a volume that size, but the usual situation is many birds in one volume.

Happy to chat some time.

Ben Weinstein (benweinstein2010@gmail.com)

2022-11-30 16:38:29

Definitely, thanks for your time. I'm using these filters, i'll post one more when its done. Very slow animations building in matplotlib. I had thought it was mosquitos or non-filtered weather. I'm going to wrap this into some functions and create a sample graph throughout the year to get a sense of the 'background bird activity' in the everglades and see if we see any additional boom as the birds come in.

    gate_filter.exclude_below("reflectivity",5)
    gate_filter.exclude_above("reflectivity",30)
    gate_filter.exclude_below("differential_phase",30)
    gate_filter.exclude_above("cross_correlation_ratio",0.8)

based on figure from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.1539

Dan Sheldon (sheldon@cs.umass.edu)

2022-11-30 16:39:51

By the way if you just want to view the data, NOAA’s weather and climate toolkit is the easiest way. It’s a bit arcane but has good functionality, allows you to browse the inventory, select files, make animations, etc.: https://www.ncdc.noaa.gov/wct/install.php.

ncdc.noaa.gov

NOAA's Weather and Climate Toolkit (Viewer and Data Exporter)

NOAA NCDC's Weather and Climate Toolkit - Visualization and Export of Weather Radar, Satellite and Drought Data. Download/Installation page.

Original URL: https://www.ncdc.noaa.gov/wct/install.php

👍 Ben Weinstein

Dan Sheldon (sheldon@cs.umass.edu)

2022-11-30 16:45:06

For thresholding with dual-pol, people usually just use correlation coefficient < 0.95 = biology, otherwise weather.

Olof Mogren (olof.mogren@ri.se)

2022-11-30 04:33:59

A well-known figure here, @Sara Beery is giving a talk in our Learning Machines Seminar tomorrow Thursday, at 15:00 CET. Auto Arborist: Towards Mapping Urban Forests Across North America. The seminar is free and open to all, and you just connect using Zoom, no registration necessary! For info about upcoming seminars, there is an optional mailing list.
https://www.ri.se/en/learningmachinesseminars/sara-beery-mit-urban-forest-mapping

RISE

Sara Beery, MIT: Urban Forest Mapping | RISE

At RISE Learning Machines Seminar on Dec 1, we have the pleasure to listen to Sara Beery from MIT give her talk: "Auto Arborist: Towards Mapping Urban Forests Across North America". – Generalization to novel domains is a fundamental challenge for computer vision. We propose urban forest monitoring for capturing real world complexity.

Original URL: https://www.ri.se/en/learningmachinesseminars/sara-beery-mit-urban-forest-mapping

:thumbsup_all: Frederic Fol Leymarie, Andrew Schulz, Aleksis Pirinen, Toryn Schafer, Nicolas Arrieta Larraza, Valentin Gabeff, Jason Holmberg (Wild Me), Jose Ruiz-Munoz, Yseult Hb

👍 Shir Bar, Aleksis Pirinen, Taiki Sakai - NOAA Affiliate, Jason Holmberg (Wild Me), Victor Anton, John Martinsson

👍:skin_tone_3: Pen-Yuan Hsing

😎 Jon Van Oast, David McClosky, Jason Holmberg (Wild Me), Eddie Zhang

🌲 Ben Weinstein, Jason Holmberg (Wild Me), Yseult Hb

Olof Mogren (olof.mogren@ri.se)

2022-11-30 04:36:22

We also have a series of recordings of previous seminars on youtube, with speakers such as @Devis Tuia and Frederik Kratzert which are definitely worth checking out:
https://www.youtube.com/watch?v=eP79uTWTTjY&list=PLqLiVcF3GKy1tuQFoDu5QKOM6S33t_4R1&index=8&t=1063s https://www.youtube.com/watch?v=Wrv-IS3wf80&list=PLqLiVcF3GKy1tuQFoDu5QKOM6S33t_4R1&index=1

YouTube

RISE Research Institutes of Sweden (https://www.youtube.com/@RiSeSweden)

Devis Tuia: Machine learning supporting ecology supporting machine learning

Original URL: https://www.youtube.com/watch?v=eP79uTWTTjY&list=PLqLiVcF3GKy1tuQFoDu5QKOM6S33t_4R1&index=8&t=1063s

YouTube

RISE Research Institutes of Sweden (https://www.youtube.com/@RiSeSweden)

Frederik Kratzert: Rainfall-runoff modelling

Original URL: https://www.youtube.com/watch?v=Wrv-IS3wf80&list=PLqLiVcF3GKy1tuQFoDu5QKOM6S33t_4R1&index=1

👍 Aleksis Pirinen, Omiros Pantazis

Olof Mogren (olof.mogren@ri.se)

2022-12-08 04:01:12

For those of you who missed this very interesting talk by @Sara Beery, the recording is on youtube! https://www.youtube.com/watch?v=Ob7XfUPmKu4&list=PLqLiVcF3GKy1tuQFoDu5QKOM6S33t_4R1&index=1&t=891s

YouTube

RISE Research Institutes of Sweden (https://www.youtube.com/@RiSeSweden)

Sara Beery: Urban forest mapping

Original URL: https://www.youtube.com/watch?v=Ob7XfUPmKu4&list=PLqLiVcF3GKy1tuQFoDu5QKOM6S33t_4R1&index=1&t=891s

❤️ Sara Beery

Majid Mirmehdi (m.mirmehdi@bristol.ac.uk)

2022-12-01 12:27:56

AI for Nature Conservation - Fully funded PhD positions at Bristol University and other member institutions in the Wilddrone.eu project! See https://wilddrone.eu/recruitment/

WildDrone - Drones for Nature Conservation

Recruitment - WildDrone

Original URL: https://wilddrone.eu/recruitment/

❤️ Lucia Gordon, Josh Seltzer, Devis Tuia

🐘 Blair Costelloe

Devis Tuia (devis.tuia@epfl.ch)

2022-12-02 02:29:33

of someone needs information about the project (we have 13 phd projects after all!), the application process, etc, please reach out to me, @Blair Costelloe or @Tilo Burghardt !

Ameya Patil (ameyapatil249@gmail.com)

2022-12-02 08:11:02

Hello everyone! I am Ameya Patil, PhD student at the University of Washington, Seattle advised by Dr. Leilani Battle. My research area is in databases and data visualization, along with a motivation to work towards environmental health. Accordingly, I am working on building interactive data analytics systems for environmental science data or contexts - weather/climate, animal tracking, waste management, sustainability, etc. I would be happy to chat if anyone has any such datasets for which they would like to build a exploration and analysis interface. Also, I am looking for internships for Summer 2023 in the same domain, so I would appreciate any leads on that front!

👋 Jon Van Oast, Dan Morris, Sara Beery

👏 Vanessa Suessle

Wenqi Su (wenqi.su@mcgill.ca)

2022-12-03 19:21:36

Hi everyone! My name is Wenqi Su. I am currently majoring in computer science and statistics at McGill University. I am super interested in the socially impactful application of computer science and actively looking for research opportunities to get involved in this domain as early as possible. I am currently interning as a data scientist in industry and working as a RA and TA for computer classes at McGill. I already took ML and AI courses and have a solid understanding of the core concepts. I am eager to discover how to apply theoratical knowlegde to application through research. If anyone need extra help for their projects/research and is willing to bring me in, trust me I will only be helful and not a burden even though I am just an undergrad!!

❤️ Carly Batist, Sara Beery, Eddie Zhang

👍:skin_tone_3: Pen-Yuan Hsing

👍 Kangyu Zheng, Michael Bunsen

:thumbsup_all: Frederic Fol Leymarie

Kostas Papafitsoros (k.papafitsoros@qmul.ac.uk)

2022-12-05 10:22:17

I have a fully funded PhD position in Queen Mary University of London, on the topic “Data-driven Image Processing Methods with Applications to Wildlife conservation”. Deadline: 31 January 2023. I am happy to answer any questions that you might have 🙂
https://www.qmul.ac.uk/maths/postgraduate/postgraduate-research/phd-projects/current[…]rocessing-methods-with-applications-to-wildlife-conservation/

👍 Sara Beery, Lukas Picek, Alessandra Sellini, Malte Pedersen, Alexander Robillard, Carly Batist, Vanessa Suessle

🙌 Omiros Pantazis

🎉 Lukas Picek, Alexander Robillard, Aleksis Pirinen

karen bakker (karen.bakker@ubc.ca)

2022-12-06 09:48:26

Just wanted to share some exciting news: my new book, which explores the implications of AI for analyzing non-human communication, has just been published: The Sounds of Life: How Digital Technology is Bringing Us Closer to the World of Animals and Plants (Princeton University Press, October 2022).

The book synthesizes decades of research (including work by the members of the AI for Conservation community), explores applications to conservation, and discusses potential implications for digitally-mediated interspecies communication. The book got a full page review in Science, was covered in The Guardian, and was chosen as the book of the month by the NPR Science Friday Book Club.

If anyone is interested in learning more or would like to host a talk or webinar, please let me know!

🙌 Stephanie O'Donnell, Josh Seltzer, Declan, Robin Zbinden, Sara Beery, Viktor Domazetoski, Katelyn Morrison, Talia Speaker, Yseult Hb, Ted Schmitt, Jason Holmberg (Wild Me), Matt Weldy, Drew Blount, Lindsey Dukles, Ameya Patil

👍 gvanhorn, Omiros Pantazis, Justin Kay, Katelyn Morrison, Jason Holmberg (Wild Me)

😎 Jon Van Oast, Jason Holmberg (Wild Me)

Sankaran (shun-ka-run) (sankaranv@cs.umass.edu)

2022-12-06 16:53:29

Hi everyone! My name is Sankaran, and I am a PhD student at UMass Amherst working in causal inference, probabilistic ML, and RL. I am looking for ways to apply ideas from these fields to problems in human-elephant conflict and broadly in wildlife conservation. Excited to be involved in this group (thanks to @Elizabeth Bondi)! From the few papers I have read in this field, the idea of trying to approach conservation problems on my own felt very daunting, and I feel lucky to have found this community. By reading all of your work, I am hoping to learn more about how wildlife conservation problems can be cast as AI problems and how data from the field is collected, so I can accordingly direct my future research!

I am hoping to contribute to projects related to elephant conservation or human-animal conflict, and would also appreciate any pointers, readings or recommendations for how I can get started! I’m also happy to chat about anything related to causal inference and experiment design, if there ways you think they can be useful in conservation problems.

👋 gvanhorn, Elizabeth Bondi, Josh Veitch-Michaelis, Katelyn Morrison, Nicolas Arrieta Larraza

Anselm Bradford (ans@anselmbradford.com)

2022-12-06 18:17:47

Are you set on elephants? I’m wondering if maybe there could be large animal conservation projects regionally closer to Umass that might make the logistics easier for you. For instance the Wolf Conservation Center in South Salem, NY, ~2hr drive from you. I have no affiliation, but just a thought of something to look into. There might be similar regionally-relevant issues to explore in regard to bobcats, coyotes, bears, etc. that conflict with human activities.

Sankaran (shun-ka-run) (sankaranv@cs.umass.edu)

2022-12-06 18:23:22

I think elephants were where I started looking into wildlife conservation (from growing up in South India) but I'm not set on it! These are actually really great, having resources and expertise in proximity is very helpful when I'm still new.

👍 Anselm Bradford

Amber De Neve (adeneve@umass.edu)

2022-12-12 16:01:39

Hi Sankaran- I am a PhD student at UMass Amherst too! I actually work on plant monitoring using ML, but happy to chat if you ever want to brainstorm stuff with a biologist

Timm Haucke (timm@haucke.xyz)

2022-12-07 12:37:41

Hi everyone, there is a new camera trapping dataset on lila.science: Lindenthal Camera Traps. This dataset was captured in a wildlife park by a stereo camera, which provides not only color / IR, but also distance information. Feel free to contact me if you have any questions about the dataset. Thanks @Dan Morris for the hosting and help!

LILA BC

Lindenthal Camera Traps - LILA BC

Overview This data set contains 775 video sequences, captured in the wildlife park Lindenthal (Cologne, Germany) as part of the AMMOD project, using an Intel RealSense D435 stereo camera. In addition to color and infrared images, the D435 is able to infer the distance (or “depth”) to objects in the scene using stereo vision. Observed...

Written by

lilawp

Est. reading time

2 minutes

Original URL: https://lila.science/datasets/lindenthal-camera-traps/

🎉 Jon Van Oast, Josh Veitch-Michaelis, Dan Morris, Peter Bull, Rowan Converse, Alexander Robillard, Ando Shah, Viktor Domazetoski, Sara Beery, Carly Batist, Kakani Katija, Vanessa Suessle, Tiziana Gelmi Candusso, Sofía Miñano

👍 Atul Ingle, Rowan Converse, Alexander Robillard, gvanhorn, Jose Ruiz-Munoz, Peter van Lunteren, Vanessa Suessle, Ștefan Istrate, Shir Bar, Rita Pucci

💯 Valentin Gabeff

👍:skin_tone_3: Pen-Yuan Hsing

Peter Bull (peter@drivendata.org)

2022-12-07 12:54:59

Awesome, congrats on the release @Timm Haucke!

🙌 Timm Haucke

Ando Shah (ando@berkeley.edu)

2022-12-07 15:13:50

Hi folks! Im a second year PhD student at UC Berkeley, relatively new to conservation, and applying ML to conservation problems (currently developing deep learning SDMs) that can be operationalized quickly. Im wondering:

What are some of the journals / conferences that computer science folks working in conservation and conservation policy should be paying attention to?

How do most folks deal with the interdisciplinary nature of this work - where do they feel their academic home community is - is it in their original domains (CS, Conservation Biology, Policy, etc), or is there a new interdisciplinary home emerging?

😍 Sara Beery, Rebecca

👍 Carly Batist, Peter van Lunteren, Carl Boettiger, Shir Bar

Katelyn Morrison (kcmorris@andrew.cmu.edu)

2022-12-07 15:15:05

ACM COMPASS is one conference that is relevant https://compass.acm.org/. I've also seen a lot of CompSust work shared at AAAI and IJCAI venues

🥳 Ando Shah

👍 Boyu Zhang

Carl Boettiger (cboettig@berkeley.edu)

2022-12-07 15:57:48

Hey @Ando Shah, nice to see you! On the conservation side, high-profile work is still concentrated in Science/Nature/PNAS which all obviously have interdisciplinary readership, though there are many discipline specific journals like Conservation Letters, Conservation Biology, or OneEarth (the relatively new flagship Cell press journal in this area). Conference-wise, AGU is a fair-sized conference and probably a nice venue for such work (long a focus of geospatial informatics), though this would be well-received at conferences like ESA and narrower ones like ISEC https://imstat.org/meetings-calendar/the-6th-international-statistical-ecology-conference/.

🙌:skin_tone_5: Ando Shah

Dan Morris (agentmorris@gmail.com)

2022-12-07 20:10:44

This sort of the opposite of what you asked about, but Methods in Ecology and Evolution has been very friendly to papers about new ML techniques for conservation data processing:

https://besjournals.onlinelibrary.wiley.com/journal/2041210x

If I had to generalize based on what I've seen, I would say journals like Methods in E&E that are interdisciplinary but "ecology-first" have a high bar for usefulness, but don't need to see algorithmic innovation; CS venues like AAAI/CVPR/etc. have an inherent high bar for innovation but are more comfortable with domain-specific case studies that aren't necessarily practical yet (i.e., "our technique could be applied to this interesting conservation problem").

😎 Carl Boettiger

❤️ Emilio Luz-Ricca, Ando Shah

Rebecca (rebeccayap92@uchicago.edu)

2022-12-08 00:49:25

Not an answer to your question, but I love the second question especially because I am dealing with this and trying to pivot and wondering what degree to place myself in.

Ando Shah (ando@berkeley.edu)

2022-12-09 13:49:44

Thank you all for those super helpful thoughts and insights!

Olof Mogren (olof.mogren@ri.se)

2022-12-08 02:16:01

Today, William Lidberg from Swedish University of Agricultural Sciences is giving a talk on AI-based forest landscape mapping using remote sensing. 15:00 CET on Zoom. https://www.ri.se/en/learningmachinesseminars/william-lidberg-slu-geographical-intelligence

RISE

William Lidberg, SLU: Geographical intelligence | RISE

At RISE Learning Machines Seminar on Dec 8, we have the pleasure to listen to William Lidberg, SLU, give his talk: "Geographical intelligence" – My research focuses on using AI to map the forest landscape but I also ask questions about social and ethical challenges.

Original URL: https://www.ri.se/en/learningmachinesseminars/william-lidberg-slu-geographical-intelligence

🙌 Aleksis Pirinen, Josh Seltzer, Michael Bunsen, Jon Van Oast

😍 Sara Beery

🙌:skin_tone_5: Ando Shah

Sean Nachtrab (sean.nachtrab@gmail.com)

2022-12-08 10:38:14

This was interesting to listen to! Thank you

Olof Mogren (olof.mogren@ri.se)

2022-12-09 07:42:01

Good to hear @Sean Nachtrab! For those of you who missed this inspiring talk about an important application of state-of-the-art AI (maps of soil and wetlands using LIDAR data and machine learning), it's available on youtube! https://www.youtube.com/watch?v=Ezwgsjh2Oh8&list=PLqLiVcF3GKy1tuQFoDu5QKOM6S33t_4R1&index=1

YouTube

RISE Research Institutes of Sweden (https://www.youtube.com/@RiSeSweden)

William Lidberg: Geographical intelligence

Original URL: https://www.youtube.com/watch?v=Ezwgsjh2Oh8&list=PLqLiVcF3GKy1tuQFoDu5QKOM6S33t_4R1&index=1

Vanessa Suessle (vanessa.suessle@h-da.de)

2022-12-08 07:55:04

Hello all.
I am a PhD student from Darmstadt, Germany and I just started.

My topic is 'AI applications for wildife conservation' with a focus in Computer Vision and Individual Identification for fur-patterned species.

During my master's thesis I tried a first approach to automatically identify leopards from video data without pre-labeled data.
My plan is to build up on this and make it more robust and useable.

I am always happy to meet new people and a have a chat on CV and AI in wildlife conservation.

👋 Timm Haucke, Josh Seltzer, Jose Ruiz-Munoz, Nicolas Arrieta Larraza, Jason Holmberg (Wild Me), Taiki Sakai - NOAA Affiliate, Katelyn Morrison

Josh Seltzer (jyseltz@gmail.com)

2022-12-08 08:17:07

Welcome Vanessa! I've just started working on a project to identify jaguars and other Central American felines in video data, so I'd love to connect some time 🙂

👍 Vanessa Suessle

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2022-12-08 12:26:13

Hi @Josh Seltzer and @Vanessa Suessle At Wild Me we have data and benchmarks that might help for jaguars, leopards, cheetah, and snow leopards. Imagery not video. But happy to collaborate on this.

🙌 Josh Seltzer, Taiki Sakai - NOAA Affiliate

Josh Seltzer (jyseltz@gmail.com)

2022-12-08 13:20:29

Hey @Jason Cremerius that sounds awesome! I had been looking at the Leopard ID 2022 dataset and wondering how well that might support transfer learning to jaguars, it'd be a pleasure to collaborate 🙏

👍 Jason Holmberg (Wild Me)

Vanessa Suessle (vanessa.suessle@h-da.de)

2022-12-09 03:59:08

Hi @Josh Seltzer and @Jason Holmberg (Wild Me) I would be very happy to collaborate as well.
The ideas sound great.
I would love to chat with you on those topics. I am in CET timezone. Feel free to suggest a time and date.

Josh Seltzer (jyseltz@gmail.com)

2022-12-16 17:57:44

@Vanessa Suessle hey, sorry for the delayed response! I am pretty flexible next week, I can do 9am (15:00 CET) or any time around then any day except Thursday. Let me know how that sounds, and @Jason Holmberg (Wild Me) it'd be great if you could join and/or we could fill you in after.

Cheers and have a great weekend 🙇

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2022-12-20 19:36:26

I am available to meet his week or early in the new year. Let me know.

Vanessa Suessle (vanessa.suessle@h-da.de)

2023-01-03 10:45:56

Hello,
I am so sorry. I missed the notification of the message.
Suggest any time for this or next week I can make it work.

Vanessa Suessle (vanessa.suessle@h-da.de)

2023-01-27 04:26:56

Hello all,
I would still be interessted to meet and have a chat.
In the upcoming weeks I dont have much fixed meetings and could arrange to meet at almost any time.

Sorry that my first reply took so long. I am kinda new on Slack and missed notifications.
Vanessa :)

Burak Ekim (burak.ekim@unibw.de)

2022-12-08 10:09:40

Hey everyone!
I am Burak, a PhD student from Munich, Germany. I am working on a project in which we (with Michael Schmitt and Ribana Roscher) raise the open question of what makes nature wild for better understanding of wilderness areas and, ultimately, the concept of wild. I mainly use explainable machine learning, uncertainty quantification, and data fusion methods to get the most out of multi-modal earth observation data. 📡🌲 Feel free to ping me and check my website for more information.

🌲 Nico Lang, Josh Seltzer, Jason Holmberg (Wild Me), Katelyn Morrison

👏 Vanessa Suessle, Andrzej Białaś

👀 Andrzej Białaś

👍 Jan Kees

Josh Seltzer (jyseltz@gmail.com)

2022-12-08 12:48:30

Sounds really cool! I'm taking a look at MapInWild but since I don't think it's addressed there (maybe I am mistaken), I'd be really curious to hear your thoughts on the temporal aspect of what it means to be 'wild'. For example, in Latin America there seems to be a common trend of "reforestation reversals" where reforestation efforts are only sustained for a few years, before land is again repurposed for other uses. I imagine it must be very difficult to account for these complexities.

(Not trying to poke any holes - I am new to forest ecology, so just really curious if/how these kinds of complications are considered!)

Frontiers

Reversals of Reforestation Across Latin America Limit Climate Mitigation Potential of Tropical Forests

Carbon sequestration through tropical reforestation and natural regeneration could make an important contribution to climate change mitigation, given that forest cover in many tropical regions increased during the early part of the 21st century. The size of this carbon sink will depend on the degree to which second-growth forests are permanent and protected from re-clearing. Yet few studies have assessed permanence of reforestation, especially not at a large spatial scale. Here, we analyzed a 14-year time series (2001–2014) of remotely sensed land-cover data, covering all tropical Latin America and the Caribbean, to quantify the extent of second-growth forest permanence. Our results show that in many cases, reforestation in Latin America and the Caribbean during the early 21st century reversed by 2014, limiting carbon sequestration. In fact, reversals of reforestation, in which some or all gains in forest cover in the early 2000s were subsequently lost, were ten times more common than sustained increases in forest cover. Had reversals of reforestation been avoided, forests could have sequestered 0.58 Pg C, over four times more carbon than we estimate was sequestered after accounting for impermanence (0.14 Pg), representing a loss of 75% of carbon sequestration potential. Differences in the prevalence of reforestation reversals across countries suggest an important role for socio-economic, political, and ecological context, with political transitions and instability increas...

Original URL: https://www.frontiersin.org/articles/10.3389/ffgc.2020.00085/full

Burak Ekim (burak.ekim@unibw.de)

2022-12-08 14:32:04

Thanks Josh! Yes, in this paper we present our initial efforts in addressing the task of wilderness mapping by forming both sensitivity analysis and semantic segmentation setups on spatial/spectral levels. Although I have not addressed the task from that perspective yet, temporal aspect could introduce a new dimension where discriminative patterns are more noticable and provide valuable hints that are covered otherwise, I would say.

Yes, undisturbance is a vital criteria for wilderness areas in both gradients (spatial and spectral). Probably not what you want to hear but I would approach the reforestation reversal effects from the CV/ML and formalize an optimisation task (I would simply put it on the shoulder of a learner).

A side note: While doing that, I would also not create a strong connection between wilderness and forested areas (as wilderness areas can be found in all kinds of ecological/climate zones).

Ethan Shafron (ethan.shafron@gmail.com)

2022-12-09 14:51:20

Sounds intriguing! As a side note, there is a ton of literature on this from indigenous and political science perspectives, which I think can be very useful to contextualize the concept of wilderness/wildness in your work. Also, as a result of the "Wilderness Act" in the US, there are actually a lot of documents that outline quantitative metrics for managing wilderness areas here, even if the concept itself is more of a narrative goal for managers. In many ways, these metrics are viewed as loss functions - things to minimize in order to maintain "wilderness character". See below - cheers!

https://www.pnas.org/doi/epdf/10.1073/pnas.2022218118

https://wilderness.net/practitioners/toolboxes/wilderness-character/default.php

https://d1wqtxts1xzle7.cloudfront.net/75523678/Indigenous_20Knowledges-libre.pdf?1638420131[…]JrrzlJDn2Fdx8D-NWPA__&Key-Pair-Id=APKAJLOHF5GGSLRBV4ZA

Burak Ekim (burak.ekim@unibw.de)

2022-12-10 13:09:02

Great sources and perspectives Ethan, thanks a bunch for your input! Exactly, wilderness areas are home to many indigenous communities and we should do our best to keep those areas undisturbed. But, as a cultural construct that changes in definition from country to country, it is not really easy to delineate those areas from space. That is why I find the studies on the quantification of wilderness management quite valuable for our investigation.

Graeme Phillipson (graeme.phillipson@bbc.co.uk)

2022-12-08 13:10:15

Could I ask, what is the state of the art for position tracking of animals in urban environments? Especially if you wanted to track a large number? (edited: referring to geographical position data)

Josh Seltzer (jyseltz@gmail.com)

2022-12-08 13:15:21

Do you mean based on ground-level images?

Graeme Phillipson (graeme.phillipson@bbc.co.uk)

2022-12-08 13:52:37

I meant on there positions, like GPS data, but yes at ground level.

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2022-12-08 15:38:37

Do you mean which gps-collars? What type of animal?

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2022-12-08 15:43:10

GPS have out faced VHS as the technology has gotten better and relatively cheaper. GPS collars are more convenient. There other mechanisms to track animals too, one was being developed lately, the ICARUS initiative using a specific satellite but Russia cut the chord lately so we’ll see what happens there. My advice is search “telemetry” and your organism of interest and see what have people have used in the latest literature. I think someone made a tracker with raspberry PI too, or maybe i dreamed it but so far GPS are the go-to protocol especially for mammals at least. Good luck!

👍 Graeme Phillipson, Sara Beery

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2022-12-09 13:19:44

Here is a really cool paper from this year on all telemetry related to animal movement and ecology. It's a good starting point.

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2022-12-09 13:26:48

I just noticed you wanted to track specifically in urban environments, when it comes to tracking devices for terrestrial animals, your main concern will be body weight and budget. Landscape wont have a big impact on your choice. The ones we used were Lotek, and they were pretty good, some stayed on for 1-2 years on urban coyotes. https://www.lotek.com/

Lotek |

LotekLotek

Advancing wildlife science. GPS collars. Our innovative and internationally recognized satellite, radio, acoustic, and archival monitoring solutions allow researchers to ...

Original URL: https://www.lotek.com/

Graeme Phillipson (graeme.phillipson@bbc.co.uk)

2022-12-11 08:24:42

Thanks @Tiziana Gelmi Candusso!

❤️ Tiziana Gelmi Candusso

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2022-12-10 17:02:55

Quick question: Is there any work done on automatic identification of butterflies in images, either identifying species and/or even unique individuals? I was recently looking at some work on butterfly observations and it piqued my curiosity.

Josh Seltzer (jyseltz@gmail.com)

2022-12-10 17:11:25

Hey @Pen-Yuan Hsing! I saw this shared on Twitter last week—I am guessing there is a lot of other work dealing specifically with species/individual identification, but it seems like there is very robust/comprehensive imaging techniques available, in case you're interested: https://www.nature.com/articles/s42003-022-04282-z

Nature

A high-throughput multispectral imaging system for museum specimens

Communications Biology - An accessible approach to multispectral imaging analysis of museum butterfly specimens is presented using consumer technology.

Original URL: https://www.nature.com/articles/s42003-022-04282-z

Katelyn Morrison (kcmorris@andrew.cmu.edu)

2022-12-11 03:24:53

WildMe https://www.wildme.org/ doesn't do this for butterflies yet, but I bet their pipeline could do it given some initial images to set it up. WildMe does species identification and identifies unique individuals for species like whales, turtles, amphibians, and various carnivores 🙂. You might want to look into the Hotspotter Algorithm for individual identification.

wildme.org

Wild Me

Wild Me builds open software and artificial intelligence for the conservation research community. We are machine learning experts and software professionals supporting you in the fight against...

Original URL: https://www.wildme.org/

Yves Bas (yves.bas@gmail.com)

2022-12-11 16:38:33

INaturalist is pretty good for butterflies auto id in a large part of the world: https://www.inaturalist.org/

inaturalist.org

iNaturalist

iNaturalist is a social network for naturalists! Record your observations of plants and animals, share them with friends and researchers, and learn about the natural world.

Original URL: https://www.inaturalist.org/

🦋 Alex Borowicz, Katelyn Morrison

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2023-01-17 12:33:12

Sorry for my late response here, thanks for the suggestions I'll take a look! I have a butterfly colleague to whom I will also share this.

Katelyn Morrison (kcmorris@andrew.cmu.edu)

2023-01-17 12:34:08

Hey! I am talking to a group at Mila who is working on this - let’s touch base soon? They are also in this slack actually!

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2023-01-17 12:35:53

Hi @Katelyn Morrison cool! The colleague I know is Denise Dalbosco Dell'Aglio (part of a group at the Univ of Bristol where I happen to also be, but part of a unrelated project) who has a mountain of butterfly footage from Panama. Do you know them?

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2023-01-17 12:36:01

I'd love to touch base about this!

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2023-01-17 12:36:38

My background is in ecology/conservation but not butterflies specifically, but would be interested if we can make something interesting out of this.

Katelyn Morrison (kcmorris@andrew.cmu.edu)

2023-01-17 12:37:18

Ohhh awesome! I don’t know them, but yes I’d love to touch base. I am starting up a study to understand how lepidopterists collaborate with XAI to determine the ID of the butterfly. My background is computer science and human-AI collaboration. :)

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2023-01-17 12:37:51

That's so cool! Lemme reach out to Denise to see if I can get a few sample clips to share.

🙌 Katelyn Morrison

Andrzej Białaś (andrzej@appsilon.com)

2022-12-12 07:50:18

👋 Hello everyone! I’m Andrzej Białaś and I am the Data4Good Lead at Appsilon. I joined this Slack community a while ago and mostly lurked around. The intro is well overdue!

Happy to connect with others working on using technology for the benefit of planet & people (Add me on LinkedIn or DM here).

Some more info about me and our work:
Appsilon, delivers high-quality data analysis and technical support to help our clients make the most of their data. We specialize in building advanced enterprise R Shiny applications, and we have a passion for using our skills to make a positive impact on the world.

In my work at Appsilon, I oversee our Data4Good program, which realized that mission. One of our keystone projects coming from the Data4Good initiative is Mbaza AI - an open-source algorithm that allows rapid biodiversity monitoring at scale.

PS: I am excited to learn from and collaborate with all of you in this community!
PPS: I am much more responsive on LinkedIn as I have only a handful of Slack channels on my mobile device).

👋 Viktor Domazetoski, Jose Ruiz-Munoz, Carly Batist, Katelyn Morrison, Dan Morris, Lucia Gordon, Sara Beery, Jon Van Oast, Amber De Neve, Eddie Zhang, Timm Haucke, Lindsey Dukles, Mikey Tabak, Antón Álvarez, Yseult Hb

Austin Greene (austin.greene@whoi.edu)

2022-12-13 09:19:00

Hey all! My name is Austin Greene and I am a postdoctoral investigator at Woods Hole Oceanographic Institution. I mostly work on coral reefs and other threatened marine habitats as a disease ecologist by training. Over the past few years I started building low-cost camera systems (CoralCam and KiloCam) and that brought me into the world of conservation technology. I'm hoping to develop a new iteration on these that has embedded ML and am keen to work on any other projects people have that make conservation tech more accessible to those with the least resources. Nice to meet you all!

👍 Aleksis Pirinen, Fagner Cunha, Timm Haucke, Alexander Robillard, Sam Kelly, Katelyn Morrison, Eddie Zhang, Justin Kay, Jose Ruiz-Munoz, Declan, Sara Beery, Lindsey Dukles, Armi Tiihonen, David Will, Valentin Ștefan, Talia Speaker, Anthony Bao, Andrew Schulz, Yseult Hb, Henrik Cox (Sentinel)

👏 Silvia Zuffi, Lindsey Dukles, Lucia Gordon, Ted Schmitt, Toryn Schafer

👋 Sara Beery, Atul Ingle, Viktor Domazetoski, Jon Van Oast, Carly Batist, Ed Miller

👍:skin_tone_3: Pen-Yuan Hsing

Josh Seltzer (jyseltz@gmail.com)

2022-12-13 09:20:31

This sounds really cool! Welcome @Austin Greene 👋

Alexander Robillard (RobillardA@SI.EDU)

2022-12-13 09:31:59

Sounds awesome!

Austin Greene (austin.greene@whoi.edu)

2022-12-13 14:36:27

Thanks so much for the warm welcome!

Valentin Ștefan (valentin.stefan.vst@gmail.com)

2022-12-13 15:31:50

Hi Austin, I am interested in such custom cameras for pollinator monitoring. We wanted to order some Raspberry Pi gear and follow the guidelines of @Maximilian Sittinger, but I could't find any available on the market at the moment. I am therefore interested in alternatives / other microcomputers. Could you share some info with me about your models? How easy are the components to order, costs, put together and add custom code on them?

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-12-13 15:53:47

You should talk to the Sentinel team at ConservationX! They’re working on ML-enabled camera traps/edge AI with Edge Impulse. Edge Impulse has also been working with the Audiomoth folks to develop edge computing for acoustic recorders. You could also check out other resources related to this through the Conservation Tech Directory (KiloCam is already in there 🙂)

👍 Valentin Ștefan

Ed Miller (ed@hypraptive.com)

2022-12-14 16:49:06

Tagging the Sentinel team: @Henrik Cox (Sentinel), @Sam Kelly

👍 Carly Batist

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-12-14 19:25:38

totally forgot to do that 🤦‍♀️. thanks Ed!

👍 Ed Miller

Henrik Cox (Sentinel) (henrik@conservationxlabs.org)

2022-12-14 21:01:56

Thanks @Ed Miller and @Carly Batist! This looks fantastic @Austin Greene, it would be great to see how we can cross wires. I'll DM

👍 Ed Miller

sam maxwell (maxwellsamjm@gmail.com)

2022-12-13 16:01:54

Hi all,
I am Sam and work on a partnership research project with SUEZ and the LaBRI (Computer Science Research Lab of Bordeaux) in France on the case of Sound Event Detection and Underwater Acoustic. This project aims to develop an automatic system to process and analyze acoustic data from underwater environment monitoring, like marine and freshwater ecosystems, but also in the case of sanitary systems environment. Having first studied Biology and ecology, I am interested in conservation topics. Happy to join this group :)

👋 Suzanne Stathatos, Armi Tiihonen, Mikey Tabak, Jose Ruiz-Munoz

👍 Talia Speaker, Déva Sou

Talia Speaker (talia.speaker@wildlabs.net)

2022-12-13 16:06:40

Hi everyone! Carly shared a while ago, but wanted to flag again that our WILDLABS State of Conservation Technology Survey 2022 is open until Dec 30th. The aim of this annual research is to track and inform the evolution of conservation tech, and we'd love to have more AI for Conservation community perspectives represented. Last year, our global assessment showed that ML/computer vision tools were viewed as having the highest untapped potential, so it will be interesting to see if and how that shifts this year, and how needs are changing. Thanks so much in advance for your time and please share widely! More background info here.

colostate.az1.qualtrics.com

WILDLABS Conservation Tech Survey 2022

Contribute your thoughts to an evaluation of the State of Conservation Technology by researchers at WILDLABS & Colorado State University

Original URL: https://colostate.az1.qualtrics.com/jfe/form/SV_0HTOP765wvVTkii

❤️ Carly Batist, Sara Beery, Pen-Yuan Hsing

Sasha Luccioni (sasha.luccioni@huggingface.co)

2022-12-13 17:41:36

Hi all! My name is Sasha and I'm a researcher at Hugging Face 🤗, an AI startup that aims to democratize AI and make datasets and models more accessible to more communities.
I recently worked with @Dan Morris to add the LILA camera trap data to the HuggingFace Hub (the central place where people can share resources), and I'm looking forward to adding more datasets and more modalities 🖼️ 🔊 📃 🐘 :giraffe_face: 🐋 🌴
Reach out if this sparks any ideas on your end!

👋 gvanhorn, Ben Weinstein, Elizabeth Bondi, Caleb Robinson, Dan Morris, Declan, Jason Parham, Katelyn Morrison, Nico Lang, Cameron Trotter, Timm Haucke, Viktor Domazetoski, Josh Veitch-Michaelis, Mikey Tabak, Carly Batist, Eddie Zhang, Ștefan Istrate, Jose Ruiz-Munoz, Sara Beery, Lindsey Dukles

😎 Jon Van Oast, Jason Parham, Timm Haucke, Sara Beery, Lindsey Dukles

🤗 Suzanne Stathatos, Jason Parham, Timm Haucke

❤️ Justin Kay, Caleb Robinson, Jason Parham, Timm Haucke

Ben Weinstein (benweinstein2010@gmail.com)

2022-12-13 17:43:21

We have tree data and a model that would be nice to contribute. https://deepforest.readthedocs.io/, https://github.com/weecology/NeonTreeEvaluation. Plus tree species data that will be around soon. Some overlap with datasets that torchgeo @Caleb Robinson have done a great job compiling.

weecology/NeonTreeEvaluation

Benchmark dataset for tree detection for airborne RGB, Hyperspectral and LIDAR imagery

Stars

Language

Python

Sasha Luccioni (sasha.luccioni@huggingface.co)

2022-12-13 17:44:09

Sure, let me know if I can help you upload it! The documentation is pretty good, FWIW : https://huggingface.co/docs/datasets/index

huggingface.co

Datasets

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Original URL: https://huggingface.co/docs/datasets/index

👍 Ben Weinstein

Caleb Robinson (calebrob6@gmail.com)

2023-02-01 01:28:22

(reviving a 2 month old thread out of absolutely nowhere) @Ben Weinstein FYI we've started putting some torchgeo stuff on HuggingFace (https://huggingface.co/torchgeo). In particular we're rehosting some datasets here, pros:

WayYyy faster download speeds than rehosting stuff on Zenodo!

We've hit a couple of cases now where one our dataset stops working because the SSL certificate of wherever it was hosted had expired. Most recently with the EuroSAT dataset https://github.com/phelber/EuroSAT/issues/10. We're betting that HuggingFace knows way more about sys admin-ing than random academics 🙂

cons:

no idea, seems good

🤗 Sasha Luccioni

👍 Ben Weinstein

Sasha Luccioni (sasha.luccioni@huggingface.co)

2023-02-01 09:38:48

haha let me know if you have any questions, y'all :facewithcowboy_hat:

❤️ Caleb Robinson

Katie Wetstone (she, her) (katie@drivendata.org)

2022-12-14 16:46:04

Hi y'all! 👋 I'm a data scientist at a small consulting company focused on social good, and wanted to share a programming competition we just posted! Competitions are generally a great way to break into the programming space and get some python creds 🐍

Today we launched an exciting computer vision challenge in partnership with NASA :female_astronaut: to identify harmful algal blooms in important water sources. Even better, it's punnily named the Tick Tick Bloom: Harmful Algal Bloom Detection Challenge!

🌊 🌿 The goal of the competition is to help public health officials detect and address algal blooms in water sources like lakes and reservoirs, which can pose serious dangers to both human and ecosystem health
📡 You can create new state-of-the-art methods using satellite imagery to protect communities across the US
💸 There's a $30,000 prize pool for the top performing teams!

You can learn more and sign up to participate by visiting the challenge here. Feel free to share with anyone else you think might be interested, and reach out if you have any questions! Happy coding 🙂

DrivenData

Tick Tick Bloom: Harmful Algal Bloom Detection Challenge

Harmful algal blooms occur all around the world, and can harm people, their pets, and marine life. Use satellite imagery to detect dangerous concentrations of cyanobacteria, and help protect public health!

Original URL: https://www.drivendata.org/competitions/143/tick-tick-bloom/page/649/

🙌 Josh Seltzer, Lucia Gordon, Sara Beery, Casey Youngflesh, Peter Bull, Aleksis Pirinen, Jose Ruiz-Munoz, Tiziana Gelmi Candusso, Timm Haucke, Stephanie O'Donnell, Carl Boettiger, Yseult Hb, Nicolas Arrieta Larraza

📡 Carl Boettiger

Kateryna Nekhomiazh (kateryna.nekhomiazh@mail.utoronto.ca)

2022-12-15 15:56:01

Hi all! My name is Kate, I’m first year MSc in computer science at the University of Toronto, doing research in reinforcement learning.

I recently discovered the potential for AI to help with environmental conservation efforts. I am now intensively learning about this topic and am eager to apply my skills in machine learning and AI to make a positive impact on the world. I am particularly interested in the impact of unconscious consumption and war on the Earth’s ecosystem.

If you have any projects that you need help with, or to which you plan to attract students, I would love to participate!

👋 Sara Beery, Josh Seltzer, Suzanne Stathatos, Andrew Schulz, Lucia Gordon, Dan Morris, Aleksis Pirinen, Timm Haucke, Lindsey Dukles

😎 Jon Van Oast

Josh Seltzer (jyseltz@gmail.com)

2022-12-15 16:01:07

Hey @Kateryna Nekhomiazh welcome!! I'm a UofT alum, so nice to see you here 🙇 I'm in the early stages of working on something you might be interested in, I'll reach out once it's a bit more concrete! Since you're interested in RL (and things at least tangential to policy), you might be interested in AI Economist (they also have an active slack community and lots of events) as well

salesforce/ai-economist

Foundation is a flexible, modular, and composable framework to model socio-economic behaviors and dynamics with both agents and governments. This framework can be used in conjunction with reinforcement learning to learn optimal economic policies, as done by the AI Economist (<https://www.einstein.ai/the-ai-economist>).

Website

<https://www.einstein.ai/the-ai-economist>

Stars

915

Kateryna Nekhomiazh (kateryna.nekhomiazh@mail.utoronto.ca)

2022-12-15 16:03:45

thank you so much, I will look forward to it!

Peggy Bevan (ucbtbev@ucl.ac.uk)

2022-12-18 05:56:06

Hi all 🙂 I’m wondering if anyone has some recommendations for ML textbooks useful for postgrad students (Master’s level)? We are creating a reading list for students learning ML for Ecology but they will be from a range of backgrounds, some not computer scientists. Thanks!

👍 Omiros Pantazis, Kakani Katija

Ethan Shafron (ethan.shafron@gmail.com)

2022-12-18 10:24:00

This is a great book for general ML concepts and algorithms, and it's all free - https://www.statlearning.com/

🙏 Peggy Bevan

Matt Weldy (matthewjweldy@gmail.com)

2022-12-18 10:32:24

I enjoyed working through this one. Although, it is 'deep learning' focused so there isnt any text space on other learning approaches. It has code available in the three largest training frameworks. https://d2l.ai/

🙌 Peggy Bevan

Graeme Phillipson (graeme.phillipson@bbc.co.uk)

2022-12-18 11:17:21

I often use the google ML crash course for that kind of thing https://developers.google.com/machine-learning/crash-course

Google Developers

Machine Learning | Google Developers

Original URL: https://developers.google.com/machine-learning/crash-course

👍 Peggy Bevan

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2022-12-18 16:47:17

I personally found the presentation in Introduction to Statistical Learning quite dry. It does cover R, which ecologists like (and if you need them to learn R, then that could be important), and it does cover the theory from first principles. Despite the "introduction", it is still quite mathsy and I think you need to read quite a lot of it before you can actually use the information. I would say it's an approachable theory book to accompany more practical books. There is also a MOOC available from the authors.

Book-wise, I like Machine Learning with Applications in Scikit-Learn and Tensorflow by Aurelien Geron. It uses Tensorflow/Keras and it covers "classical" ML too. There's also the Fast AI book if you want something PyTorch focused (and I think is free?) Both are O'Reilly. I've heard good things about d2l.

Be warned that there is a lot of cruft in O'Reilly's offerings these days and I think their editorial standards aren't as high as they used to be. Manning has some nice books - I'm reading Human-In-The-Loop Machine Learning at the moment.

In terms of courses, Fast AI is enduringly popular. My only complaint is that they rely a lot on their library which I think adds unnecessary fragmentation when things like Pytorch-Lightning exist. But, the course is presented at a good level for beginners, the API is easy, and there are a load of little tips and tricks for Python/Jupyter.

For Deep Learning theory, I really think Andrej Karpathy's CS231 lectures are some of the best out there even though they're quite old (specifically the season where he taught). Also hey, it's 2022 and ResNets are still popular. He's also started a new series on YouTube which gives a fantastic overview of building a DL library from scratch. His talent is convincing you that actually the code that runs these mega networks is conceptually quite simple. The videos also have accompanying exercises in Colab, which is great. Like Fast AI he throws in a lot of neat Python stuff which you might not be familiar with (and it's interesting to see how he approaches problems and tooling). Of course there's Andrew Ng's course on ML which is the OG 🙂

🙏 Peggy Bevan, Dhruv Sheth

Silvia Zuffi (silvia@mi.imati.cnr.it)

2022-12-19 04:35:58

I would recommend Bishop’s book https://www.microsoft.com/en-us/research/people/cmbishop/prml-book/

Microsoft Research

Christopher Bishop at Microsoft Research

Christopher Bishop is a Microsoft Technical Fellow and Director of Microsoft Research AI4Science. He is also Honorary Professor of Computer Sc…

Original URL: https://www.microsoft.com/en-us/research/people/cmbishop/prml-book/

👍 Peggy Bevan, Dhruv Sheth

➕ Atul Ingle

Toryn Schafer (tschafer@tamu.edu)

2022-12-19 10:40:13

Not specifically textbooks or ML, but the Ecological Forecasting Initiative have put together a great list of educational resources and I'm sure there is intersection with ML: https://ecoforecast.org/resources/educational-resources/

👍 Peggy Bevan, Carly Batist, Dhruv Sheth

Peggy Bevan (ucbtbev@ucl.ac.uk)

2022-12-19 11:10:07

Thanks so much everyone! This is incredibly helpful 🤗🤗

Sankaran (shun-ka-run) (sankaranv@cs.umass.edu)

2022-12-21 07:56:40

I would suggest using lecture notes and slides from a course (there’s so many, Stanford CS 229 seems to be popular though) rather than a textbook to help build a foundation in ML, and then use another resource specifically for deep learning. I learned from Kevin Murphy’s book (Machine Learning: A Probabilistic Perspective) which gave me a great foundation but is very huge and covers a lot of methods in depth - not sure is the best use of time for someone trying to get ahead to applied work.

I really liked this resource for deep learning. I wasn’t a fan of the textbook from Goodfellow, Bengio and Courville that is pretty much the standard in courses, there really isn’t another I know of that is widely adopted though. IMO the online resources and notebooks blow them out of the water, but I’ve never seen one where the sections on ML that precede the deep learning sections are very good.
https://uvadlc-notebooks.readthedocs.io/en/latest/
https://d2l.ai

Cody Kupferschmidt (kupfersc@uoguelph.ca)

2023-01-04 10:00:10

A Course in Machine Learning by Hal Daumé is free to download and quite good. http://ciml.info/

Lukas Picek (lukaspicek@gmail.com)

2022-12-19 14:08:08

Great News! The FGVC10 workshop was accepted to CVPR 2023 and will take place in Vancouver. Interested in fine-grained learning and its applications in biodiversity and conservation?
Consider submitting a paper to FGVC10 or participate in one of our competitions.

More information on the website: https://sites.google.com/view/fgvc10
#CVPR #CVPR2023 #FGVC

😍 Sara Beery, Fagner Cunha, Marek Hruz, Oisin Mac Aodha, Lukas Picek, Andrew Schulz, Elijah Cole (Deactivated), Nico Lang, Jason Holmberg (Wild Me), Yuanqi Du, Robin Zbinden, Georgia Atkinson, Timm Haucke, Justin Kay

🎉 Lukas Picek, Jon Van Oast, Omiros Pantazis, Aleksis Pirinen, Josh Seltzer, Jason Holmberg (Wild Me), Dan Morris, Diego Marcos, Cameron Trotter, Riccardo de Lutio, Yuanqi Du, Timm Haucke

:thumbsup_all: Frederic Fol Leymarie, Jose Ruiz-Munoz, Timm Haucke

👍 Kostas Papafitsoros

Nico Lang (nila@di.ku.dk)

2023-03-06 17:11:54

FGVC10's submission website is now open for your 4-page extended abstracts on fine-grained recognition and related topics:
https://sites.google.com/view/fgvc10/submission
Deadline: March 20

🙏 Oisin Mac Aodha, gvanhorn, Andy Viet Huynh, Lukas Picek, Sara Beery

🎉 Jon Van Oast, Ronny Hänsch, gvanhorn, Andy Viet Huynh, Fagner Cunha, Lukas Picek, Riccardo de Lutio, Sara Beery, Dan Morris

Josh Seltzer (jyseltz@gmail.com)

2022-12-20 12:37:57

I just came across the GPAI Biodiversity & Artificial Intelligence report, published last month - I see it has a lot of citations and contributions from some community members here, and for myself at least it's a very useful resource which I'll be using to guide future work!

😎 Jon Van Oast, Sara Beery, Yseult Hb, Hannah Murray, Wenxin Yang, Carly Batist, Emilio Luz-Ricca, Marconi Campos, Kateryna Nekhomiazh

👍 Otto Brookes, Armi Tiihonen, Monty Ammar

Devis Tuia (devis.tuia@epfl.ch)

2022-12-21 03:59:24

Seems that there will be quite some “Earth” workshops at CVPR this year! To add to the good news of FGCV10 above (and I let the cv4animals folks announce their own later 😉 ), I am also happy to announce the seventh edition of EarthVision 🌍📡!
More info here: https://www.grss-ieee.org/events/earthvision-2023/
(we are still populating the website, but at least the deadlines are on…)

GRSS-IEEE

EARTHVISION 2023 - GRSS-IEEE

EARTHVISION 2023 June 18th/19th, Vancouver, Canada in conjuction with the Computer Vision and Pattern Recognition (CVPR) 2023 Conference Aims and Scope Important Dates People Challenge Sponsors Submission Program CVPR 2022 Aims and Scope ”Earth Observation (EO) and remote sensing are ever-growing fields of investigation where computer vision, machine learning, and signal/image processing meet. The general […]

Est. reading time

4 minutes

Original URL: https://www.grss-ieee.org/events/earthvision-2023/

🎉 Lukas Picek, Oisin Mac Aodha, Nico Lang, Omiros Pantazis, Subhransu Maji, Timm Haucke, Aleksis Pirinen, Robin Zbinden, Olivier Dietrich, Adam Noach, Sara Beery, Iván Higuera-Mendieta, Emilio Luz-Ricca, Aakash Gupta, Kasirat

🍁 Burak Ekim, Frederic Fol Leymarie

🌎 Oisin Mac Aodha, Nico Lang, Valentin Gabeff, Omiros Pantazis, Aleksis Pirinen, Robin Zbinden, Majid Mirmehdi

😍 Sara Beery

Silvia Zuffi (silvia@mi.imati.cnr.it)

2022-12-21 12:02:47

Yes, CV4animals will be at CVPR again!!! Stay tuned for a proper announcement with the new website! In the meantime I can anticipate the list of speakers, as I am super happy about it: Kostas Daniilidis, Tanya Berger-Wolf, Katija Kakani, Arsha Nagrani, Albert Ali Salah!

👍 Subhransu Maji, Oisin Mac Aodha, Elijah Cole (Deactivated), Sara Beery, Felipe Parodi, Katelyn Morrison, Jason Parham, Omiros Pantazis, Aleksis Pirinen, Antón Álvarez, Jose Ruiz-Munoz, Timm Haucke, Cameron Trotter, Devis Tuia, Robin Zbinden, Majid Mirmehdi, Otto Brookes, Yseult Hb, Justin Kay, Paul Janson, Levi Cai, Genevieve Moat

🐕 Sara Beery, Robin Zbinden, Emilio Luz-Ricca

🎉 Sofía Miñano

Devis Tuia (devis.tuia@epfl.ch)

2023-01-10 02:39:29

details details 🙂, we want details!

Silvia Zuffi (silvia@mi.imati.cnr.it)

2023-01-11 05:09:54

Yes, soon!

Silvia Zuffi (silvia@mi.imati.cnr.it)

2023-01-18 07:21:35

Here it is! https://www.cv4animals.com/ Please note that we accept work-in-progress papers, as we want our workshop to be an opportunity for people to interact and have feedback on their work. Hope to see many submissions from this community!

cv4animals.com

https://connectedconservation.foundation/news/satellites-for-biodiversity-award/

About

Original URL: https://www.cv4animals.com/

🙌 Stephanie O'Donnell, Vincent Christlein, Cameron Trotter, Oisin Mac Aodha, Dan Morris, Emilio Luz-Ricca, Valentin Gabeff, Sara Beery, Mahir Patel, Sofía Miñano

:squirrel: Andrew Schulz

Carly Batist (cbatist@gradcenter.cuny.edu)

2022-12-27 14:00:34

CCF

Satellites for Biodiversity Award

Connected Conservation Foundation and The Airbus Foundation are inviting individuals and not-for-profit organisations to apply to a new grant programme.&#160; Winners will receive access to the most advanced optical satellite imagery at 30cm spatial resolution from Airbus Pl&#233;iades Neo and 50cm from Pl&#233;iades, alongside $5,000 USD of financial funding.

Original URL: https://connectedconservation.foundation/news/satellites-for-biodiversity-award/

👍 Wenxin Yang, Robin Zbinden, Jose Ruiz-Munoz, Sara Beery, Declan

👍:skin_tone_5: Ando Shah

Andrew Schulz (akschulz@gatech.edu)

2022-12-28 14:08:14

Did not know if people are interested in this group, but this is the syllabus from the Conservation Tech course at Georgia Tech that is for undergraduate engineers, computer scientists, and biologists. I'm working over the winter break to get more of the content online and when it is online it will all be linked on this syllabus! https://zenodo.org/record/7470257#.Y6yTW3aZND9

Zenodo

Conservation Technology Project Based Course Undergraduate Syllabus

There is a need for the development of new techniques to aide in the conservation of wildlife to prevent extinction and the decrease of biodiversity. Examples of such techniques include the use of drones to mitigate rhino poaching or low-cost equipment that can deter elephants or predators from entering rural communities. The goal of this VIP is to expose Georgia Tech students to the field of conservation, and to bring different disciplines together to aide various species. All projects are proposed and supported by real experts in the field, who can provide the background knowledge and advice as required. After this course is completed, a short curriculum will ensure that students learn how to perform background research, conduct ethical experiments with animals, and be able to document their results. In addition to the short curriculum that the class offers, students will grow by communicating with international leaders in wildlife and species conservation, Furthermore, continued participation in this VIP will allow students to take projects from an idea to a tested and implemented solution.

Original URL: https://zenodo.org/record/7470257#.Y6yTW3aZND9

👍 Jose Ruiz-Munoz, Dan Morris, Ted Schmitt, Tiziana Gelmi Candusso, Wenxin Yang, Alessandra Sellini, Cameron Trotter, Sara Beery, Dhruv Sheth, Sankaran (shun-ka-run), George Colaco, Kalindi Fonda, Timm Haucke

❤️ Carly Batist, Antón Álvarez, Yseult Hb, Dhruv Sheth, Rebecca Wilks

Lindsey Dukles (lindseydukles@gmail.com)

2022-12-30 11:22:08

Hi All!👋

I'm Lindsey and am new here! @Michael Bunsen introduced me to this group and it has been inspiring to see what everyone is working on!

I'm an NYC-based Software Engineer with a background in conservation biology. I recently completed a coding bootcamp and am looking to land my first software engineering role with a company whose mission I believe in. I have a background in chemical lab management, outdoor education, and the service industry. Outside of work you can find me birding 🐦, botanizing 🌼, climbing 🧗‍♂️, or reading 📕.

Excited this group exists and to connect with you all. Feel free to reach out to me here or linkedin!

👋 Josh Seltzer, Sara Beery, Eddie Zhang, Carly Batist, Yves Bas, Kateryna Morhun, Dan Morris, Antonio Ferraz, Jose Ruiz-Munoz, Jorrit van Gils

Alexander Kobald (avkobald@gmail.com)

2023-01-02 10:20:12

Hi everyone! I'm Alex, I co-direct the Design Across Scales Lab, an architecture research lab at Cornell University. @Sara Beery pointed me to this group when we met to share notes on some of our work on detecting and modeling urban forests. Excited to see what everyone is working and happy to be here!

😎 Jason Holmberg (Wild Me), Sara Beery, Jorrit van Gils

Ben Weinstein (benweinstein2010@gmail.com)

2023-01-02 18:29:48

cool demo! Are we looking at species predictions or ground truth? If predictions, from LiDAR only? Point density? Any publications/white paper to go with these? We work on similar themes using HSI. https://www.biorxiv.org/content/10.1101/2022.12.07.519493v1. Working on semi-supervision now.

bioRxiv

Capturing long-tailed individual tree diversity using an airborne multi-temporal hierarchical model

Measuring forest biodiversity using terrestrial surveys is expensive and can only capture common species abundance in large heterogeneous landscapes. In contrast, combining airborne imagery with computer vision can generate individual tree data at the scales of hundreds of thousands of trees. To train computer vision models, ground-based species labels are combined with airborne reflectance data. Due to the difficulty of finding rare species in a large landscape, the majority of classification models only include the most abundant species, leading to biased predictions at broad scales. Extending classification models to include rare species requires targeted data collection and algorithmic improvements to overcome large data imbalances between dominant and rare taxa. In addition, large landscapes often require multiple acquisition events, leading to significant within-species variation in reflectance spectra. Using a multi-temporal hierarchical model, we demonstrate the ability to include species predicted at less than 1% frequency in landscape without losing performance on the dominant species. The final model has over 75% accuracy for 14 species with improved rare species classification compared to a baseline deep learning model. After filtering out dead trees, we generate landscape species maps of individual crowns for over 670,000 individual trees at the Ordway Swisher Biological Station within the National Ecological Observatory Network. We estimate the relative abundance of the species within the landscape and provide three measures of uncertainty to generate a range of counts for each species. These maps provide the first estimates of canopy tree diversity within NEON sites to include rare species and provide a blueprint for capturing tree diversity using airborne computer vision at broad scales. ### Competing Interest Statement The authors have declared no competing interest.

Original URL: https://www.biorxiv.org/content/10.1101/2022.12.07.519493v1

Joe Ferdinando (jgf94@cornell.edu)

2023-01-03 12:28:02

These are "ground truth" from the NYC 2015 tree census

👍 Ben Weinstein

Devis Tuia (devis.tuia@epfl.ch)

2023-01-03 07:46:44

Little reminder, deadline January 9th: 💥 MEGA PROJECT and JOB ALERT: Hello everyone! We (U. Schultz, @Devis Tuia, @Blair Costelloe @Tilo Burghardt, M. Wikelski, B. Risse and many more) are happy to share that we will start early next year the project WILDDRONE (wilddrone.eu)! It is a Marie Curie Network, meaning that we will build a network of 13 PhD students :femalestudent:🧑‍🎓 across Europe around themes of Drones for conservation in Africa :zebraface:, with PhD topics from computer vision, to robotics and ecology! This also means that we need your help 🆘, dear members, to recruit.. can you help us out by sharing this link among your peers/students/friends? https://wilddrone.eu/recruitment/

WildDrone - Drones for Nature Conservation

Recruitment - WildDrone

Original URL: https://wilddrone.eu/recruitment/

💥 Robin Zbinden, Josh Seltzer, Sara Beery, Kateryna Morhun, Yseult Hb, Eddie Zhang, Emilio Luz-Ricca, Jorrit van Gils

😎 Jon Van Oast

📝 Remi Gosselin

benjamin de charmoy (benjamin.decharmoy@gmail.com)

2023-01-03 13:19:08

This looks super cool. It wasn’t immediately obvious if it’s open to folks from non European countries. I’m from South Africa and would love to apply. I’ll reread the details to check. Just reading the well structured themes and projects is interesting/exciting. Thanks for sharing

Blair Costelloe (blaircostelloe@gmail.com)

2023-01-03 13:22:08

It is open to all nationalities!

Blair Costelloe (blaircostelloe@gmail.com)

2023-01-03 13:22:54

The only limit is the MSCN mobility rule, dependent on the country of the hosting institution – ‘No residence or main activity (work, studies etc.) in the country of the recruiting beneficiary for more than 12 months in the 36 months before their recruitment date’.

benjamin de charmoy (benjamin.decharmoy@gmail.com)

2023-01-03 13:24:15

Awesome 😎 ah! I see that now. Thank you

Blair Costelloe (blaircostelloe@gmail.com)

2023-01-03 13:25:27

No problem! Let us know if you have any other questions as you write the application

✊ benjamin de charmoy

Devis Tuia (devis.tuia@epfl.ch)

2023-01-03 14:39:35

definitely, there is no nationality restriction, apart from not having lived recently in the country that will host you!

🥸 benjamin de charmoy

🥳 benjamin de charmoy

Remi Gosselin (remipgosselin@gmail.com)

2023-01-03 14:47:38

Finishing my application later today! :femalestudent::femaletechnologist:

🙏 Devis Tuia

🙌 Blair Costelloe, benjamin de charmoy

Sagar Nagaraj Simha (sagarnagarajsimha@gmail.com)

2023-01-03 23:55:24

Thank you for this wonderful consortium. I am excited to apply. I had a few general queries while uploading the application materials on the portal

Project description - Does this refer to the 2 page motivation letter?

List of publications - I already have a list described in my CV. Should I also list in a separate document and upload?

Publication - I assumed I should upload a chapter from my thesis here, as mentioned in the requirements. Can I also just upload my entire thesis since the file size is within 10 Mb?

References - I list out the details of my referees in a document ? Since they would be contacted in the second stage.

Devis Tuia (devis.tuia@epfl.ch)

2023-01-04 02:44:13

Hello,

yes, it is the motivation for you to join the consortium and for the (up to) three projects you apply for among the 13

I guess it does not harm to upload a 2nd one

Perfect. Otherwise try to compress it a bit or send a link to the full version if it is too big for the SDU server

Yes, a one pager with the name and addresses of up to three references

Sagar Nagaraj Simha (sagarnagarajsimha@gmail.com)

2023-01-04 02:53:27

Noted. Thank you for all the clairifications Professor.

👍 Devis Tuia

Emilio Luz-Ricca (eluzricca@email.wm.edu)

2023-01-08 22:15:02

I am having some trouble with the SDU portal for submitting my application… I have tried submitting multiple times today, but have not received a confirmation receipt yet. The portal says that this means that the application has not been registered properly. Is this an issue that other applicants are experiencing? Is there any other way to submit my application by tomorrow’s deadline if the portal continues having issues? Thank you!

Emilio Luz-Ricca (eluzricca@email.wm.edu)

2023-01-08 23:16:57

Looks like the application went through in the end. Apologies if multiple submissions appear!

Devis Tuia (devis.tuia@epfl.ch)

2023-01-09 03:57:31

no problem!

Devis Tuia (devis.tuia@epfl.ch)

2023-01-09 03:57:43

I would have contacted the SDU admin if the problem persisted

benjamin de charmoy (benjamin.decharmoy@gmail.com)

2023-02-02 02:31:39

@Emilio Luz-Ricca have you heard back regarding your application? I submitted as well and haven’t received anything further. Just curious. Thanks 😊

Devis Tuia (devis.tuia@epfl.ch)

2023-02-02 02:51:02

hello Benjamin. We are going through all the application as we speak. It takes time, and especially since we are coordinating over 13 PIs. You should hear back from us in a week or two. Thanks for your patience.

benjamin de charmoy (benjamin.decharmoy@gmail.com)

2023-02-02 04:12:30

Awesome, I fully understand. And thank you 😊 I was just curious not really expecting anything so soon. Thanks for the swift response.

Sagar Nagaraj Simha (sagarnagarajsimha@gmail.com)

2023-02-07 15:43:59

https://wilddrone.eu/kick-off-meeting/

WildDrone - Drones for Nature Conservation

Kick-off Meeting - WildDrone

Kick-off Meeting All WildDrone DC-hosting partners got together for the project’s kick-off meeting in Denmark on 16-17th of January 2023. We had a successful meeting where we had the chance to discuss the organization of work packages, learn about gender equality in the hiring process, and start the reviewing process of the 189 applications that

Original URL: https://wilddrone.eu/kick-off-meeting/

Sagar Nagaraj Simha (sagarnagarajsimha@gmail.com)

2023-02-07 15:44:18

@benjamin de charmoy I have been curiously waiting too. This link may help follow the happenings at Wilddrone

Sagar Nagaraj Simha (sagarnagarajsimha@gmail.com)

2023-02-22 09:11:30

@benjamin de charmoy @Emilio Luz-Ricca Did you receive any responses so far, apart from the assignment of review committee maybe ?

benjamin de charmoy (benjamin.decharmoy@gmail.com)

2023-02-22 14:44:53

I did. I was unsuccessful 🦑 Got a nice kind message 😊 keen to see the results some day as it looks like a strong initiative 🚀

Sagar Nagaraj Simha (sagarnagarajsimha@gmail.com)

2023-02-22 15:22:13

Oh! That’s unfortunate. Thanks for the reply. I am yet to hear the decision. I wish you good luck with other opportunities!

Thijs (thijs@q42.nl)

2023-01-04 04:00:43

I'm pretty happy to say that a paper I co-authored is online now! http://doi.org/10.1111/2041-210X.14036

It's about a pilot we did in Gabon with realtime AI camera's to validate a system to mitigate Human-Elephant conflicts.

It was the first time I co-authored a scientific paper, was really fun to do and learn about the process.

🎉 Cameron Trotter, Jose Ruiz-Munoz, Declan, Lucia Gordon, Dan Morris, Sara Beery, Carly Batist, Rita Pucci, Emilio Luz-Ricca, Mark Fisher, Timm Haucke, Pen-Yuan Hsing, Alexander Robillard, Rebecca Wilks

Thijs (thijs@q42.nl)

2023-01-04 04:32:58

Se this post for more info and a cool video about our project: https://www.linkedin.com/posts/tsuijten_ai-enabled-camera-traps-for-protecting-elepha[…]238213607424-B0e6?utm_source=share&utm_medium=member_desktop

linkedin.com

Thijs Suijten on LinkedIn: AI-enabled camera traps for protecting elephants (2)

I&#39;m excited to share that a scientific paper I co-authored is published! It&#39;s about using realtime AI wildlife cameras (satellite connected) to mitigate…

Original URL: https://www.linkedin.com/posts/tsuijten_ai-enabled-camera-traps-for-protecting-elephants-activity-7016335238213607424-B0e6?utm_source=share&utm_medium=member_desktop

🎉 Jon Van Oast, Carly Batist, Ed Miller, Rita Pucci, Sam Kelly, Dhruv Sheth, Abhay

🎶 Jon Van Oast, Dhruv Sheth

Ed Miller (ed@hypraptive.com)

2023-01-04 18:08:55

Very cool. We are hoping to do something similar with the BearID Project as some point. I sent you a connection request on LinkedIn!

Jorrit van Gils (vangilsjorrit@gmail.com)

2023-01-06 05:06:30

Great work Thijs! 👏

❤️ Thijs

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-01-07 12:44:24

Read your paper. Awesome work with the use of a mini computer. I am curious what will be the cost of a single compute unit?

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2023-01-08 18:42:24

Super cool work! Sorry haven't read the full paper yet, but what's the power consumption like for the AI and satellite connection?

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-01-10 23:35:18

@Jorrit van Gils @Thijs super interested in these figures. Cost and power consumption. I have understanding of AI models and deploying them. But I don't know a lot about embedded compute. Will be great if you can shed light on it or point to some relevant resources (if possible) 🙂

Thijs (thijs@q42.nl)

2023-01-11 02:55:18

@Aakash Gupta there are many technical details in the supplementary materials of the paper.

👍 Aakash Gupta

Thijs (thijs@q42.nl)

2023-01-04 04:34:05

In this video you can see we actually witnessed a Human-Elephant conflict when we were rolling out our technology in Gabon. That is not something I experience every day 😇

Sara Beery (sara.beery@gmail.com)

2023-01-09 02:06:43

Very exciting new faculty position(s) in conservation science at Berkeley 🙂

"Please help us spread the word about our exciting new search for one or possibly more Assistant Professors in Conservation Science in the Department of Environmental Science, Policy and Management at UC Berkeley.

We view conservation science as interdisciplinary and applied by necessity; combining aspects of biology, ecology, geography, data science and environmental policy, management, and justice to dissect threats and generate solutions related to biodiversity loss, climate change, land conversion, unsustainable consumption, water scarcity, wildfire, disease, invasive species, and other challenges. We seek applicants whose work engages conservation using (but not limited to) approaches from population or community ecology, wildlife, fisheries or habitat conservation, forest or rangeland science, computer/data science or ecoinformatics, protected area or working lands management, and conservation governance, planning, policy or effectiveness.

We appreciate any help you can provide in getting this advertisement out to your groups and professional networks.

The full position description is available here: https://aprecruit.berkeley.edu/JPF03613

The deadline for applications is February 2, 2023."

aprecruit.berkeley.edu

Assistant Professor -Conservation Science - Department of Environmental Science, Policy, and Management

University of California, Berkeley is hiring. Apply now!

Apply by

Feb 2, 2023

Department

Environ Sci, Policy & Mgmt - College of Natural Resources

Original URL: https://aprecruit.berkeley.edu/JPF03613

😊 Aleksis Pirinen, Antón Álvarez, Fagner Cunha, Yseult Hb

👍 Oisin Mac Aodha, gvanhorn, Shivam Shrotriya

😎 Jon Van Oast

Benno Simmons (benno.simmons@gmail.com)

2023-01-10 09:21:37

Hi all. Just a quick question and potential collaboration opportunity on a grant application. Is anyone aware of/own any labelled camera trap datasets from UK woodlands? Any taxa is fine.

Cathy Atkinson (cathy.atkinson@highlandsrewilding.co.uk)

2023-01-12 04:28:37

I might be able to help, depending on what you need

Mahir Patel (mahirp@bu.edu)

2023-01-10 16:10:23

Hi everyone, I just wanted to share our recent work on a species-independent 3D Pose Optimization toolkit that works not only on lab settings (our Rodent3D dataset, Deepfly3D, Rat7M) but also on open-field settings (AcinoSet).
It was recently published in IJCV, and we also presented it in the last CV4Animals workshop. Since then, we have been working on a pose optimization software (work in progress) that complements widely used toolkits such as DeepLabCut and DANNCE.
It generates robust 3D tracking and postural analysis as of now only for calibrated camera systems; I hope it will be helpful to people from diverse fields once it matures.
https://www.cs.bu.edu/faculty/betke/OptiPose

🙌 Josh Seltzer, Timm Haucke, Viktor Domazetoski, Aleksis Pirinen, Vincent Christlein, Paul Janson, benjamin de charmoy, Aamir Ahmad, Sara Beery, Yseult Hb, Genevieve Moat

👍 Ed Miller

Vincent Christlein (vincent.christlein@fau.de)

2023-01-11 03:39:59

Hi everyone!

👋 Nora Gourmelon, Robin Zbinden, Declan, Jose Ruiz-Munoz, Suzanne Stathatos, Mahir Patel, Aleksis Pirinen, Rita Pucci, Sara Beery, Lindsey Dukles, Yseult Hb, Ronan Wallace

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-01-11 14:34:05

Hi everyone! We at the Flight Robotics and Perception Group just released SMALify: https://github.com/robot-perception-group/smalify -- a PyTorch based method to estimate pose and shape of animals for the SMAL animal model from multi-view images of the animal as input. For now it is tested with videos of Przewaskis horses and Grévy's zebras recorded with multiple drones flying simultaneously around them. It is easy to use and well documented, with examples and demos. It is developed for multi-view drone images but obviously not limited to it. One can use any number of camera sources. Will be happy to help anyone who wants to use it!

robot-perception-group/smalify

pytorch-based articulated animal model (SMAL) fitting

Language

Python

Last updated

42 minutes ago

🙌 Sofía Miñano, Sara Beery, Paul Janson, Antón Álvarez, Moira Shooter

Silvia Zuffi (silvia@mi.imati.cnr.it)

2023-01-12 02:02:14

Nice, but maybe not the best choice of name. There is already an existing Smalify method from Benjiamin Biggs that is quite popular, maybe you can rename with SMALifymv for multiview? https://github.com/benjiebob/SMALify

benjiebob/SMALify

This repository contains an implementation for performing 3D animal (quadruped) reconstruction from a monocular image or video. The system adapts the pose (limb positions) and shape (animal type/height/weight) parameters for the SMAL deformable quadruped model, as well as camera parameters until the projected SMAL model aligns with 2D keypoints and silhouette segmentations extracted from the input frame(s).

Stars

Language

Python

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-01-12 02:21:13

Thanks! yes, good idea! too many similar names around 🙂

Tjomme Dooper (tjomme@fruitpunch.ai)

2023-01-16 05:35:03

Hi friends,

FruitPunch AI 🍉 has teamed up with researchers from Cornell & Stanford, and tech partners like AWS, Edge Impulse and RFCx to work on acoustic monitoring of elephant rumbles. using AI. Up to 50 AI enthusiasts and experts from all over the world will work together for 10 weeks to improve an audio processing pipeline, optimize ML models and implement them on-edge, starting in a month from now.

With the AI for Forest Elephants Challenge, we'll develop a prototype that can be deployed in the field for follow-up research into elephant populations and to prevent poaching and human-wildlife conflict.

We have some spots left for anyone with experience in Python and about a day per week to spare between February 17th and May 2nd. Registrations are open until February 17th. I'd love to see some of you there 🙂

fruitpunch.ai

AI for Forest Elephants Challenge

Detect forest elephant rumbles on sound recordings and optimize the model to implement on-edge

Original URL: https://www.fruitpunch.ai/challenges/ai-for-forest-elephants

🙌 Stephanie O'Donnell, Ali Johnston, Lindsey Dukles, Ronan Wallace, Sofía Miñano, Ed Miller, Yseult Hb, Dan Morris, Abhay, Aakash Gupta, Sara Beery, Dhruv Sheth, Alexander Kobald, Nicolas Arrieta Larraza

😎 Jon Van Oast, Rebecca Wilks

🎉 Jan Kees

Nick Giampietro (giampiet@pdx.edu)

2023-01-16 23:27:05

Is there anyone actively working on using AI/ML and remote-sensing techniques to aid/simplify the field work required for government reports like the US NEPA Environmental Assessment? There's surely a lot of overlap with many more general projects folks are doing, but I'm specifically looking for people using them to aid with stringent government bureaucratic processes 🙂

Declan (declan.pizzino@consbio.org)

2023-01-17 11:23:18

This is in line with the sort of work that we do at CBI- providing science, analyses, and tools to support decision-makers. Unfortunately the fun RS + AI/ML projects happen (or are funded) less often, but we've got a couple that are active for the upcoming year or two.

Our current big project spinning up right now is for the USDA Conservation Reserve Program. The RS + AI/ML is just one component of many for this project.

We'll be doing some updates for habitat models supporting the Stephens' kangaroo rat Rangewide Conservation Plan

Conservation Biology Institute

Decision Support for the USDA Conservation Reserve Program - Conservation Biology Institute

The USDA Conservation Reserve Program (CRP) works with private landowners to advance conservation on their lands. This voluntary program currently comprises over 20 million acres making it an extremely important […]

Est. reading time

2 minutes

Original URL: https://consbio.org/projects/decision-support-for-the-usda-conservation-reserve-program/

Conservation Biology Institute

Rangewide Management & Monitoring Plan for the Stephens' kangaroo rat - Conservation Biology Institute

Agencies and scientists coordinate to protect Stephens kangaroo rat - an endemic species of Southern California.

Est. reading time

2 minutes

Original URL: https://consbio.org/projects/rangewide-management-monitoring-plan-for-the-stephens-kangaroo-rat/

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2023-01-17 12:08:22

Hi everyone,

I am conducting a ~10 minute survey ⏱️📋 which I hope you can share with your networks 🤝:

🔗 https://ec.europa.eu/eusurvey/runner/research-baseline-survey

🔗 shortened link: https://t.ly/GWLb

This survey asks questions about the research culture in your field/discipline, including possible barriers to doing open science such as publishing ideas, methods, data, etc. I believe it is relevant here because better open research practices will enable the innovation that's needed for conservation!

⏳ The deadline for submission is end-of-day on Sunday, February 5 2023 (AoE).

:firstplacemedal: Everyone who completes and submits this survey can choose to enter a drawing for a prize worth GBP £25!!!

Anonymised results will be summarised in a report I am writing for UK Research and Innovation (UKRI), the primary public research funder in the United Kingdom. It will inform a better understanding of, and possible reforms to, research institutions and the development of new research publishing platform(s). I am doing this wearing my hat 👒 as an open science researcher at the University of Bristol in the Octopus project 🐙.

Most importantly, please share this survey among your networks! I hope to get responses from researchers representing diverse disciplines in the natural and social sciences, especially those who are NOT strongly aware of or advocates for open science. Or, if you can suggest other places to share this survey do let me know.

Please respond if you have any questions/concerns, and thank you in advance for your time! 🙏🙇‍♀️

✅ Remi Gosselin, Sara Beery

Catherine Villeneuve (catherine.villeneuve.9@ulaval.ca)

2023-01-17 16:41:03

Are you interested in Predator-Prey Ecology | Machine learning | Movement analysis | Modelling?
BIOS2 (A Canadian computational ecology training program) and Sentinel North (A large trandisciplinary Arctic research and training strategy funded by the Canada First Research Excellence Fund) are organizing an advanced field school in computational ecology 🤖🐻‍❄️ from May 19 to 26 2023 in the beautiful setting of the Couvent de Val-Morin, Canada. Join us in a cozy cottage, by a beautiful lake, in the Boreal Forest 🌲

The field school is aimed at international grad students and post-docs in ecology, familiar with R or Python (but no need to be an expert!). Once at the cottage, we will create an original data set by playing a predator-prey interaction game called TrophIE (Trophic Interaction Experiments), where bio-logged players impersonate predators and preys in a real-life context. We will then analyze the data through 5 days of intensive workshops led by experienced mentors.

The goal is to delve into predator-prey theory and gain skills in machine learning, behavioural classification, movement analysis and advanced modelling of interaction networks.

For more information and to register: https://sentinellenord.ulaval.ca/en/ecology2023
Applications are open until 1st February 2023 (with a potential for a short extension of the deadline) and is limited to approximately 35 graduate students and post-docs.

Feel free to reach out to me if you are interested and have any questions 🙂

😮 Jon Van Oast, Arjun Subramonian (they/them), Benjamin Hoffman, Toryn Schafer, Yseult Hb, Sara Beery

🦊 Suzanne Stathatos, Arjun Subramonian (they/them), Benjamin Hoffman, Sara Beery

🐰 Suzanne Stathatos, Arjun Subramonian (they/them), Benjamin Hoffman

👍 Savannah Bissegger O'Connor

Savannah Bissegger O'Connor (savannah.bisseggeroconnor@gmail.com)

2023-01-18 15:42:53

Hi everyone! 👋

@Tim Elrick and I are happy to join the group! We are from McGill University and are estimating the scale of white-tailed deer overpopulation at the Gault Nature Reserve in Mont-Saint-Hilaire, Quebec using a drone equipped with a thermal camera. After detecting deer by eye in our ~550 thermal images, we are starting to look into the use of machine learning to detect deer signatures in our thermal images. We look forward to exchanging ideas here in Slack and would be happy if you could point us to studies or literature on this topic. ☺️

👋 Declan, Catherine Villeneuve, Aamir Ahmad, Dan Morris, Suzanne Stathatos, Jon Van Oast, Timm Haucke, Valentin Gabeff, Ali Johnston, Sara Beery, Björn Lütjens, Aakash Gupta

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2023-01-18 18:04:58

I built a drone-based system for thermal rhino detection a few years ago - basically the same thing. You will probably get good results provided you have decent contrast between the ground and the animal, especially if you can do it by eye. Here's a paper from my old group that discusses thermal animal detection at a high level:

https://www.astro.ljmu.ac.uk/~aricburk/Burke2018.pdf

Otherwise you can fine-tune normal "colour" object detection models and they mostly just work. I would recommend trying to classify radiometric images rather than the min-max scaled images, there is very little published research on this, but it should give you an improvement. Might require some custom dataloading because you have to load the 16-bit images and then normalise them.

I also have labelled data from a few flights of various ungulate species in the North of the UK which might be useful, not under forest cover though.

I think the most important thing to consider is your flight altitude vs size of the animal in pixels in the image. Try and make sure that your deer are at least 30 px. That can be tricky if you're flying over forest and especially with a thermal cam because you only get 640x512 at best. Otherwise it really depends if there are only deer in the images (e.g. can you just look for bright stuff and then confirm later).

Feel free to DM me., I'm happy to have a call if you want to discuss further!

👍 Jon Van Oast, Savannah Bissegger O'Connor

Savannah Bissegger O'Connor (savannah.bisseggeroconnor@gmail.com)

2023-01-19 20:03:08

Hi Josh - thanks for your reply! I am actually already aware of the paper you sent and have referred to it in the past, especially the observing strategy optimization web tool!

For the "colour" object-detection models, do you have any software or algorithms in mind that we should look to? I am rather new to the machine learning world and the only software I personally have experience with is ENVI and their object-oriented classification / feature extraction tool. The tool was able to pick out the deer signatures in one of our higher contrast images, though I am not sure if this is the best tool for the job.

To touch on some of the things you mentioned: due to an unknown error, all our images appear to have turned out non-radiometric despite having a radiometric camera, and yes, we are lucky that we only had deer in our survey area (or rather, they are the only larger-sized mammal on the reserve).

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2023-01-20 05:15:57

Back in the day we used YOLO (for speed), but I think any object detector would be fine (e.g. Faster-RCNN, SSD, YOLOv8). There are quite a few "no code" solutions if you just want to upload a dataset and get a model out. It's more dependent on your dataset than the model architecture. I also think 90+% of people use non-radiometric, so that shouldn't be a problem even if it isn't optimal 🙂

What do you want to do with the detections? That may have some impact on which tool you choose, e.g. if you want to integrate with ENVI this is probably as good as any: https://www.l3harrisgeospatial.com/docs/ENVIDeepLearningTutorialObjectDetection.html

👍 Savannah Bissegger O'Connor

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-01-21 23:42:31

This is very interesting! Thanks for sharing. For this particular use-case you should think about how the model will generalize for out-of-sample images. Does the data points allow you to identify deer vs non-deer objects in the image?

👍 Savannah Bissegger O'Connor

Savannah Bissegger O'Connor (savannah.bisseggeroconnor@gmail.com)

2023-01-23 20:33:00

@Josh Veitch-Michaelis Thanks for the suggestions! I will look into YOLO and the others you mentioned. We essentially have ~550 non-radiometric RGB images (like the one I attached) to iterate through, each having different levels of thermal contrast due to environmental and other factors. So, I assume we will have to take into consideration that the deer signatures will have different RGB values.

Savannah Bissegger O'Connor (savannah.bisseggeroconnor@gmail.com)

2023-01-23 20:38:10

@Aakash Gupta Thanks for your comment! Please forgive my lack of expertise in this domain, but could you specify what you mean by the model generalizing out-of-sample images? (What is out-of-sample?).

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2023-01-24 19:31:10

@Savannah Bissegger O'Connor out-of-sample refers to how well your training data represents new, unseen, data you're likely to run the model on. A typical example would be you train the model in one forest, but deploy it in another and for some reason all the deer are a few degrees warmer and the model struggles to detect them.

Essentially as long as your training data looks sufficiently similar to where you'll deploy the model, you should be OK. But if you get poor performance when you deploy it, it's something to look out for.

👍 Savannah Bissegger O'Connor, Aakash Gupta

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-02-03 21:44:45

@Savannah Bissegger O'Connor , I believe Josh has explained it very well. To solve it you can use a technique known as data augmentation. You typically use image manipulation techniques to create variations of the image. This increases the total size of your training dataset as well as generalizing the model. But you should be vary of data leakage, while implementing it.

👍 Savannah Bissegger O'Connor

Jake Burton (jake.burton@fauna-flora.org)

2023-01-19 10:35:34

Hi everyone!

It’s great to be part of this group! I’m Jake and I started working as Project Officer at WILD LABS late last year.

I just wanted to share that applications are now open for our AI for Conservation Office Hours, where we are teaming up with @Dan Morris (from Google's AI for Nature and Society program) to offer conservationists facing AI or ML challenges in their work the chance to get advice from AI specialists in 1:1 virtual sessions.

If you think you could benefit from one of these sessions, then apply now! We would also be hugely grateful if you could share this in any of your networks that might be interested.

The deadline to apply is Friday 27 January 2023. Visit WILD LABS for more info and how to apply.

Thanks very much! 🐯

🙌 Sasha Luccioni, Stephanie O'Donnell, Alessandra Sellini, Toryn Schafer, Dan Morris, Sara Beery, Ed Miller, Timm Haucke

❤️ Carly Batist, Pen-Yuan Hsing, Jon Van Oast, Suzanne Stathatos, Talia Speaker, Timm Haucke

aruna (arunas@mit.edu)

2023-01-19 12:03:19

if you have 5m and you care about climate change, can you please help classify these pictures? https://forms.gle/H54vppXauJLNNMut5. Thank you! 🍀

Google Docs

Classify these pictures!

Classify the following pictures into one of 5 categories based on the most prominent features of the picture 1. Person 2. Environment 3. Industry 4. Research 5. Nationalism

Original URL: https://forms.gle/H54vppXauJLNNMut5

😍 Björn Lütjens, Kateryna Nekhomiazh

🙌 Björn Lütjens

🌍 Björn Lütjens

Sasha Luccioni (sasha.luccioni@huggingface.co)

2023-01-19 12:03:46

what's the purpose of this?

Björn Lütjens (bjoern.luetjens@gmail.com)

2023-01-19 12:06:49

done - very fun and took less than 5min @Sasha Luccioni aruna is a kick-ass grad student in climate misinformation at MIT. i'm assuming this is for some of her research.

👍 Sasha Luccioni

❤️ aruna

aruna (arunas@mit.edu)

2023-01-19 12:07:55

Thanks bjorn!

🙌 Björn Lütjens

aruna (arunas@mit.edu)

2023-01-19 12:08:04

Sasha, yes, it's for a climate misinformation project. 🙂

👍 Sasha Luccioni

aruna (arunas@mit.edu)

2023-01-19 12:39:45

Thanks everyone! ❤️ all good here, just closed the form. 🙂

Suzanne Stathatos (suzanne.stathatos@gmail.com)

2023-01-19 16:26:17

Has anyone here used Google Coral before for a project and if so, would you be willing to chat with me about it?

Alan Papalia (alanpapalia@gmail.com)

2023-01-19 19:23:57

spent a decent amount of time on one this summer - overall takeaways here so others can see but happy to chat more as well!

Strengths: great at everything it's stated to be good at (fast, efficient, relatively easy to use if you can get a model that is compatible)

Ease of use: have to spend some time wrangling models to get them to compile to the format the corals accept (don't remember the exact format, but there were quite a few operations used in many models that were not supported). Best-case is you're using a model that already has a well-demonstrated ability to compile to the coral format (e.g. YOLO)

Model size: must use very small models. I'd have to look up docs to remember how small this is but I think it was somewhere in the range of yolo-small or yolo-tiny. Definitely saw some performance loss due to needing very small networks, but maybe could have clawed back some performance using clever tricks elsewhere.

👍 Jon Van Oast

Ed Miller (ed@hypraptive.com)

2023-01-19 20:09:02

@Sam Kelly @Henrik Cox (Sentinel) Did you use Google Coral?

Sam Kelly (sam@conservationxlabs.org)

2023-01-19 20:16:18

Hi Suzanne -
Yes, we use the Coral SoM in the Sentinel. Would be happy to chat and share any learnings! Overall it’s pretty good, just some little quirks (I concur with what Alan mentioned) that I would be willing to chat about, definitely would love to hear more about how you are planning on using it.
I know that @Thijs uses the Coral in some of their work.

Suzanne Stathatos (suzanne.stathatos@gmail.com)

2023-01-20 12:41:09

Thank you all for your responses! Cool, right, I was chatting with a group that is trying to use it as an end-to-end system to monitor animals in an urban setting. From what everyone has said so far, it seems superb if you can use their out-of-the-box models and pretty good if you can use small models that can work via tflite.

Thijs (thijs@q42.nl)

2023-01-23 09:59:41

Please let us know how it works for you @Suzanne Stathatos or if we can help in any way!

Ronan Wallace (rwallace@macalester.edu)

2023-01-22 12:39:43

Hi everyone! I'll be going on a little roadtrip soon, and I'm looking for some good podcasts on conservation tech. Are there any out there that you've enjoyed? (thank you so much!!) 😄

Josh Seltzer (jyseltz@gmail.com)

2023-01-22 13:07:42

I'd love to hear some recommendations as well! The Mongabay podcast sometimes covers things related to conservation tech, but outside of that i haven't been able to find many good ones. There's a ton of great conservation and climate tech podcasts i've been listening to (happy to recommend if anyone's interested) but conservation tech seems definitely underrepresented!

Ștefan Istrate (stefan.istrate@gmail.com)

2023-01-22 16:11:09

I really enjoyed @Roland Kays’ Wild Animals: https://open.spotify.com/show/1M5daW5fcOtXOjYkHkrZQF

Spotify

Wild Animals

Original URL: https://open.spotify.com/show/1M5daW5fcOtXOjYkHkrZQF

🙏 Roland Kays

Roland Kays (rwkays@ncsu.edu)

2023-01-22 18:02:31

thanks - about to release season 3!

Roland Kays (rwkays@ncsu.edu)

2023-01-22 18:04:18

Also working on a new you tube channel, 2 episodes up so far https://www.youtube.com/channel/UCTMtNawoWB8-7M8waKnuiqg

YouTube

Wild Animals

Stories about wild animals and wild places, told with the help of 1000's of cameras.

Original URL: https://www.youtube.com/channel/UCTMtNawoWB8-7M8waKnuiqg

🙌 Josh Seltzer, Ștefan Istrate

aruna (arunas@mit.edu)

2023-01-22 23:42:50

Thank you so much for participating in my last survey. Here's another survey that should take <5m on classifying images. Some of the images in this survey can be triggering so please participate only if you are able to. Thanks in advance: https://forms.gle/xyjtGPnU6b7eygvc8. Please direct any questions to me via DM (as opposed to on this channel/on the thread) so as to not bias others about the survey.

Google Docs

Classify These Pictures

Classify the following pictures into one of 5 categories based on the most prominent features of the picture 1. Emissions 2. Extreme climate events 3. Plastic pollution 4. Activism 5. Renewable energy

Original URL: https://forms.gle/xyjtGPnU6b7eygvc8

👍 Omiros Pantazis, Vijay Karthick

Thijs (thijs@q42.nl)

2023-01-23 10:00:43

Are there happen to be people here who deployed a YOLO(v5) model in the cloud? I'm trying to deploy one on Google AppEngine and I'm looking for some best-practices.

Dan Morris (agentmorris@gmail.com)

2023-01-23 12:13:23

Some folks I know who are using MegaDetector v5 (which is just a trained YOLOv5) in the cloud, though AFAIK none via AppEngine: @Nicholas Osner, @Nathaniel Rindlaub, @Matt Hron

Thijs (thijs@q42.nl)

2023-01-23 13:22:51

Awesome, thanks for the references. I'm trying to figure out what the most "lightweight" way is to load and run the model. For "native YOLOv5" it seems I have to include a ton of libraries.

I also exported my model to tflite but I cannot really find examples on how to load and run that model (my default tflite scripts don't seem to work).

I'm also experimenting with openvino

Dan Morris (agentmorris@gmail.com)

2023-01-23 13:57:21

How are you measuring "lightweight"? If you measure by the disk footprint of your dependencies and/or the install time of your dependencies, yes, YOLOv5's native inference path is heavy, mostly because PyTorch itself is very large. But if you define "lightweight" by engineering complexity, I would consider eating the disk footprint of YOLOv5's dependency list. Their dependency setup has been tested in lots of environments, plus if you export to other formats, you likely won't get exactly the same results, which may complicate your debugging.

But if you are in an environment where disk space and/or install time are constrained, or where you already have lots of dependencies that are already compatible with one of the other export formats (e.g. TFLite), those are IMO great reasons not to use the native inference path.

Curious what you end up doing!

Thijs (thijs@q42.nl)

2023-01-23 14:19:04

Yeah, currently my app won't even deploy because of the huge list of dependencies...

OpenVino seems to be nicer / quicker and less bulky. But I find it very hard to find the correct python script to run the model with OpenVino.

@Dan Morris you mean, try to peel away as many dependencies as I can and try to stick with pytorch?

Thijs (thijs@q42.nl)

2023-01-23 14:20:33

Sometimes I really wonder how people het sh**t done in this space 😅 I've been a software engineer for over 20 years, but the whole ML space seems to be the most frustrating so far 😇 Especially if you don't "just" wanna train a model, but actually want to use it somewhere.

Dan Morris (agentmorris@gmail.com)

2023-01-23 16:32:28

It's hard for me to weigh in on whether you should try to eliminate individual dependencies or just make it work as-is in your deployment environment. Whatever issues you encountered in deployment, it may be easier to fix them and keep the YOLOv5 inference environment totally intact, as opposed to modifying the inference environment to fit your application. But, that's easy for me to say, since I don't know what those issues are and I don't have to fix them. 🙂

My simplified mental model is that all application environments are just Linux VMs that support arbitrary Docker containers, in which case it probably is easiest to keep YOLOv5 exactly as it is, dependencies and all. But that model is clearly oversimplified, and if the payloads your execution environment can run have limitations, maybe it makes sense to either prune problematic dependencies away from the YOLOv5 code or switch to another runtime.

Good luck!

Thijs (thijs@q42.nl)

2023-01-24 09:22:48

It seems like this is the best option for me: https://learnopencv.com/object-detection-using-yolov5-and-opencv-dnn-in-c-and-python/

It's exporting the model to ONNX format and then I only need the opencv-python-headless dependency to do inferencing.

LearnOpenCV – Learn OpenCV, PyTorch, Keras, Tensorflow with examples and tutorials

Object Detection using YOLOv5 OpenCV DNN in C++ and Python

A comprehensive guide to Object Detection using YOLOv5 OpenCV DNN framework. Learn how to run YOLOv5 inference both in C++ and Python. OpenCV YOLOv5.

Original URL: https://learnopencv.com/object-detection-using-yolov5-and-opencv-dnn-in-c-and-python/

Michael Bunsen (notbot@gmail.com)

2023-01-23 19:46:21

Has there been any talk about upgrading this Slack workspace so we don't loose all the awesome resources older than 90 days? Perhaps we can crowd source the cost or get a sponsorship?

👍 Jose Ruiz-Munoz, Nicolas Arrieta Larraza

💯 Carly Batist, Katelyn Morrison

Andrew Schulz (akschulz@gatech.edu)

2023-01-24 03:41:59

So I believe non-profits gets one free professional slack account and that would likely be the best way.

❤️ Katelyn Morrison

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-01-24 04:37:08

non profits do!

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-01-24 04:37:27

you just need to find a properly registered non profit that will give you their account

Eddie Zhang (ete@ucsb.edu)

2023-01-24 01:55:58

I wonder how expensive it would be?

Nick Giampietro (giampiet@pdx.edu)

2023-01-24 02:03:11

About $10k/mo assuming Pro tier and no special deal from Slack 😨

Eddie Zhang (ete@ucsb.edu)

2023-01-24 04:51:46

Doesn’t seem too likely in that case 😁

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-01-24 04:54:36

Definitely go find an ngo one you can use

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-01-24 04:54:52

our slack workspace for wildlabs is free

Michael Bunsen (notbot@gmail.com)

2023-01-24 13:26:19

Is the Slack for Wildlabs public? Or is that just for your internal use?

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-01-24 13:40:41

Oh it's our internal(ish) one. It also has all the people we're collaborating with on various projects and all our community organisers/group managers in there as well

Michael Bunsen (notbot@gmail.com)

2023-01-24 18:58:54

Ah cool!

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-01-24 04:55:23

You just need a registered NGO/charity that will let you use their free workspace

👍 Jose Ruiz-Munoz

Jose Ruiz-Munoz (jfruizmu@unal.edu.co)

2023-01-24 06:34:10

Would we have to switch then to a new workspace (with a new name)?

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-01-24 06:34:39

no no

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-01-24 06:35:51

To upgrade this workspace you (the workspace owner - sara?) need to be able to input a charity number and say it's their workspace, then slack will upgrade it.

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-01-24 06:37:32

I'm sure one of the 1374 people in this group is associated with a group that isn't making use of their free account - and so would kindly donate it to the cause

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-01-24 06:38:04

OR - someone goes through the hassle (cost?) of registering 'Ai for conservation' as a charitable group?

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-01-24 06:38:44

OR someone knows someone at slack that will do us a solid and just give us a free workspace. SURELY in this group you guys know people working there

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-01-24 07:45:37

Just my 2 cents: seems like Slack free upgrade for NGO, etc. is limited to 250 users only. For more than that, there is an 85% discount -- so still some costs. the last option @Stephanie O'Donnell floated sounds best at the moment 🙂

👀 Stephanie O'Donnell

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-01-24 08:11:11

ooooh i didn't realise there was an upper limit. Also Andrew called this idea first!

Sara Beery (sara.beery@gmail.com)

2023-01-24 11:40:15

There's a limit, I've looked into this but it would still be $$$ for a community this size 😞

Matt Allen (mja78@cam.ac.uk)

2023-01-24 07:35:37

Hello! I am putting together a benchmark dataset for tree crown segmentation - do you have data and would you like to join our effort?

Call for data: we're looking for exact tree crown segmentations (rather than bounding boxes) collocated with some form of (very) high resolution aerial imagery - drone or satellite. Segmentation derived from TLS or ALS data is absolutely ideal, but high quality manual labels are also ok.

In return, we can offer:
- Co-authorship on a publication (likely at some point in 2023)
- Access to the dataset for analysis prior to the full release

If there's anything else you think that I/we might be able to offer in return, feel free to ask!

Link for registering interest: bit.ly/crownseg
Contact for questions: mja78@cam.ac.uk

Matt

👍 Stephanie O'Donnell, Emily Lines, Felipe Parodi, David Russell, Tiziana Gelmi Candusso, Sara Beery, Robin Zbinden

🌳 Tjomme Dooper

Benjamin Kellenberger (benjamin.kellenberger@yale.edu)

2023-01-24 08:01:51

@Ben Weinstein

Mikey Tabak (tabakma@gmail.com)

2023-01-24 08:38:05

Hi Matt, This sounds exciting. I don’t have the data you’re looking for, but I’m curious, what do you mean by very high resolution?

Felipe Parodi (parodifelipe07@gmail.com)

2023-01-24 10:07:13

In case you do have bounding box annotations for each tree, you could use something like boxinstseg to predict the segmentation labels

Matt Allen (mja78@cam.ac.uk)

2023-01-24 13:27:51

Re: resolution - at the moment my thinking is it would probably need to be ~30cm (I think that's the lowest resolution that it's currently possible to get? Someone feel free to correct me if I'm wrong though) to separate crowns at the individual level in areas where the species distribution is pretty homogeneous (in very diverse canopies it's quite easy to separate trees using hyperspectral data or stretching the RGB bands when you're labelling). Maybe 50cm ish would work.

👍 Mikey Tabak

Matt Allen (mja78@cam.ac.uk)

2023-01-24 13:31:19

On boxintseg etc - For the purpose of assembling the dataset I'm really looking for ground truth data rather than unsupervised predictions. It would be very interesting to see whether weakly supervised approaches like this could reproduce the manual/laser scanning derived labels. I'd need the ground truth labels to check how well it was doing though

Ben Weinstein (benweinstein2010@gmail.com)

2023-01-24 13:35:21

@Matt Allen, what's the goal? Check out the last couple datasets and the https://google.github.io/auto-arborist/ https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1009180.

journals.plos.org

A benchmark dataset for canopy crown detection and delineation in co-registered airborne RGB, LiDAR and hyperspectral imagery from the National Ecological Observation Network

Author summary Combining RGB, LiDAR and hyperspectral sensor data from the USA National Ecological Observatory Network’s Airborne Observation Platform with multiple types of evaluation data, we created a benchmark dataset to assess crown detection and delineation methods for canopy trees covering dominant forest types in the United States. This benchmark dataset includes an R package to standardize evaluation metrics and simplify comparisons between methods. The benchmark dataset contains over 6,000 image-annotated crowns, 400 field-annotated crowns, and 3,000 canopy stem points from a wide range of forest types.

Original URL: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1009180

Ben Weinstein (benweinstein2010@gmail.com)

2023-01-24 13:37:53

I think to have a new dataset there should be some kind of new/different goal. Do you have field based ground validation? You can try coarsening the 10cm data we have in that benchmark to get a sense for accuracy. I'm not sure individual tree segmentation can be rigorously done at 30/50cm. Really really hard.

Ben Weinstein (benweinstein2010@gmail.com)

2023-01-24 13:39:14

We strongly suspect that the 80% accuracy of deepforest (https://deepforest.readthedocs.io/en/latest/landing.html) at 10cm is limited by human annotators inability to do better than that. We have cross validated across many observers, included people in the field with tablets (https://peerj.com/preprints/27182.pdf). We are really unsure if we can do better with hand labeling.

Ben Weinstein (benweinstein2010@gmail.com)

2023-01-24 13:40:59

There are hundreds (hundreds) of papers on this, many include zenodo links. Some big ones are going to get integrated here when I have a second. https://github.com/weecology/DeepForest/issues/340

#340 Integrate other large datasets.

This issue is meant for cataloging data that should eventually go into a deepforest baseline. Happy to have contributions from the community on this issue. <a href="https://arxiv.org/pdf/2208.10607.pdf">https://arxiv.org/pdf/2208.10607.pdf</a> <a href="https://github.com/jonathanventura/urban-tree-detection-data">https://github.com/jonathanventura/urban-tree-detection-data</a> A portion of this could be used. <a href="https://google.github.io/auto-arborist/">https://google.github.io/auto-arborist/</a> More forest data <a href="https://lila.science/datasets/forest-damages-larch-casebearer/">https://lila.science/datasets/forest-damages-larch-casebearer/</a> Along with this issue we need a strategy for updating the baseline model and potential tradeoffs to new model weights. <a href="https://github.com/nightonion/yosemite-tree-dataset/issues/2">nightonion/yosemite-tree-dataset#2</a> Roadmap 1. Format train and test data to meet deepforest inputs. 2. Assess test, or portion of train, on existing backbone. 3. Document a training strategy on new data 4. Assess test on new model 5. Assess combined model on NeonTreeEvaluation benchmark to assess tradeoffs.

Comments

Ben Weinstein (benweinstein2010@gmail.com)

2023-01-24 14:14:38

Also check out large work being done https://openforestobservatory.org/ by Derek Young. I spoke with him a couple weeks ago and he was interested in contributing data.

openforestobservatory.org

Open Forest Observatory

Drone-based forest maps and mapping tools for ecologists and forest managers

Original URL: https://openforestobservatory.org/

🙌 Emily Lines

Emily Lines (erl27@cam.ac.uk)

2023-01-24 15:40:11

Thanks @Ben Weinstein we'll look into all of those ☺️

Ben Weinstein (benweinstein2010@gmail.com)

2023-01-24 15:41:44

Happy to zoom whenever if I can be helpful.

Ben Weinstein (benweinstein2010@gmail.com)

2023-01-24 15:44:57

I also know that the world resource institute has been doing a ton of this work for labeling/validation. By coincidence we are talking this week about data through https://www.globalforestwatch.org/. I'll know more next week.

globalforestwatch.org

Forest Monitoring, Land Use & Deforestation Trends | Global Forest Watch

Global Forest Watch offers free, real-time data, technology and tools for monitoring the world’s forests, enabling better protection against illegal deforestation and unsustainable practices.

Original URL: https://www.globalforestwatch.org/

Matt Allen (mja78@cam.ac.uk)

2023-01-24 15:46:18

Thanks for all this, appreciate it - apologies for being slow to reply; it's getting (somewhat) late here 😅

Ben Weinstein (benweinstein2010@gmail.com)

2023-01-24 15:46:35

no rush.

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2023-01-24 19:33:58

We [Restor/ETH] have around 5k 2kx2k images @ 10cm with instance segmentation labels, and we're currently field trialling. Images are from global drone data, so we generalise pretty well. Hopefully this will also be published sometime in the near future. Validation is by far the hardest problem, so we're working with partners to get ground truth where we can (it's still incredibly hard to do at the tree level, beyond asking people to eyeball images to check that the model is correct).

We opted to not segment closed canopy, because we're sceptical that you can get accurate instance segmentation with RGB alone for exactly the reason that Ben mentioned. We basically treat it as a crowd class and have the model try and distinguish between 1 vs many (and you can maybe use a different downstream task to segment the closed areas).

The difficulty with relying on other data sources that would help you disentangle is that they're hard to get consistently from restoration practitioners. LIDAR is pretty rare, though there is some evidence you can do OK with crown extraction from photogrammetry. Ideally we're targeting something that will work reasonably well on high-resolution satellite imagery which is mainly RGB (+ maybe NIR).

Would also be great to join a Zoom, I'd be happy to share where we're at.

Ben Weinstein (benweinstein2010@gmail.com)

2023-01-24 19:40:23

Sounds great, let me know when they are available and we can add them. Our model was trained only on US data and it gets applied broadly, we have no way of knowing its performance at a global scale. If you can try the release model, that would always be of interest. One very strange caveat is that we have yet to find a LiDAR + RGB integration that outperforms RGB alone. NEON's LiDAR cloud is sparse (5 points /m), which makes sense for the huge area it covers, often more than 4 million trees. Again, given lack of field validation, hard to know if because we annotated in RGB, we do better in RGB. We often see papers that use very dense LiDAR coverage and get good results, but they don't really reflect large scale field usage. This was what designers at OpenForestObservatory and I were discussing. We had a one off paper https://ieeexplore.ieee.org/abstract/document/9387530 that looked at post-hoc HSI splitting of RGB crowns. We have always meant to get back to it, but spent too long focusing on species ID.

ieeexplore.ieee.org

Multisensor Data Fusion for Improved Segmentation of Individual Tree Crowns in Dense Tropical Forests

Original URL: https://ieeexplore.ieee.org/abstract/document/9387530

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2023-01-24 19:58:28

Yes, I've spoken to other folks and they've echoed the LIDAR bit. Weird. Not something we've tried though and our images were sourced from Open Aerial Map who unfortunately don't provide surface models.

Currently I'm looking at options to QA our labels (paid-for and automated). We're also looking at crowdsourcing for label verification, something along the lines of "place a keypoint in the centre of every tree in the image". Our masks are already pretty good, but we have a lot of annotator confusion between single/mutiple trees. I think there is also promise in sub-classing the labels to try and force the models to learn better distinctions like pine/leafless/broadleaf.

It'd be great to have a chat about benchmarking (both our data on deepforest and vice versa). We do have a fair number of orthomosaics+DSMs from partner sites so one idea would just be to look at agreement between RGB model predictions and 3D model predictions. There are also a few datasets that can be used for tree count verification e.g. there's some post-disaster imagery over Tonga that has individual trees labelled (including species).

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2023-01-25 06:19:36

@Ben Weinstein I was speaking to Justdiggit today and the LIDAR question came up. Do you know what sort of pre-processing people have tried to integrate either LIDAR returns or DSMs?

Emily Lines (erl27@cam.ac.uk)

2023-01-25 06:28:37

Thanks @Ben Weinstein @Josh Veitch-Michaelis we do have some TLS-segmented data in a few ecosystems and are interested in this issue of accurately identifying where there is ground verification. @Matt Allen has been looking at other methods of ground verification too (e.g. using trunk locations from DGPS/total stations)

Matt Allen (mja78@cam.ac.uk)

2023-01-25 08:03:59

On the issue of the annotations being verifiable - I think the degree to which manual labels etc. are acceptable really depends on the ecosystem - for a lot of plantation type forests I think hand annotations should be fine. On ALS, I should clarify that I really meant drone-based ALS (ULS might be the right acronym) where the point density is much higher than say from scanners mounted to planes. In very dense canopy the labels should probably be derived from TLS, unless they can be verified in some other way. The aim is to see whether sensors derived from these labels can be reproduced using the RGB data alone. I don't think I would reasonably expect it to be spot on but it would be interesting quantify exactly how far off they are (similarly, could also see how close human annotators tend to be)

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2023-01-25 08:09:48

One thought we had was that if we can only delineate the denser areas (whilst also identifying standalone trees), then that's still a good step towards handing off to another algorithm/method. Simple/classical CV methods can work quite well in certain environments, but are very sensitive to background inclusion (e.g. if you give an area that has no trees in, they'll happily try and predict). If you can also constrain where to run a LIDAR/photogrammtry/DSM-based algorithm they should have fewer false positives.

I'm hoping that with some crowdsourcing we might be able to better quantify human annotator disagreement in different scenes (different biomes, urban vs natural, etc).

Sara Beery (sara.beery@gmail.com)

2023-01-24 11:42:04

Re: the discussions about making the slack more permanent, I would LOVE any ideas folks have! I've thought about this a bit but so far don't see any solution that is sustainable, without having a very expensive monthly payment

❤️ aruna, Katelyn Morrison, Michael Bunsen

Titus (titus@colossal.com)

2023-01-24 11:44:03

Its not terribly expensive - you could start an ngo for AI for Conservation. It would cost a couple hundred bucks a year, but would require a little bit more overhead on your part.

👍 Sara Beery

Sara Beery (sara.beery@gmail.com)

2023-01-24 11:45:33

Well, I've been thinking about it already. If there are folks with experience/cycles to help with this maybe reply in the chat and we can look into it?

Titus (titus@colossal.com)

2023-01-24 11:46:26

Ive never done it for an ngo, but I've started a few LLCs throughout the years. Happy to chat about it and see if I can help.

👍 Sara Beery

Sara Beery (sara.beery@gmail.com)

2023-01-24 11:48:13

@Aamir Ahmad expressed interest as well. I can ask Priya Donti, they just went through this process for CCAI. And maybe I can get help from MIT lawyers once I start in the fall.

👏 Titus, Katelyn Morrison

Graeme Phillipson (graeme.phillipson@bbc.co.uk)

2023-01-24 12:32:43

There are other opensource/academic communities which use Discord instead of slack (I’m in the NeRF studio discord https://docs.nerf.studio/en/latest/ for example). Discord is very much more orientated towards social communities rather than professional (so has a weird look and lots of things about emojis and gifs), but the past messages are free. Of course given the number of people already in the slack instance I’m sure people would rather not migrate!

docs.nerf.studio

Contents

Nerfstudio Documentation

Original URL: https://docs.nerf.studio/en/latest/

Silvia Zuffi (silvia@mi.imati.cnr.it)

2023-01-24 13:54:16

How much would the fee be? Could we find some sponsorship?

Jose Ruiz-Munoz (jfruizmu@unal.edu.co)

2023-01-24 14:20:15

Pro: $7.25 USD per person/month, when billed yearly $8.75 USD per person/month when billed monthly 😲

😧 Tiziana Gelmi Candusso

Silvia Zuffi (silvia@mi.imati.cnr.it)

2023-01-24 14:21:18

Thanks, that is a lot indeed!

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-01-24 15:39:19

For a group of our size, we might need Business. With 85% discount, in € I calculate about 2300€/month for our ~ 1300 strong group. Plus, our group keeps growing!

😧 Tiziana Gelmi Candusso

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-01-24 15:42:17

But it might be worth contacting the Slack sales team to get a fair idea regarding pricing? @Sara Beery as the creator of the workspace, perhaps you could drop in a query?

🙌 Kakani Katija

Jose Ruiz-Munoz (jfruizmu@unal.edu.co)

2023-01-24 16:00:09

No sure if this might apply https://slack.com/help/articles/206646877-Apply-for-the-Slack-for-Education-discount

Slack Help Center

Apply for the Slack for Education discount

The Slack for Education program offers an 85% discount on our Pro or Business+ paid plans. To see the non-discounted cost per member in your currency, head to our pricing page. If your application ...

Original URL: https://slack.com/help/articles/206646877-Apply-for-the-Slack-for-Education-discount

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-01-24 16:40:40

Indeed, this is the discount I relied on 🙂

👍 Jose Ruiz-Munoz

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2023-01-25 07:22:59

For what it's worth, I've had success with Element (which runs on the Matrix network) for collaboration within a 3-year big EU-wide research project, and use it extensively in various other communities.

Uniquely, it's entire tech stack is 100% open source which is more socially responsible, and it has seen adoption by the French and German governments for internal communications, so it's not just another random app.

The free-of-charge personal tier is sufficient for our needs, and because it is fully open source there is always to option to self-host.

I suspect most would prefer to remain on Slack so as to not create another account. But for cases where we start a new community elsewhere, IMO this is much better - ethically and socially speaking - than Discord.

👍 Dan Stowell, Sara Beery, Tiziana Gelmi Candusso, Yseult Hb, Jon Van Oast

Kakani Katija (kakani@mbari.org)

2023-01-25 13:33:31

@Sara Beery I just hired someone for OVAI who has a degree in non-profit administration, and may be similarly having to set up a non-profit. I'll know more in two weeks and happy to compare notes

🙌 Sara Beery

👍 Silvia Zuffi

Bourhan (bourhan@rfcx.org)

2023-01-26 18:33:27

Hey everyone! Bourhan here, CEO of the non-profit Rainforest Connection (rfcx.org), I think the process of incorporating a company and filing for the 501c3 status is quite extensive and lengthy especially if the primary goal is to get a subscription discount, etc. The IRS sometimes takes upwards of a year to approve a non-profit status, at times even longer.

Bourhan (bourhan@rfcx.org)

2023-01-26 18:36:33

Couple of things I can help with. We currently have a free slack subscription for non-profits and our organization was selected this year to be part of the Salesforce accelerator. Salesforce (owners of Slack) offers a 2 year free subscription as part of the accelerator for us. Given that we already have the non-profit edition, we wouldn't have any use for it so we could possibly give that up for this workspace. The key would be to convince Salesforce to do that which I don't necessarily see any issues with, though we would definitely have to make the case to them.

Bourhan (bourhan@rfcx.org)

2023-01-26 18:36:48

Let me know if y'all are interested in that and we can talk about the next steps.

Sara Beery (sara.beery@gmail.com)

2023-01-26 18:37:19

Definitely interested! I'm happy to help make that case

Bourhan (bourhan@rfcx.org)

2023-01-26 18:39:50

Awesome! I'll talk to our team at Salesforce about it. If you have any small write up on the purpose of this workspace etc... it would be super helpful

🙏 Silvia Zuffi, Dan Morris, Vijay Karthick

Sara Beery (sara.beery@gmail.com)

2023-01-25 13:41:28

New possibly relevant workshop announcement!

Call for Contributions: Workshop on Probabilistic Approaches in Weather and Climate Science (8-9 May, Kigali)

We are excited to be hosting the first Workshop on Probabilistic Approaches in Weather and Climate Science (climate-workshop23.github.io) at AIMS, in Kigali, on the 8th and 9th of May 2023. In fields such as climate science, where the impact of inaccurate and untrustworthy predictions could be extreme, there is a need for probabilistic approaches that robustly quantify uncertainty. This workshop aims to bring together researchers across a range of disciplines who are interested in probabilistic machine learning methods to tackle pressing environmental issues. The workshop will provide an engaging forum for researchers to present and discuss their work, as well as the opportunity to hear from a number of keynote speakers working in the area.

We invite submissions of short abstracts (up to 2 pages plus references) for consideration as posters or short presentations during the workshop. Submissions that explore probabilistic methods in any area of weather and climate science are welcome. We reference a few areas of interest below:

Weather forecasting

Climate model emulation

Data assimilation

Extreme events

Constraining uncertainty in climate

Climate model ensembling

Hydrology

Clouds and aerosols

Air quality

Downscaling

Details

Submission link: https://forms.gle/CsT7GGnDQv6WAb2KA

Submission deadline: 1 March 2023

Acceptance notification: 8 March 2023

We particularly welcome submissions that focus on climate and weather related issues that affect lower and middle income countries. This workshop will not publish proceedings, and previously published work is welcome.
If your abstract is accepted, we expect you to present your work in person at the event.

accounts.google.com

Google Forms: Sign-in

Access Google Forms with a personal Google account or Google Workspace account (for business use).

Original URL: https://forms.gle/CsT7GGnDQv6WAb2KA

👍 Oisin Mac Aodha, Jon Van Oast

🙌 Stephanie O'Donnell, Mahir Patel, Hamed Alemohammad

Jon Van Oast (jon@wildme.org)

2023-01-25 13:46:02

correct me if i am wrong, but isnt this the third (topic-relevant!) conference in kigali this year?? rwanda for the win! :flag_rw:

Kristina Kupferschmidt (kupfersk@uoguelph.ca)

2023-01-25 14:12:35

@Cody Kupferschmidt

Sara Beery (sara.beery@gmail.com)

2023-01-25 17:36:03

I think a lot of them are in conjunction with ICLR!

👍 Jon Van Oast

Jon Van Oast (jon@wildme.org)

2023-01-25 17:37:59

yeah the proximity in date of this one would seem to be that. i think ICCB was in june or july. kinda cool coincidence i guess. maybe a long holiday in rwanda is in order? 😏

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2023-01-25 18:55:07

Hi all, passing this opportunity on behalf of some friends and colleagues in Bonn. The abstract deadline has been extended to the 31st of Jan so there's still a week or so if you have something that you think would be good to submit. A full paper is not required at this stage. The scope is somewhat tangential to conservation, but I think the topics are extremely relevant to the work that we do and I'm sure some people here would find some overlap:

Bonn Sustainable AI conference 2023
Sustainable AI Across Borders

This is the second Sustainable AI conference organized by the Bonn Sustainable AI lab at Bonn University’s Institute for Science and Ethics (IWE). The focus of the first conference in June 2021 was to create a community of researchers in the space of Sustainable AI and to raise awareness on the topic. The second conference will focus on cross cultural perspectives to address the variety and scope of ethical issues on a global scale. An inspiration for this theme is to acknowledge the reality that certain countries play an integral role in the early production phase and the waste management but may never experience the benefits of AI.

Environmental justice throughout the development and procurement chain of AI (exploitation of planet, minerals e-waste, land and water usage)

Social justice throughout the development and procurement chain of AI (exploitation of human labor)

AI and (post)-colonialism, decolonizing AI, and digital sovereignty

AI and global healthcare

Social and environmental impacts of AI on urbanization

AI and socio-technical imaginaries/AI narratives

AI and gender

AI and Economy

AI and agriculture (food production, biodiversity)

AI and manufacturing

Conceptual foundations of sustainability

https://sustainable-ai-conference.eu/

😎 Timm Haucke, Jon Van Oast, Ronan Wallace, Kristina Kupferschmidt

Dan Stowell (dan.stowell@naturalis.nl)

2023-01-27 03:33:31

Hi all! I'm Dan and I work on AI methods for understanding animal sounds.
I'm wondering which conferences/workshops you're planning to attend this year? I know all the acoustics events (& I can suggest some), but I'd like to know where you're likely to be this year...

👋 Ali Johnston, Oisin Mac Aodha, Stephanie O'Donnell, Sara Beery, Carly Batist, Dan Morris

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-01-27 04:46:17

We're organising a movement ecology workshop at ICCB in July which will include an ML focus. We also will be organising an acoustic one with ML elements at some point this year. Will share details about both when available, and they'll also be posted in the conservation tech event calendar on wildlabs - https://wildlabs.net/events

wildlabs.net

Events | WILDLABS

Original URL: https://wildlabs.net/events

👍 Dan Stowell, Carly Batist, Jon Van Oast, Antonio Ferraz, Taiki Sakai - NOAA Affiliate

Sara Beery (sara.beery@gmail.com)

2023-01-27 11:19:38

Hi Dan! Great to see you here 🙂 I'll definitely be at CVPR and ESA, and possibly ICCV. I'm considering attending ICLR as well.

👍 Dan Stowell, Stephanie O'Donnell, Carly Batist, Dhruv Sheth

👍:skin_tone_5: Ando Shah

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-01-27 12:03:03

I’ll be at ICCB! And wildlife society (Nov). Potentially IBAC

👍 Dan Stowell, Stephanie O'Donnell, Toryn Schafer

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-01-27 18:41:18

Damn…. do I need to get myself to Portland too??

🎉 Jon Van Oast

Dan Morris (agentmorris@gmail.com)

2023-01-27 18:32:52

I just created an "issue" (really somewhere between an "issue" and a "boring blog post") on the MegaDetector repo listing a zillion MegaDetector-related todo's we haven't gotten to, plus some half-baked ideas that have come up in email threads with users, with the intention of having a better place to point people who want to get involved in making contributions to the repo:

https://github.com/microsoft/CameraTraps/issues/331

I'm not sure we'll even have bandwidth to support a lot of external contributions right now (the activation energy for doing that is high, even though the ROI is very high), so I would take this with a grain of salt wrt this repo in particular.

But I would love to have a place to point colleagues who (a) are good at writing code and (b) want to get involved in open-source projects related to conservation (or sustainability more broadly). There are some great lists of sustainability-related OSS projects out there, e.g.:

https://opensustain.tech/

...but I can't tell from that list which projects would benefit from additional non-domain-expert software engineering or data science hands on deck. E.g., to pick one at random that I don't know anything about, the "Python Toolbox for the Evaluation of Soil Moisture Observations" looks awesome, but I have no idea whether someone who doesn't know anything about soil moisture observations could contribute.

So... does anyone have favorite OSS projects (or lists of projects) where there's a high prior on someone being able to make contributions without a ton of domain expertise? E.g. maybe a couple repos that do a really good job using the GitHub "good first issue" tag?

🙌 Sara Beery, Ștefan Istrate, Viktor Domazetoski, Josh Veitch-Michaelis, Jinsu Elhance, Stephanie O'Donnell, Lindsey Dukles

🐘 Peter van Lunteren, Jinsu Elhance

Caleb Robinson (calebrob6@gmail.com)

2023-01-30 01:14:10

(self-advertising here 🙂) -- We're creating torchgeo, a "PyTorch domain library, similar to torchvision, providing datasets, samplers, transforms, and pre-trained models specific to geospatial data." to make it easy for non-domain experts to use geospatial data (mainly satellite and aerial imagery) / benchmark their cool models on geospatial datasets. An awesome first contribution for anyone is to add a dataset!

Here is the docs page to see which datasets we already have -- https://torchgeo.readthedocs.io/en/latest/api/datasets.html

Here's a list of datasets that we've put together that would be awesome to add to torchgeo -- https://docs.google.com/spreadsheets/d/1TU1T5RdVWBify6MZGVVJICOX3c7qkhfmSGN2luyFNVU/edit?usp=sharing

Also relevant, here is our "Contributing" page (https://torchgeo.readthedocs.io/en/latest/user/contributing.html) that attempts to walk potential contributors through all the extra complicated looking stuff that happens in "Github Actions".

We (5 maintainers) are pretty active on github so feel free to open an Issue if you want to discuss the details of a dataset before or while you are implementing.

👍 Dan Morris, Ștefan Istrate, Aakash Gupta

🙌 Michael Bunsen, Carl Boettiger

🎉 Burak Ekim, Aakash Gupta

Michael Bunsen (notbot@gmail.com)

2023-01-30 15:41:50

Thanks for this thread! I hadn't seen the long list of projects at opensustain.tech

But I have seen a number of ecological or domain-specific projects that could certainly use help by "general" python, database, devops and front-end engineers! Here are some ways that come to mind or that I have helped projects accomplish:

Converting code in notebooks to python modules

Finding bottlenecks: writing faster database queries, adding caching layers, turning one-at-a-time steps into batch functions.

Containerizing code (docker), making the app portable to multiple environments, making it easier for more developers to on-board & contribute

Adding support for another platform (windows, linux, or perhaps R to python)

Writing tests, implementing automation, CI/CD, making things reproducible in general

Writing and improving documentation

Utilize packages, patterns and 3rd party integrations that are common in software world, but not in the science world (black, pep8 formatting, Sentry, NewRelic, GitHub Actions, AWS S3 open license storage, etc)

Updating dependencies and testing with those versions

Improving the management of data in general, perhaps switching to a database, or making it easier to share and retrieve large files or images. Removing large files & log files from the git repo.

Improving or creating a UI

Make the project easier to configure via environment variables, CLI parameters, etc.

Adding python type annotations

Changing hard-coded local absolute paths to relative ones 🙂

Removing sensitive information from the repo & git history

🙌 Aakash Gupta, Carl Boettiger

Dan Morris (agentmorris@gmail.com)

2023-01-30 16:56:02

Those all sound like optimal ways to get new engineers involved in the conservation space. Do you have pointers to specific projects?

👀 Carl Boettiger

Tjomme Dooper (tjomme@fruitpunch.ai)

2023-01-31 09:23:57

The AI for Wildlife Lab at FruitPunch hosts 10-week Challenges accessible to anyone with some basic programming skills. There's domain experts involved, but usually in a stakeholder role and to give masterclasses to the participants. Most of the work is done by ML/DS enthusiasts with an interest in conservation, rather than conservationists with an interest in ML/DS.

Enrollment for an acoustic monitoring Challenge is open until Feb 17th. (AI for Forest Elephants) Usually a new Challenge in conservation launches every few months.

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-02-02 09:11:04

Please also check Omedna challenges:
https://omdena.com/projects/
There are projects and local chapters which you can join. Some interesting ongoing projects are related to extreme weather forecasts, monitoring air quality and preventing wild-fires

Omdena | Building AI Solutions for Real-World Problems

Omdena Projects

Omdena AI projects are the best way to build sought-after data science and machine learning skills while solving real-world problems.

Original URL: https://omdena.com/projects/

Carl Boettiger (cboettig@berkeley.edu)

2023-02-02 15:35:19

@Michael Bunsen ❤️ what a great list. I'm saving this for future reference. Do you have links to any specific examples you can share?

I'm particularly keen myself to become more familiar with streamlined / most used ways to do these things in python. having been in the R world a long time I'm familiar with which are better/more popular patterns for caching, unit tests, containers, dependency management, linting, documentation etc, but trying to cross over into python I rarely know where to start!

Dan Morris (agentmorris@gmail.com)

2023-02-10 12:59:00

One more note to add here... I mentioned opensustain.tech in my original post; I noticed that they also have a really useful table where they index a lot of GitHub metadata, including whether each repo uses the "good first issue" tag:

https://airtable.com/shr9we419r2TkpLkc/tblfcyw3opQsmaqQj/viwtjrGUtJZG6yGBH?blocks=hide

Sorting that table in descending order by "good first issue" seems like an awesome way to find projects. Just the fact that a repo uses the "good first issue" tag at all is a decent proxy for "we can help on-board new devs".

Also FWIW there are not a ton of wildlife conservation projects there, so if folks here have OSS projects that might be interested in external contributors, add your projects!

Airtable

Airtable | Everyone's app platform

Airtable is a low-code platform for building collaborative apps. Customize your workflow, collaborate, and achieve ambitious outcomes. Get started for free.

Original URL: https://airtable.com/shr9we419r2TkpLkc/tblfcyw3opQsmaqQj/viwtjrGUtJZG6yGBH?blocks=hide

Michael Bunsen (notbot@gmail.com)

2023-02-13 19:52:05

Here is one Python package for estimating how many beavers a given stream can support. I got excited about after it was mentioned in a published book about beavers ("Eager"). However when I went to go check it out I found it was impossible to run. Multiple file paths are hard-coded to a person's local workstation and it must be run within ArcGIS (also Windows only). It would be awesome for someone with knowledge about spatial data & python to re-work this package to run outside of arcgis (using shapely, etc). Or perhaps just port it to QGIS.
https://github.com/Riverscapes/pyBRAT/

For Java developers, or big data engineers, it would be awesome to give GBIF some love! It's a big repository for biodiversity data from many sources (including iNaturalist). Any improvements to GBIF will help out many conservation efforts around the world.
https://github.com/gbif

Catalog of Life & Checklist Bank are also big projects that need help. These are working hard to create harmonized taxonomies for all walks of life. As well as create tools for different organizations to merge and keep up with changing taxonomies. They have a newer javascript/html portal that needs help.
https://github.com/CatalogueOfLife

Michael Bunsen (notbot@gmail.com)

2023-02-13 19:59:53

One more! I think Android / iOS apps for DIY camera traps would be very helpful for many projects. Here is one I know if now that could use some help "BioLens" (maybe formerly "AutoMoth"). Perhaps someone can add support for Tensorflow Lite models??
https://github.com/bhostetler18/BioLens

Beckett Sterner (bsterne1@asu.edu)

2023-02-02 15:28:09

https://wwfus.zoom.us/webinar/register/WN_oPbJlubeS9OhkGQtnnH_Vg

Beckett Sterner (bsterne1@asu.edu)

2023-02-02 15:28:23

Artificial Intelligence and Conservation: Indigenous AI

Beckett Sterner (bsterne1@asu.edu)

2023-02-02 15:28:32

Seminar Overview
Seminar Five - Artificial Intelligence and Conservation: Indigenous AI
Indigenous people have been leveraging technology to achieve their conservation goals along with everyone else focused on conservation. However, there is a long history of exploitation and colonization of these communities that must be acknowledged and eliminated moving forward. Agreements like Free, Prior and Informed Consent and the U.N. Declaration on the Rights of Indigenous Peoples, and principles like the CARE Principles for Indigenous Data Governance, apply to data being gathered for AI efforts and must be considered. Furthermore ownership or authority over Indigenous and local data must be respected. This session will cover some of the amazing AI work being done by Indigenous groups and will touch on issues such as data sovereignty and how those impact the growth and application of AI techniques.

Speakers:
Michael Running Wolf, Indigenous AI
Mason Grimshaw, Earthrise Media

👍 Jon Van Oast, Carly Batist, Justin Kay, Sara Beery, Yseult Hb, Alessandra Sellini, Sasha Luccioni, Kristina Kupferschmidt

👀 Carl Boettiger, Sara Beery, Kristina Kupferschmidt

Clare Price (theclareprice@gmail.com)

2023-02-10 16:44:51

Hello! I was wondering if this was recorded? I would love to listen and learn, but just missed the seminar. Thanks!

Nora Gourmelon (nora.gourmelon@fau.de)

2023-02-03 07:57:54

Hi all,

the AI-newcomer award is granted by the German Association of Computer Science (Gesellschaft für Informatik) to young researchers under 30 years for innovative developments in the area of artificial intelligence.
I'm one of three finalists for the award this year in the field of natural and life sciences.
If you would like to support me and make research at the interface of sustainability and AI more visible, you can vote for me at https://kicamp.org/ki-camp-2023/ki-newcomerinnen-2023/
If you would like to take a look at my research, visit my profile on my institute's website: https://lme.tf.fau.de/person/gourmelon/

Best,
Nora

🏆 aruna, Pen-Yuan Hsing, Vincent Christlein, Viktor Domazetoski, Alan Papalia, Timm Haucke, Tiziana Gelmi Candusso, Nicolas Arrieta Larraza

🎉 Jon Van Oast

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2023-02-03 16:12:28

just voted, good luck with the grant application!

🤗 Nora Gourmelon

Maxime Cauchoix (mcauchoixxx@gmail.com)

2023-02-08 09:29:48

Hello everyone, we setup a special issue to take some time to thing more broadly about the pro and cons of automation in conservation and ecology https://www.frontiersin.org/research-topics/53052/can-technology-save-biodiversity . Feel free to propose a contribution! Thanks!

Frontiers

Can Technology Save Biodiversity?

While technological development has been a key driver of climate change and biodiversity loss, humans from high income countries continue their unbridled race for innovation and technological development. Throughout history, technological developments have allowed humans to exploit more and more natural resources. These technological developments have, themselves, consumed more and more natural resources and generated various types of pollution. This virtuous or vicious circle, depending on the point of view, culminates today with the advent of electronics and digital technology. This Research Topic aims at examining the paradoxical relationship that has been established in recent years between the flagships of new technology such as embedded systems and artificial intelligence (AI), and the entities involved in the study and protection of life such as ecological research and conservation policies. Articles may include experimental studies, opinion pieces, systematic reviews and meta-analyses on advances in conservation biology as a result of new technologies, the limits of these approaches and their potentially harmful consequences for humans and other animals as well as cost-benefit analyses such as the trade-off between the potential gain in scientific knowledge and the generation of new forms of pollution.We invite studies examining automated environmental monitoring, ecological impacts of environmental sensors and communication technology including new tools to evalu...

Original URL: https://www.frontiersin.org/research-topics/53052/can-technology-save-biodiversity

🎉 Jon Van Oast, Yseult Hb, Ando Shah, Maxime Cauchoix, Carl Boettiger, Pen-Yuan Hsing

Kalindi Fonda (kalindi.fonda@gmail.com)

2023-02-10 22:25:24

Oh nice idea! 🥳
I actually came here today with the intention of posting my calendly and asking if anyone wants to chat and meet 1:1, but I'll wait until after the group coffee. And maybe then if there are more people who are interested in something like this we can make a random-coffee-partner channel. 🌱
(and yes please I'd love to be added to the calendar invite kalindi.fonda@gmail.com)

❤️ Carly Batist

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-02-14 13:42:50

Love this idea dan! Incidentally... this just so happens to be the half hour before our Feb Variety Hour, so you can roll into an all our conservation tech gathering if you get excited! https://wildlabs.net/event/variety-hour-february

wildlabs.net

The Variety Hour: February | WILDLABS

This month, we're hearing from Yanira Pop of the Belize Forest Department and the vision to build a collaborative data sharing project to link up efforts to monitor Jaguar across the Belize landscape. Andrea Kölzsch will introduce Moveapps, a no-code analysis platform for animal tracking data and winner of the 2022 Conservation Tech Award. Abhijit Kumar Nageshkumar will share his work to develop a system to monitor green turtles in India. Finally, Dr. Gilberto Câmara will talk about important cloud services for Earth observation (EO) data analysis like Google Earth Engine (GEE) or Microsoft Planetary Computer (MPC) - but what are the strong points of each? What are their weaknesses? Join to learn more!

Original URL: https://wildlabs.net/event/variety-hour-february

Dan Morris (agentmorris@gmail.com)

2023-02-14 13:50:08

OMG, I somehow totally missed that, even though I literally searched for one and only one thing to not schedule over, and it was the WILDLABS variety hour. For some reason I just missed this. I 100% do not want to schedule over 30 minutes of that, and two hours of community-ing is a lot, so I'm going to move this hypothetical AI4C coffee a day earlier. My bad!!!

Editing my original post to change to Tuesday 2/21. Sorry!

❤️ Stephanie O'Donnell

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-02-14 13:50:35

Well 100% my fault as I literally only just published it

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-02-14 13:51:51

but as it's apparently helpful - future reference, our event schedule is as follows 😂

Stephanie O'Donnell (stephanie.odonnell@wildlabs.net)

2023-02-14 13:54:05

March 29, April 26, May 31, June 28, July 26, August 30, Sep 27, oct 25, Nov 29. Always the last wednesday of the month. AND - if anyone would like to speak about their work at any events or has someone we should feature, please let me know!

Daniel Velasco (daniel.elias.velasco@gmail.com)

2023-02-15 10:40:06

This sounds like fun! I’d love to join: Daniel.Elias.Velasco@gmail.com

Dan Morris (agentmorris@gmail.com)

2023-02-20 16:31:17

Meeting link for tomorrow's (Tuesday 2/21) AI4C Coffee "event" @ 7:30am PT:

meet.google.com/nib-pjiy-dwc

"Event" is in quotes because, as a reminder, I have no plan at all for this hour. This is not like a WILDLABS event where there are well-curated games and sound effects, this is just "here's a meeting link, YMMV". For the same reason, there is a distant possibility that we exceed the number of participants in a regular meeting, in which case... see above, YMMV.

That said, here's my ask to everyone who joins: when you join the call, post one sentence in the chat about what you do for work, and two sentences in the chat about what you do for fun. If there is awkward silence, I will pick someone at random and ask more about your hobbies. If it is a total cacophony, I won't try that hard to fix it, but it wouldn't hurt if others were ready to spontaneously create their own meetings for side chats, e.g. if I find someone who only wants to talk about Fender basses, I won't hesitate to talk about Fender basses for an hour, so eventually you'll all have to say "hey, Dan, take this to a side chat".

YMMV!

👍 Daniel Velasco, Kalindi Fonda, Sara Beery, Peter Bermant, Taiki Sakai - NOAA Affiliate, Yseult Hb, Ștefan Istrate, Akash Jaiswal, Risa Shinoda

🎉 Carly Batist, Sara Beery, Michael Bunsen, Stephanie O'Donnell, Viktor Domazetoski, Nora Gourmelon, Jon Van Oast

🎸 Matt Weldy, Michael Bunsen, Nora Gourmelon

😂 Stephanie O'Donnell, Talia Speaker

👋 Eelke

Dan Morris (agentmorris@gmail.com)

2023-02-21 11:27:06

Thanks everyone for joining! We learned that AI for Conservation folks like to climb, and that underwater rugby is a thing.

❤️ Stephanie O'Donnell, Sara Beery, Jon Van Oast, Suzanne Stathatos, Akash Jaiswal, Carly Batist

🧗 Taiki Sakai - NOAA Affiliate

🎉 Jon Van Oast

☕ Jon Van Oast

Daniel Velasco (daniel.elias.velasco@gmail.com)

2023-02-21 11:27:39

Fun meeting. it was great meeting you all!

👍 Dan Morris

Kalindi Fonda (kalindi.fonda@gmail.com)

2023-02-21 11:30:21

🌊
Oh yes underwater rugby 💙, here's the map with the teams: https://www.uwrmap.com/ let me know if you are actually interested in trying it out and I can connect you with the people, but even if you just show up, people are usually very welcoming.

🙌 Suzanne Stathatos, Dan Morris

Jon Van Oast (jon@wildme.org)

2023-02-21 12:08:17

oh drat. only just now noticed this got moved ahead to today. 😭 very sorry to miss you all. i guess i will wait til next one. (at least i have consolation that i have met some of you, including dan, in person, right? wahhh) 😁

👍 Michael Bunsen, Dan Morris, Timm Haucke

Suzanne Stathatos (suzanne.stathatos@gmail.com)

2023-02-21 13:12:10

Thanks for organizing, Dan! If anyone wants to chat about marine stuff 🐟 or hiking 🧗‍♀️, don’t be a stranger

🎉 Jon Van Oast

🐋 Taiki Sakai - NOAA Affiliate, Dan Morris

Peter Lawrence (peter.lawrence@cumbria.ac.uk)

2023-02-13 13:04:10

Hi all, really happy to join. I have a working knowledge of AI in various conservation settings but hoping to develop some clearer projects and will be on the hunt for collaborators 🙂

👋 Sara Beery, Sankaran (shun-ka-run), Daniel Velasco

Sara Beery (sara.beery@gmail.com)

2023-02-13 15:57:25

New NASA Applied Remote Sensing Training Program (ARSET) relevant to this community: NASA ARSET has opened a new open, online intermediate webinar series: Biodiversity Applications for Airborne Imaging Systems. This four-part webinar series will focus on NASA Earth Observations (EO) that can be used to characterize the structure and function of ecosystems and to measure and monitor terrestrial and aquatic biodiversity.

If you have any questions, feel free to email natasha.r.johnson-griffin@nasa.gov.
More info on the training and how to register for free can be found here: https://nam02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgo.nasa.gov%2F3DkCnvs&data=05%7C01%7Ccbradley%40biologicaldiversity.org%7C64efe90aaab84844613b08db0b93084b%7C95c0c3b8013c435ebeea2c762e78fae0%7C1%7C0%7C638116498368338551%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=R%2BPkvKOgpk%2FS8EswcwjlMjwVfHqn5b8tbFqPijmfhT4%3D&reserved=0

appliedsciences.nasa.gov

ARSET - Biodiversity Applications for Airborne Imaging Systems | NASA Applied Sciences

Original URL: https://nam02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgo.nasa.gov%2F3DkCnvs&data=05%7C01%7Ccbradley%40biologicaldiversity.org%7C64efe90aaab84844613b08db0b93084b%7C95c0c3b8013c435ebeea2c762e78fae0%7C1%7C0%7C638116498368338551%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=R%2BPkvKOgpk%2FS8EswcwjlMjwVfHqn5b8tbFqPijmfhT4%3D&reserved=0

‼️ Justin Kay, Timm Haucke, Ronan Wallace, Wenxin Yang, Mikey Tabak

❤️ Jon Van Oast, Carly Batist, Timm Haucke, Ronan Wallace, Alex Brace, Viktor Domazetoski, Cathy Atkinson, Lucia Gordon, Talia Speaker, Kristina Kupferschmidt, Ando Shah

📡 Ronan Wallace

Justin Kitzes (justin.kitzes@pitt.edu)

2023-02-16 11:19:33

Hi everyone, our group is once again searching for one or more Research Assistants to contribute to our bioacoustics research projects. A description of the opening and how to apply can be found here - https://www.kitzeslab.org/research-assistant-position-available/. Feel free to be in touch with any questions!

The Kitzes Lab

Research Assistant positions available

The Kitzes Lab at the University of Pittsburgh is seeking several Research Assistants to contribute to analysis of acoustic biodiversity data sets. Our lab's research focuses on understanding how human alteration of natural habitat impacts species abundance and diversity, motivated strongly by an interest in informing conservation and habitat management. As part of this work, our

Original URL: https://www.kitzeslab.org/research-assistant-position-available/

👀 Stephanie O'Donnell, Risa Shinoda

❤️ Carly Batist, Yseult Hb, Sara Beery, Clare Price

🐦 Taiki Sakai - NOAA Affiliate

👍 Sara Beery

Alasdair Davies (alasdair@shuttleworthfoundation.org)

2023-02-18 15:04:46

Hi all. It's world pangolin day! so no better a time to share the news that the Paul Allen Family Foundation have funded a six year programme called Operation Pangolin to do something serious to address their decline, starting in West and Central Africa. The good news for this community is that it will include the use of open source conservation technology, with Arribada leading on development, but importantly will aim to develop reliable in-field data streams for AI / ML ingestion upstream (think regular inference data from camera traps, acoustic recorders etc but on a regular basis). I hope many of you can get involved to help the humble pangolin not to be turned into medicine and ornaments. More here: https://gfjc.fiu.edu/operation-pangolin/index.html

arribada.org

Arribada Initiative | Open Source Conservation Technology

We believe conservation technologies should be open, affordable and accessible to all.

Est. reading time

9 minutes

Original URL: https://arribada.org

gfjc.fiu.edu

Operation Pangolin

Original URL: https://gfjc.fiu.edu/operation-pangolin/index.html

😎 Jason Holmberg (Wild Me), Ted Schmitt, Aamir Ahmad, Stephanie O'Donnell, Sara Beery, Talia Speaker

🎉 Jason Holmberg (Wild Me), Carly Batist, gvanhorn, Dan Morris, Cameron Trotter, Yseult Hb, Aamir Ahmad, Dhruv Sheth, Henrik Cox (Sentinel), Stephanie O'Donnell, Antón Álvarez

❤️ Viktor Domazetoski, Carly Batist, Dhruv Sheth, Joanna Turner, Henrik Cox (Sentinel), Stephanie O'Donnell, Rebecca Wilks, Antón Álvarez

🙌 Antón Álvarez

Maciej Adamiak (adamiak.maciek@gmail.com)

2023-02-20 04:16:14

Hi everyone 👋! I'm a machine learning engineer and a geographer specialized in computer vision, spatial analysis and remote sensing. If you have something interesting to do, be sure to let me know 🙂

👍 Jose Ruiz-Munoz, Josh Seltzer, Ben Weinstein, Daniel Velasco, Aleksis Pirinen

👋 Jon Van Oast, Sara Beery, Caleb Robinson, Dan Morris, Aleksis Pirinen

Daniel Velasco (daniel.elias.velasco@gmail.com)

2023-02-21 11:29:06

Hi, does anyone know of any resources (repos, datasets, open source models/tools) that one could use to learn more about bioacoustics in ML/deep learning? Thanks!

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-02-21 11:36:09

Hi! Some useful resources to look into - the Conservation Tech Directory, WILDLABS Acoustic Monitoring Group, and I’ve got some resources/lists of review papers/etc on my website 🙂.

conservationtech.directory

Acoustic Monitoring | WILDLABS

Visit the post for more.

Original URL: https://conservationtech.directory/

wildlabs.net

Bioacoustic monitoring is one of our biggest and most active groups, with members collecting and analysing acoustic data from every type of wildlife, from birds and bats to big cats, and even reptiles!

Original URL: https://wildlabs.net/groups/acoustic-monitoring

👍 Daniel Velasco

Taiki Sakai - NOAA Affiliate (taiki.sakai@noaa.gov)

2023-02-21 11:38:56

Dan Stowell's paper is a good place to start

arXiv.org

Computational bioacoustics with deep learning: a review and roadmap

Original URL: https://arxiv.org/abs/2112.06725

💯 Carly Batist, Daniel Velasco, Yves Bas

Daniel Velasco (daniel.elias.velasco@gmail.com)

2023-02-21 18:15:18

Thank you!

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-02-21 15:59:45

Just saw that Appsilon released a new version of Mbaza AI - https://appsilon.com/mbaza-ai-update/

appsilon.com

Mbaza AI 2.0 Release - Capturing Biodiversity Data with AI - Data For Good, Machine Learning

Mbaza AI 2.0 has been released with new and improved features. Take your camera trap imagery and biodiversity monitoring to the next level!

Estimated reading time

5 minutes

Original URL: https://appsilon.com/mbaza-ai-update/

🙌 Antón Álvarez

Andrzej Białaś (andrzej@appsilon.com)

2023-02-22 03:08:53

Hey @Carly Batist, it was updated in October of last year but due to [many reasons here] I never got to write that up for our blog, we shared it with the current userbase. There is a lot of exciting stuff coming to Mbaza soon**, and once that happens I'll make sure to share it with the community better. Stay tuned and so on.

👍 Carly Batist, Sara Beery, Cara Appel, Antón Álvarez

Drea Burbank (drea@savimbo.com)

2023-02-22 18:45:16

Hey guys, super new here. I'm a project developer from the Colombian Amazon who uses GEE to help smallfarmers get paid directly for conservation with carbon credits. We're also launching a biodiversity credit with CME Group, Google/Ripple, etc. this spring. I joined the group because we have a weird ML we're working on. Photorecognition ML of iPhone panos taken by smallfarmers to identify forest integrity below the canopy. Think Google streetview for the Amazon. Curious to hear what you guys think of it.

👀 Josh Seltzer

👋 Maciej Adamiak, Marconi Campos

👍 Prach Sri

Jose Ruiz-Munoz (jfruizmu@unal.edu.co)

2023-02-22 19:37:30

It sounds interesting to me. I work at a university in Colombia and have heard of some related efforts. Feel free to DM me if you would like to chat about it

Drea Burbank (drea@savimbo.com)

2023-02-25 12:18:41

Yay thank you! Colombia has some amazing innovation happening in this space.

Prach Sri (prach@todreamalife.com)

2023-02-28 15:51:02

Savimbo and KUNGFU.AI Partner to Bring AI to Rainforest Conservation

https://hubs.li/Q01zYZRt0

Savimbo

Savimbo — Fair-trade carbon credits

Savimbo sells fair-trade carbon credits. We work directly with subsistence farmers in tropical forests to create, certify, and sell carbon credits to help entities of any size go carbon neutral. No middlemen, fast and friendly, global good. Grow something great!

Original URL: https://www.savimbo.com/?via=sp

Dan Stowell (dan.stowell@naturalis.nl)

2023-02-23 04:39:15

A naive question for the "fine-grained" folks... how can I tell if my task is "fine-grained"? In some descriptions, it seems like any species-ID task would be called fine-grained. But maybe it's all relative? For example we're also working on individual-ID which is REALLY fine-grained!

👍 Burooj Ghani

Oisin Mac Aodha (macaodha@caltech.edu)

2023-02-23 04:46:48

The term is often used and abused e.g. some species tasks are not necessarily fine-grained e.g. a duck versus a pigeon.

We often show the image below at our fine-grained workshop.

There is also a huge literature in human vision related to subordinate classification that you might have seen. https://link.springer.com/article/10.1007/s004260050047

SpringerLink

Subordinate-level object classification reexamined

Psychological Research - The classification of a table as round rather than square, a car as a Mazda rather than a Ford, a drill bit as 3/8-inch rather than 1/4-inch, and a face as Tom have all...

Original URL: https://link.springer.com/article/10.1007/s004260050047

Dan Stowell (dan.stowell@naturalis.nl)

2023-02-23 04:53:44

This image is useful, thanks - I like the way "instance recognition" is included, which is perhaps the best-defined level in the hierarchy - we don't want to generalise across instances, at that level. I still find unclear the distinction between the top 2 levels. Perhaps because human visual perception is not a relevant starting point for me! 😉

Oisin Mac Aodha (macaodha@caltech.edu)

2023-02-23 04:57:49

Even instance is not totally ambiguous e.g. what if I buy two copies of the exact same chair ... But agreed, it is likely the most well defined.

👍 Dan Stowell, Sara Beery

Sara Beery (sara.beery@gmail.com)

2023-02-23 09:55:44

I often think of it as a "how easy is it to tell these things apart" spectrum with no clear boundaries, that is somewhat individual rooted in each person's expertise/experience. For example, folks at the Lab of O find it VERY easy to tell apart bird species that I would easily lump together, and I'm pretty good at pretty niche stuff like matching ballet choreography to specific choreographers, etc. Course categories would be things that don't require any training to distinguish. But there's definitely cultural and regional differences in what those might be, which is interesting.

👍 Jon Van Oast

Jon Van Oast (jon@wildme.org)

2023-02-23 12:38:09

curious which species you are working with re: individual id. we have quite a bit of experience with it over at wildme.org and i would definitely say it falls into "fine-grained".

Elijah Cole (Deactivated) (ecole@caltech.edu)

2023-02-23 12:39:11

One nice attempt to formalize “granularity” in a machine learning context can be found here:

https://arxiv.org/abs/1912.10154

arXiv.org

Measuring Dataset Granularity

Despite the increasing visibility of fine-grained recognition in our field, "fine-grained'' has thus far lacked a precise definition. In this work, building upon clustering theory, we pursue a...

Original URL: https://arxiv.org/abs/1912.10154

👍 Oisin Mac Aodha, Dan Stowell, Sara Beery, Ben Weinstein

Dan Stowell (dan.stowell@naturalis.nl)

2023-02-23 12:40:20

Thanks all! Jon - we're working on developing individual-ID methods that work across many species - so the focus is very much on the ML methodology. Having said that, we're working with a collection of different bird datasets, as well as various terrestrial mammals
(...based on sound!)

😎 Jon Van Oast

👍 Jon Van Oast

Sara Beery (sara.beery@gmail.com)

2023-02-23 17:13:22

@Peter Kulits is also digging into multi-species re-ID, maybe you two should sync up and see if the efforts are complementary?

🎉 Jon Van Oast

👍 Peter Kulits

Dan Stowell (dan.stowell@naturalis.nl)

2023-02-24 03:57:31

Thanks! Elijah - thanks for the paper. I was actually quite surprised at their way of formalising the notion: essentially how neatly-clustered are the classes. I'll take some time to think about whether it's a good measure for us. Intuitively, I'm more drawn to the directly hierarchical formulation. Thanks

👍 Sara Beery

Howard L Frederick (simbamangu@gmail.com)

2023-02-23 10:55:55

Has anyone got experience with detecting large bird nests from aerial imagery?

Vulture nests are something we’d like to start looking for in our aerial survey imagery (oblique photos from aerial surveys of large mammals, image footprints ~ 150 x 200m). Some examples below - lots of questions around this, not least of which the paucity of training data. I wonder if we could / should photoshop some of these nests into other trees and backgrounds for training?

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-02-23 10:57:05

@Ben Weinstein

👍 Rowan Converse, Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2023-02-23 23:45:45

happy help, but heading to Colombia. @Howard L Frederick message me again in a week or so. https://deepforest.readthedocs.io/en/latest/bird_detector.html

👍:skin_tone_4: Howard L Frederick

Andrzej Białaś (andrzej@appsilon.com)

2023-02-24 03:41:56

👋 Hi @Howard L Frederick - we are working on a model to automate the counting of cormorant (shag) nests in drone imagery from Antarctica, here's a summary of a quick POC we did last year:
➡️ https://appsilon.com/yolo-counting-nests-antarctic-birds/

Now last time I checked the final model was almost ready, and we should be able to share more on this soon. A paper is also in the works, but as with those, it's coming in several months at best.

Perhaps not 1 to 1 what you are looking for but I expect you can get some good insights and I'll be happy to pass any specific questions to our ML team and see if we can help (email me at andrzej@appsilon.com, DM here, or via LinkedIn).

appsilon.com

Counting Nests of Shags With YOLO To Assess the Wellbeing of the Antarctic Ecosystem - Data For Good

Can Machine Learning be used to count nests of Antarctic birds? YOLO object detection algorithm says it can!

Estimated reading time

12 minutes

Original URL: https://appsilon.com/yolo-counting-nests-antarctic-birds/

Howard L Frederick (simbamangu@gmail.com)

2023-02-24 04:17:14

That's a very good start - will be in touch directly!

👍 Andrzej Białaś

Howard L Frederick (simbamangu@gmail.com)

2023-02-24 04:25:59

I suspect the chaotic backgrounds from African savannahs will be problematic but this is a good basis to start with

Ben Weinstein (benweinstein2010@gmail.com)

2023-03-04 09:31:02

okay, back from Colombia. Can you provide a single full size image and we can test our bird detector. It has been trained on 200,000 images, so you only need a tiny handful to customize to your situation. We tend to do annotations in QGIS. There is a video in the link above.

Adam Benzion (adam@edgeimpulse.com)

2023-02-23 12:21:38

Hi, I’m Adam from Edge Impulse. Helping SmartParks with ElephantEdge, AI powered tracking collars, the folks from Conservation X who created Sentinel, an AI powered nature camp, and WildLabs with wildlife research grants. Here to learn and help as I can. I help these organizations on a few projects and always want to learn and do more.https://techcrunch.com/2020/11/20/can-artificial-intelligence-give-elephants-a-winning-edge/

TechCrunch

Can artificial intelligence give elephants a winning edge?

With great technology, collaboration and a commitment to stop the Ivory trade that leads to most elephant deaths, there’s a real chance to save these singular creatures.

Written by

Walter Thompson

Est. reading time

6 minutes

Original URL: https://techcrunch.com/2020/11/20/can-artificial-intelligence-give-elephants-a-winning-edge/

🐘 Suzanne Stathatos, Dhruv Sheth, Toryn Schafer, Ando Shah, Sara Beery, Sara Olsson, Timm Haucke, Ed Miller, Edward Bayes

❤️ Stephanie O'Donnell, Talia Speaker, Dhruv Sheth, Sara Olsson, Timm Haucke

🎉 Carly Batist, Dhruv Sheth, Dan Morris, Dante Wasmuht, Timm Haucke

👋 Ed Miller

Sara Beery (sara.beery@gmail.com)

2023-02-23 19:29:45

From Tilo Burghardt:

Bristol AI and Nature Week

A week full of “AI for Nature” talks with a top-notch opening event on Monday 6pm at the Great Hall by Tanya Berger-Wolf (USA) who is a pioneering scientist in AI for Conservation and Robert Dawes from the BBC about AI use in their Natural History Productions. Lots of stuff to attend for all interested in AI and Nature- the full event website is at http://people.cs.bris.ac.uk/~burghard/ai_nature_week .

🎉 Jon Van Oast, Justin Kay, Suzanne Stathatos, Graeme Phillipson, Timm Haucke, Kalindi Fonda, Andrew Schulz, Reshma Ramesh Babu

👍 Michael Bunsen, Risa Shinoda, gvanhorn, Timm Haucke, Nathan Fox, Shir Bar

🙏 Andrzej Białaś, Prabath Gunawardane

Thijs (thijs@q42.nl)

2023-02-24 03:10:05

I just want to share with you that my TEDx talk is online, you can watch it here! 🎉

It's about how we are using smart cameras with machine learning to try and mitigate human-elephant-conflicts in Gabon.

Please let me know what you think of it! 🙏

linkedin.com

Thijs Suijten on LinkedIn: From code to conservation: a nerd&#39;s quest for meaning | Thijs Suijten |…

🎉 YES! My TEDx talk is online! Let me know what you think of it 🙏. I explain what Human-Elephant-Conflicts are and how I use #AI and my tech-skills to try…

Original URL: https://www.linkedin.com/posts/tsuijten_from-code-to-conservation-a-nerds-quest-activity-7034794567472160768-O9hJ?utm_source=share&utm_medium=member_desktop

🙌 Peter van Lunteren, Stephanie O'Donnell, Carly Batist, Kalindi Fonda, Talia Speaker, Cathy Atkinson, Gracie Ermi, Sara Beery, Dan Morris, Ed Miller, Andrew Schulz, Shir Bar, Rita Pucci

❤️ Jon Van Oast

Thijs (thijs@q42.nl)

2023-02-24 08:34:13

I look a bit scared in the YouTube preview image 😂

🐘 Kalindi Fonda

Peter van Lunteren (peter@addaxdatascience.com)

2023-02-26 06:16:47

Great stuff! What kind of model do you use for detecting the elephants/persons? And how long does it take to process one image on the embedded device?

Thijs (thijs@q42.nl)

2023-02-27 04:32:47

@Peter van Lunteren We use a TensorFlow Lite image classification model, it takes ~100ms to classify a single image.

Peter van Lunteren (peter@addaxdatascience.com)

2023-02-27 06:20:57

I once (very briefly and not at all thoroughly) played with the idea of running the MegaDetector model on a raspberry pi with a camera and motion detector attached. The idea was similar, to detect poachers or animals. One of the problems was the processing speed on the embedded device (especially for such a huge model like MD). You’d have to create something like a queue and hope for not too much false positives. Or train your own light weight model, but I didn’t really want to give up accuracy. And the other thing is that the existing commercial camera traps will always be better and than some raspberry pi with a camera attached. Anyhow, long story short… good that you found a way to deploy this! Very impressive

Thijs (thijs@q42.nl)

2023-02-27 07:46:44

Thanks Peter! I think there are some people that got the most recent version of MD running on a Pi. But I don't think this is an ideal solution.

We did use MD to preprocess out dataset and we basically trained our own model to detect elephants and humans. (Together with appsilon)

👍 Peter van Lunteren, Andrzej Białaś

🙏 Andrzej Białaś

Andrzej Białaś (andrzej@appsilon.com)

2023-02-28 10:09:21

Good stuff @Thijs 💪 and cheers for the shout out!

👍 Thijs

Louis Moreau (luis.omoreau@gmail.com)

2023-02-24 04:01:30

Hi everyone,
I am Louis, I lead the Developer Relations team at Edge Impulse.
I started my career developing a low-power, GPS-based, and LPWAN IoT solution to track rhinos for an African conservancy (Zimbabwe).
When not hacking hardware or automating every task I have to do manually more than 5 times, you can see me riding an electric skateboard in Lille, northern France, or scuba diving everywhere in the world 🦏 🤿
Happy to join this community!
If you have questions about Edge Machine Learning, I'd be happy to help!

👋 Andrzej Białaś, Thijs, Stephanie O'Donnell, Timm Haucke, Carly Batist, Valentin Gabeff, Suzanne Stathatos, Dan Morris, Ed Miller, Andrew Schulz, Daniel Velasco, Josh Veitch-Michaelis, Sara Olsson, Dhruv Sheth, Nicolas Arrieta Larraza

Alasdair Davies (alasdair@shuttleworthfoundation.org)

2023-02-26 16:38:57

Hi @Louis Moreau, are you still based in France and work remotely for EI? I ask as we have a developer there and it's always nice to think about future in person meet ups

Louis Moreau (luis.omoreau@gmail.com)

2023-02-27 05:31:50

Indeed, I am based in Lille, France and I work remotely for Edge Impulse 🙂

Tilo Burghardt (tb2935@bristol.ac.uk)

2023-02-26 18:52:14

Dear all,

just a last minute invite to next week's "Bristol AI and Nature Week" at https://camtrapai.github.io/ai_nature_week.html with lots of talks streamed online for anyone who would like to join....Tanya Berger-Wolf will be around and there will be lots of talks from AI for Conservation, Science Drones, Farming, Natural History TV Productions etc. There is also a new edition of the "CamTrapAI meets Ecology Workshop" running at https://camtrapai.github.io. Anyone welcome to tune in!

Best wishes,
---Tilo

camtrapai.github.io

The Bristol AI and Nature Week

Original URL: https://camtrapai.github.io/ai_nature_week.html

camtrapai.github.io

CamTrap Ecology meets AI - International Online Workshop

Original URL: https://camtrapai.github.io

👍 Ștefan Istrate, Devis Tuia, Timm Haucke, Valentin Gabeff, Kari Kuester, Andrew Schulz, Dan Morris, Nicolas Arrieta Larraza, Sara Beery, Risa Shinoda, Catherine Wang, Jan Kees, David

:thumbsup_all: Frederic Fol Leymarie

😎 Jon Van Oast

Nicolas Arrieta Larraza (n.arrieta.larraza@gmail.com)

2023-02-27 11:04:22

Is it gonna be recorded? 🙂

Prach Sri (prach@todreamalife.com)

2023-02-28 15:49:12

Hi everyone, I’m with the project developer Savimbo,

We’re building a really cool #ML algorithm Measuring jungle from the ground with #smallfarmers. This is a super novel application of AI for conservation.

This project, which is only possible because of KUNGFU.AI, is super simple, and at the same time, vastly important. Smallfarmers scan the jungle with photographs, and the algorithm says whether that jungle is intact. The algorithm is intended to scale on-the-ground deployment of microeconomics that stop deforestation, but the dataset has wider applications for species identification.

Savimbo

Savimbo — Fair-trade carbon credits

Original URL: http://Savimbo.com

kungfu.ai

Savimbo and KUNGFU.AI Partner to Bring AI to Rainforest Conservation | KUNGFU.AI Blog

In response to the need for fairer, more effective climate action, <a href="http://KUNGFU.AI">KUNGFU.AI</a> has partnered with the environmental conservation company Savimbo to help smallfarmers conserve endangered rainforest using artificial intelligence.

Original URL: https://hubs.li/Q01zYZRt0

🎉 Jon Van Oast, Sara Beery, Dan Morris, Jose Ruiz-Munoz, Yseult Hb

Jinsu Elhance (jelhance@gmail.com)

2023-02-28 19:54:56

Hello everyone!

I am interested in pursuing a Ph.D. in computer vision methods for generating higher-resolution hyperspectral imagery from remote sensing data with lower spectral and spatial resolution. I'm looking for advice and guidance from researchers in this field, and would be grateful for any pointers or insights you can offer.

Specifically, I'm interested in labs or groups that are currently working on this topic, and any relevant papers or resources you can recommend. My research interests include addressing the geospatial digital divide by improving access to high-resolution hyperspectral imagery, and I recently completed a project on combining multispectral imagery and synthetic aperture radar for mangrove species classification along Kenya's coastline.

I would greatly appreciate any discussion on this, and feel free to ask me any questions!

Rebecca Wilks (R.C.Wilks@sms.ed.ac.uk)

2023-03-01 07:30:58

Hey @Jinsu Elhance, I'm currently a PhD student at Edinburgh/Leeds SENSE Centre for Doctral Training in Satellite Imagery for Environmental Applications, which may have projects you will be interested in.

While the projects are more application based, as opposed to focusing on creating general methods of super-resolution for satellite imagery, these kind of tasks are often baked into the processing work done for an ecological task e.g. improving satellite imagery to better identify land cover types

I believe the initial round of applications is closed for this year, however there are often a handful of projects available throughout the year, so if interested then this is the place to keep an eye on: https://eo-cdt.org/projects/

Patrizia Paci (pp4649@open.ac.uk)

2023-03-31 05:43:32

I don’t know if at the group of Prof Jan van Gemert at TU Delft (Netherlands) they work on this specific topic, but they are cool guys and Jan might be interested in a new perspective. Perhaps, have a look at their publications to get an idea?

Daniel Velasco (daniel.elias.velasco@gmail.com)

2023-02-28 20:49:05

Hi, just wanted to check in here and ask if anyone in the chat applied to Meta’s 2023 AI residency? A few other applicants and I made a discord server for interviews if anyone here is interested. Thanks!

Aaron Ferber (aferber@usc.edu)

2023-03-01 00:03:32

Hi! I would also like to join if possible!

Sara Beery (sara.beery@gmail.com)

2023-03-01 10:34:46

https://twitter.com/animaltracking/status/1630944261347725313?t=dQLwf4yDQWGhOnSa_daKxw&s=19

twitter

Migration Dept., Max Planck Inst. Animal Behavior (https://twitter.com/animaltracking/status/1630944261347725313)

EMAC23: MoveApps Coding Challenge – Participate now! Develop a useful #animaltracking analysis App for #conservation, submit it to <a href="https://twitter.com/move_apps">@move_apps</a> by 31 May 2023 and get invited to the <a href="https://twitter.com/MPI_animalbehav">@MPI_animalbehav</a>, funded by <a href="https://twitter.com/EarthRangerTech">@EarthRangerTech</a> and <a href="https://twitter.com/CBehav">@CBehav</a>. <https://www.moveapps.org>

Original URL: https://twitter.com/animaltracking/status/1630944261347725313?t=dQLwf4yDQWGhOnSa_daKxw&s=19

👀 Enis Berk Çoban, Viktor Domazetoski

😎 Jon Van Oast

Panayiotis Danassis (pdanassis@g.harvard.edu)

2023-03-01 15:18:05

Interested in Multi-agent Systems and AI for Social Good?
We invite you to submit your work at AASG 2023, the 4th International Workshop on Autonomous Agents for Social Good, I am co-organizing with Bryan Wilder, Kayse Lee Maass, and Aparna Taneja!
At AAMAS 2023, London, May 29-30
Topics of interest include
- Public Health
- Sustainability
- Climate Change
- Food Security
- Conservation
- Human Rights
- & More!
Details: https://panayiotisd.github.io/aasg2023/

panayiotisd.github.io

Autonomous Agents for Social Good 2023

Original URL: https://panayiotisd.github.io/aasg2023/

❤️ Lucia Gordon

Tilo Burghardt (tb2935@bristol.ac.uk)

2023-03-02 09:58:57

Dear all,

starting in 5min at 10am EDT / 3pm GMT / 4pm Central Europe - 2nd CamTrapAI meets Ecology Workshop streamed live on Teams at https://camtrapai.github.io/join.html

See you there,
---Tilo

camtrapai.github.io

The Bristol AI and Nature Week

Original URL: https://camtrapai.github.io/join.html

🙌 Stephanie O'Donnell, Jon Van Oast, Michael Bunsen

👍 Oisin Mac Aodha, Josh Seltzer, Valentin Gabeff

👏 Rita Pucci

Katie Millette (millettek@gmail.com)

2023-03-02 12:34:44

Hi everyone,
GEO BON is organizing a Global Conference on Biodiversity and Monitoring this fall (10-13 October 2023), in Montreal, Canada. You’re all invited to this in-person conference. Themes to include, but certainly not restricted to, “AI for biodiversity change”.

Important dates:

Call for sessions: 1 – 31 March 2023

Call for abstracts: 10 April – 14 May 2023

Registration opens: April 2023

Proposals can be for Sessions, Workshops, Mini symposia, Fireside Chats or Panel Discussions.
We hope to raise funds to support registrants from emerging and low-income countries, students/early-career.

More details to come. Happy to answer questions 🙂 https://geobon.org/geo-bon-global-conference-monitoring-biodiversity-for-action/

🙌 Stephanie O'Donnell, Sara Beery, Jon Van Oast, Justin Kay, Dan Morris, Josh Veitch-Michaelis, Emily Charry Tissier, Clare Price, Andrzej Białaś, MEIXI LIN

🎉 Carly Batist, Yseult Hb, Andrzej Białaś

🥳 Andrzej Białaś

Kakani Katija (kakani@mbari.org)

2023-03-03 13:02:56

Hi everyone! In case you want to watch a Stanford/Hopkins Marine Station seminar given by our own @Sara Beery, you can register here! I’m sure she loves it when we advertise her talks here. 😉 https://events.stanford.edu/event/sara_beery_computer_vision_for_global-scale_biodiversity_monitoring

Stanford University

Sara Beery: "Computer Vision for Global-Scale Biodiversity Monitoring"

We require a real-time, modular earth observation system that unites efforts across research groups in order to provide the vital information necessary for global-scale impact in sustainability and conservation in the face of climate change. The development of such systems requires collaborative, interdisciplinary approaches that translate diverse sources of raw information into accessible scientific insight. For example, we need to monitor species in real time and in greater detail to quickly understand which conservation efforts are most effective and take corrective action. Current ecological monitoring systems generate data far faster than researchers can analyze it, making scaling up impossible without automated data processing. However, ecological data collected in the field presents a number of challenges that current methods, like deep learning, are not designed to tackle. These include strong spatiotemporal correlations, imperfect data quality, fine-grained categories, and long-tailed distributions. Our work seeks to overcome these challenges, and includes methods which can learn from imperfect data, systematic frameworks for measuring and overcoming performance drops due to domain shift, and the deployment of human-AI systems that help experts efficiently process large scale datasets. Sara Beery is an incoming assistant professor at MIT EECS' Faculty of AI and Decision Making and CSAIL, and a Visiting Researcher at Google working on Auto Arborist.

Original URL: https://events.stanford.edu/event/sara_beery_computer_vision_for_global-scale_biodiversity_monitoring

👍 Oisin Mac Aodha, Gedeon, Silvia Zuffi, Andy Viet Huynh, Jan Kees, Jason Holmberg (Wild Me)

🎉 Jon Van Oast, Declan, Gracie Ermi, Suzanne Stathatos, Adam Noach, Viktor Domazetoski, Yseult Hb, Dan Morris, Tiziana Gelmi Candusso, aruna, Andrew Schulz, Michael Bunsen, Risa Shinoda, Malte Pedersen, Carly Batist, Hirokatsu Kataoka (AIST), Dhruv Sheth, Ted Schmitt, Andy Viet Huynh, Jason Holmberg (Wild Me)

Kakani Katija (kakani@mbari.org)

2023-03-03 13:12:11

@Sara Beery if you’re in town, we should have you come and visit MBARI

Sara Beery (sara.beery@gmail.com)

2023-03-03 13:31:17

Sure! I'm driving up on Thursday, and was planning to drive back Saturday. We could maybe do something Thursday afternoon?

Kakani Katija (kakani@mbari.org)

2023-03-03 14:05:23

Let me follow up over email. It’s an especially tough week for me to organize things since I’m out of doggie-care, but it would be great to connect. More soon.

Sara Beery (sara.beery@gmail.com)

2023-03-03 14:30:17

Sg!

James Withers (james.withers@bbc.co.uk)

2023-03-13 05:26:55

Hi @Sara Beery, Thanks for the talk on Friday, It was really interesting to hear about what you’ve been working on. I was wondering if you would be able to share a link to the talk recording and/or the slides so I could look back at some of the things you mentioned and share with others.
I was also really interested in the Alaskan fisheries project you mentioned and wondered if you might have a name or link for that?

Sara Beery (sara.beery@gmail.com)

2023-03-13 12:40:18

I'll DM you my slides!

Here is the github repo for the sonar fish counting project: https://github.com/visipedia/caltech-fish-counting

Work led by @Justin Kay!!

Holger Klinck (hk829@cornell.edu)

2023-03-07 17:28:47

Hi all, I am excited to share that the BirdCLEF 2023 competition is now live. I would love to see some of you participate in it. If you have any questions, please don't hesitate to reach out. Here is the link: https://www.kaggle.com/competitions/birdclef-2023

🐦 Sara Beery, Jon Van Oast, Carly Batist, Ben Weinstein, Viktor Domazetoski, Enis Berk Çoban, Prabath Gunawardane, Fagner Cunha, Timm Haucke, Dan Morris, Elijah Cole (Deactivated), Oisin Mac Aodha, Agnethe Seim Olsen

🙌 Enis Berk Çoban, Fagner Cunha, Jose Ruiz-Munoz

Tyus Williams (tyusdwilliams@berkeley.edu)

2023-03-08 01:23:48

Hello everyone, I hope this message finds you all well. My name is Tyus Williams, I am a PhD student at UC Berkeley studying carnivore ecology and spatial ecology hoping to tackle some outstanding questions concerning feral cats and the influece human-dominated landscapes have on the occupancy of mesopredators in the East Bay Regional Shorelines of California. I'm slightly acquainted with, Sara Beery, who invited me to join the slack channel in my interest of harnessing the technological power of AI for the data management portion of my first chapter using camera trapping to observe the presence of wildlife species in response to anthropogenic stressors direct and indirect. While I don't have a CS background I know there has to be a better way than my current method of prefiltering photos (with the incredible assistance undergrads) based on defined criteria and eventually upload the sorted raw images onto wildlife insights and confirm the ID of the wildlife species I am interested in. From prior experience or expertise that you might have with camera trapping data management if you have any input on protocols or methods that can improve the speed in which I process image I would love to hear your input. People have mentioned megadetector, but I wonder if that is really any differnet from wildlife insights. If you have any questions please let me know.

Dan Morris (agentmorris@gmail.com)

2023-03-08 09:55:26

There's no right answer; everyone's problem is a little different, and the optimal workflow depends on the details of your problem, your team, your institution's data policy, etc. But a couple links that may be helpful...

I try to keep a list of systems that exist (with or without AI, local or on the cloud) here:

https://agentmorris.github.io/camera-trap-ml-survey/

And we can usually help narrow down your choices if you can answer the questions here:

https://github.com/microsoft/CameraTraps/blob/main/collaborations.md#questions-about-specific-camera-trap-use-cases

It may be interesting to have that discussion here, but if that exceeds what's practical to type into the Slack reply box, feel free to email cameratraps@lila.science .

Camera Trap ML Survey

Overview

Everything I know about machine learning and camera traps.

Original URL: https://agentmorris.github.io/camera-trap-ml-survey/

Antonio Ferraz (antonio.a.ferraz@jpl.nasa.gov)

2023-03-09 02:01:23

Hi Tyus, interesting topic. What co-variates are you considering to characterize the human-dominated landscape that can help explain the predator’s occupancy?

Tyus Williams (tyusdwilliams@berkeley.edu)

2023-03-09 02:46:37

Great question, I am looking at housing density, population density, road distance, vegetation cover, and residential housing distance. I'm trying to see if I can gather noise levels but we will see.

👍 Antonio Ferraz, Dan Morris

Jan Kees (jankees.schakel@sensingclues.org)

2023-03-09 14:43:19

did you already check out TrapTagger? Your case sounds like a perfect fit

Toryn Schafer (tschafer@tamu.edu)

2023-03-09 12:51:37

Hello! I am supervising a master's level statistics capstone project on herpetological camera trap data. It's primarily a data analysis project (no novel methodological developments) using supervised CNN. I am hoping we can turn it into a short manuscript. If you have a suggestion for a journal or outlet, let me know! Thanks

Diego Marcos (diego.marcos.gonzalez@gmail.com)

2023-03-09 14:17:56

The GeoLifeCLEF23 competition, within FGVC and ImageCLEF, is also open! Submit you predictions for plant species presence/absence using satellite images and time-series before May 17th
https://www.kaggle.com/competitions/geolifeclef-2023-lifeclef-2023-x-fgvc10/

kaggle.com

GeoLifeCLEF 2023 - LifeCLEF 2023 x FGVC10

Location-based species presence prediction

Original URL: https://www.kaggle.com/competitions/geolifeclef-2023-lifeclef-2023-x-fgvc10/

😍 Sara Beery, gvanhorn, Oisin Mac Aodha, Venkatesh Ramesh

👍 Leonardo Viotti, gvanhorn, Riccardo de Lutio, Nina van Tiel, Elijah Cole (Deactivated)

🌍 Oisin Mac Aodha

Andrzej Białaś (andrzej@appsilon.com)

2023-03-10 06:02:38

👋 Hey Folks!

📣 Exciting news! We just published 🔗a new article on Mbaza AI.

Mbaza AI featured here has already established a strong presence in Gabon and is soon expected to expand to Kenya (thanks to Ol Pejeta Conservancy).

More details below 👇

🤖 Ol Pejeta folks needed an accurate model to confidently automate the wildlife identification task; they provided us with a subset of manually labeled data from 2018 with images from 9 different cameras in 2 migratory corridors.

🦏 Ol Pejeta Conservancy is over 90,000 acres in central Kenya committed to conserving biodiversity, protecting endangered species, driving economic growth and improving the lives of rural communities.

Please let me know what you think in the comments! 😄

If you have any questions I’ll be happy to pass them on to our ML team, feel free to DM me 🥳 (and if you would like more news like this I have a cool newsletter 😎).

appsilon.com

Ol Pejeta Conservancy and AI - Wildlife Species Classification with Camera Traps - Data For Good, Machine Learning

Appsilon and Ol Pejeta Conservancy test AI species classification of plains wildlife from camera trap imagery datasets in the Kenyan savannah

Estimated reading time

11 minutes

Original URL: https://appsilon.com/ai-camera-trap-wildlife-species-classification/

👍 Stephanie O'Donnell, Cameron Trotter, Dan Morris, Ted Schmitt, Cara Appel, Sara Beery, Matt Hron, Aakash Gupta

🎉 Jon Van Oast, Emilio Luz-Ricca, Abhay

🙌 Antón Álvarez

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2023-03-10 06:38:36

Ol Pejeta is great, they're involved with several conservation tech initiatives. I helped out with an aerial census in 2020 for Savetheelephants just before lockdown. We flew thermal + viz cameras in a cessna, part of the novelty was assessing thermal for dawn/dusk. I was stunned to learn that digital imaging is still not the norm for surveying and that manual counts from the back of a plane is more common. I got a ride back over the Rift Valley with the passenger door off (because the cameras were mounted there) which was an interesting experience.

Photo is refuelling at the end of the day. Interestingly Ol Pejeta has an airstrip inside the park where tourists pay lots of money to fly into (there's also a larger airport in Nanyuki nearby). One of the few places where "elephant on the runway" is a valid reason to delay takeoff.

🙌 Andrzej Białaś

Jon Van Oast (jon@wildme.org)

2023-03-10 12:34:00

was about to say the same thing: we work with ol pejeta and i have been lucky enough to have been there. very cool to see more tech helping out there -- congrats!

🙌 Andrzej Białaś, Jason Holmberg (Wild Me)

Andrzej Białaś (andrzej@appsilon.com)

2023-03-10 12:35:37

Agreed, great folks there no doubt. ℹ️ We are working on packing the model and shipping it to Mbaza so it can be used. Soon 🙂

🎉 Jon Van Oast

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-03-18 12:23:38

Hi Andrzej - This is cool, congratulations!
We are working on a similar tech, which is being deployed in two tiger reserves in India. Maybe we can connect and exchange notes.

🎉 Jon Van Oast, Andrzej Białaś

Andrzej Białaś (andrzej@appsilon.com)

2023-03-23 16:23:49

Cheers @Aakash Gupta, and sure! Let's talk. Sent you a connection request via LinkedIn 🤝

👍 Aakash Gupta

Andrzej Białaś (andrzej@appsilon.com)

2023-03-24 04:12:33

Hey Folks, me again!

Mbaza 2.1.0 with the Ol Pejeta model (described in the blogpost above) added is live! The installer links and instructions can be found in our GitHub.

We'll move on to helping Ol Pejeta team to make the best use of the tool, and later focus on bringing more features (andupdate manuals, training materials and so on, it's about time 😮‍💨). Anyway, stay tuned, exciting stuff on the horizon, but as usual, lots of work to be done first 🤞.

🙌 Sara Beery, Cara Appel, Dan Morris

❤️ Matt Hron, Jon Van Oast

Jaroslav Bezdek (jaroslav.bezdek@strv.com)

2023-03-12 10:27:50

Hello, fellow machine learning enthusiasts! 👋

My name is Jaroslav, and I’m a machine learning engineer with 5 years of experience. While I love my current job, I crave more meaningful work. As a result, I’m currently looking for a side job that allows me to make a positive impact and contribute to something meaningful. 🏞️ 🐘

If you or anyone you know is looking for a skilled and dedicated ML engineer to lend a hand, I’m available for up to 20 hours per week. Please note that my previous experience and projects can be found on my LinkedIn profile. Feel free to reach out to me via comments, DMs, or email - whichever is most convenient for you. 🙏

Looking forward to hearing from you!

👋 Sara Beery, Dan Morris, Stephanie O'Donnell, Timm Haucke, Andrew Schulz, Jason Holmberg (Wild Me)

😍 Sara Beery, Kalindi Fonda, Jason Holmberg (Wild Me)

👋:skin_tone_3: Pen-Yuan Hsing

🎉 Jon Van Oast, Jason Holmberg (Wild Me)

:thumbsup_all: Frederic Fol Leymarie, Jason Holmberg (Wild Me)

Magali Frauendorf (magali.frauendorf@slu.se)

2023-03-23 04:35:09

Hi Jaroslav, maybe contributing to such as challenge would be an interesting option for you?! https://www.fruitpunch.ai/challenges/ai-for-european-wildlife

fruitpunch.ai

AI for European Wildlife by FruitPunch AI

Detect and classify different species of European wildlife in camera trap images

Original URL: https://www.fruitpunch.ai/challenges/ai-for-european-wildlife

Jaroslav Bezdek (jaroslav.bezdek@strv.com)

2023-03-23 09:25:42

Hello Magali! 👋 Thank you for the suggestion! I would rather find a project for longer cooperation, but I will definitely check fruitpunch! 🙂

George Darrah (george.darrah@systemiq.earth)

2023-06-07 07:29:56

Jaroslav - one of the most exciting companies in the world is hiring in this space... https://basecamp-research.homerun.co/?lang=en

basecamp-research.homerun.co

Basecamp Research

Basecamp Research is a London-based biological data company revolutionising protein discovery from biodiverse natural environments.

Original URL: https://basecamp-research.homerun.co/?lang=en

🙏 Jaroslav Bezdek

Kristina Kupferschmidt (kupfersk@uoguelph.ca)

2023-03-16 10:34:24

👋 Hi all, I hope you're having a great Thursday!

I am working on a project where we need to create an object detection algorithm for individual leaves from overhead images. In order to do this we will have to manually identify and annotate our images.

Does anyone have any recommendations for tools that make manual annotation easy? Thank you in advance! ☺️ 🙏

Felipe Parodi (parodifelipe07@gmail.com)

2023-03-16 10:38:59

What type of annotation? bounding box?

Lukáš Adam (lukas.adam.cr@gmail.com)

2023-03-16 10:41:51

I used https://github.com/opencv/cvat to annotate turtles. But after annotating 1000 turtles, I saw them whenever I closed my eyes. Good luck with leaves 😄

👍 Kristina Kupferschmidt, Maciej Adamiak, Josh Veitch-Michaelis, Vincent Christlein

🍃 Kristina Kupferschmidt, Sara Beery

❤️ Felipe Parodi, Ando Shah

🐢 Andrzej Białaś, Cody Kupferschmidt, Yseult Hb, Lukas Picek

Kristina Kupferschmidt (kupfersk@uoguelph.ca)

2023-03-16 10:42:29

@Felipe Parodi, yes I was thinking of using bounding box!

Felipe Parodi (parodifelipe07@gmail.com)

2023-03-16 11:14:42

i’ve found roboflow to be a great (free) bbox tool, and they have an AI-assisted labeler (once you label a handful of imgs)

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2023-03-16 11:16:04

If you have an edu account, look at Segments.ai they some neat assisted tools which are free for students/academics.

CVAT is also decent for free software - it's not as polished as the commercial offerings in my experience, but it's fine for bounding boxes. See also label studio, etc.

Eric Price (eric.price@ifr.uni-stuttgart.de)

2023-03-16 12:14:02

if you are working collaboratively, there's a number of web based solutions where you can have a lot of annotators operate on the same dataset, such as label-studio (https://labelstud.io/ ) we used that in the past and it was quite useful, although setting up the task has a bit of overhead - definitely worth it for large datasets and many annotators

Label Studio

Open Source Data Labeling | Label Studio

A flexible data labeling tool for all data types. Prepare training data for computer vision, natural language processing, speech, voice, and video models.

Original URL: https://labelstud.io/

Ben Weinstein (benweinstein2010@gmail.com)

2023-03-16 12:38:23

I also use label studio.

Ben Weinstein (benweinstein2010@gmail.com)

2023-03-16 12:38:37

Out of pure curiosity, can we see a sample image with annotations?

Ben Weinstein (benweinstein2010@gmail.com)

2023-03-16 12:39:13

what resolution can you see individual leaves? We've been talking about identifying trees from drone imagery to supplement ground data.

Dan Morris (agentmorris@gmail.com)

2023-03-16 22:27:04

I recently needed to do some viewing and editing of bounding boxes in a simple, local app, and landed on this one:

https://github.com/mfl28/BoundingBoxEditor

I've been really happy with it. I think if I was doing a Big Serious Project I would second everyone's vote for Label Studio, but to just quickly edit a small number of boxes, and/or to preview a large number of boxes, BoundingBoxEditor has been quite good.

mfl28/BoundingBoxEditor

A JavaFX desktop application for creating image-object-annotations with bounding boxes.

Stars

Language

Java

🙌:skin_tone_5: Ando Shah

Vincent Christlein (vincent.christlein@fau.de)

2023-03-17 03:27:23

We used LabelBox in the past (fine for small projects otherwise it gets costly), but during the last two years, we used only CVAT.

Ephantus Kanyugi (ndungu.kanyugi@gmail.com)

2023-03-31 15:26:19

CVAT is a great tool for polygon, polylines, bounding boxes and keypoint annotation. If you only need Bounding boxes, then LabelIMG is a great free tool for that. Incase you need someone to do the manual annotation I can help with that as well as that is my area of expertise.

Chirag Nagpal (chiragn@andrew.cmu.edu)

2023-03-17 14:22:39

Hello AI For Conservation 👋👋

I am Chirag Nagpal, a PhD candidate at CMU with an interest in Time-to-Event and Survival Analysis problems could be of immediate interest to wildlife conservation.

My methodological research has been applied to a large number of clinical applications and I also maintain a repository of reusable python tools for time--to-event and survival regression,
https://autonlab.org/auton-survival/

I am interested in participating and seeking potential collaborations in this space !

☺️ Sara Beery, Lucia Gordon, Jason Holmberg (Wild Me), Katelyn Morrison

👋 Mark Goldwater, Jason Holmberg (Wild Me), Katelyn Morrison

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-03-18 12:31:53

Hello! 👋
St. Patrick's Day greetings to you all☘️☘️

My name is Aakash Gupta, I am a climate entrepreneur and an AI4Good advocate.
My team has developed an AI-enabled platform for bio-diversity estimation of the wild population. The platform is being deployed in two wildlife sanctuaries in India; viz. Kawal Tiger Reserve and Amrabad Tiger sanctuary.
The system has processed more than 3.85mn camera trap events spread over 4 census years

This project has given me the opportunity to interact with a number of decision-makers and wild-life conservation experts in India and abroad. And their vision and passion for wildlife conservation and adopting new technology has inspired me. Look forward to meeting more like-minded researchers, entrepreneurs, and enthusiasts.

https://www.linkedin.com/posts/telangana-ai-mission_ai-activity-7041721797955788801-CFoS?utm_source=share&utm_medium=member_android

linkedin.com

Telangana AI Mission (T-AIM) on LinkedIn: #ai

🌳 In August 2022, we concluded the Forest AI Grand Challenge (<http://taim-gc.in/forest>) in association with Capgemini. The challenge saw interest from ~60…

Original URL: https://www.linkedin.com/posts/telangana-ai-mission_ai-activity-7041721797955788801-CFoS?utm_source=share&utm_medium=member_android

🙌 Josh Seltzer, Lucia Gordon, Eddie Zhang, Sara Beery, Prabath Gunawardane, Felipe Montealegre-Mora, Jason Holmberg (Wild Me), Akshay Paruchuri

🎉 Jon Van Oast, Rebecca Wilks, Aditee Kumthekar, Jason Holmberg (Wild Me)

Ameya Patil (ameyapatil249@gmail.com)

2023-03-23 03:29:36

A GSoC equivalent for Earth sciences, it is only open for EU residents though
https://climate.copernicus.eu/ecmwfs-code-earth-2023#:~:text=Code%20for%20Earth%20is%20an,and%20Destination%20Earth%20(DestinE).

🌏 Rebecca Wilks

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2023-03-23 11:02:19

Hello everyone 👋

I am Christoph Praschl, and I am an Assistant Professor for Computer Vision and Software Engineering at the University of Applied Sciences Upper Austria. Within my research I am mostly focusing on computer vision methodologies in the context of conservation. Currently, I am working on the drone based detection of animals in forests using airborne light field samples (http://www.bambi.eco/; unfortunately the page is currently available in german, but when there is time an english translation will follow ^^).

If you have any questions or see any potential for collaborations don't hesitate to contact me 😊

👋 Suzanne Stathatos, Majid Mirmehdi, Dan Morris, Eric Price, Adam Noach

Chris Lang (chrislang@ucsb.edu)

2023-03-23 12:46:32

Hi everyone!

My name is Chris Lang, I am a software engineer at the Benioff Ocean Science Laboratory working on a handful of applied marine science projects including some using machine learning (vision).

We're hiring a Software Data Engineer to support the Clean Currents Coalition and other marine technology projects at the Benioff Ocean Science Laboratory at UC Santa Barbara. Learn more about this opportunity or share it with your network here [https://recruit.ap.ucsb.edu/JPF02400]. The estimated salary range for this position is $90,000 up to $120,000. Applications are due April 7th, 2023. UCSB is an AA/EOE, including disability/vets.

recruit.ap.ucsb.edu

Assistant or Associate Specialist (Software Engineer), Benioff Ocean Science Laboratory, Marine Science Institute, UC Santa Barbara

University of California Santa Barbara is hiring. Apply now!

Apply by

Apr 7, 2023

Department

Marine Science Institute - Office of Research

Original URL: https://recruit.ap.ucsb.edu/JPF02400

🐳 Suzanne Stathatos, Sara Beery, Jaroslav Bezdek, Toryn Schafer, Yseult Hb, Felipe Montealegre-Mora, Chris Lang

🎉 Jon Van Oast, Declan, Dan Morris, Sara Beery, Chris Lang

Chris Lang (chrislang@ucsb.edu)

2023-03-23 14:03:08

The website incorrectly lists the anticipated start date as February. It will now be late May-June but will still be negotiable.

Steve Haddock (haddock@mbari.org)

2023-03-23 17:57:08

set the channel topic: Contact info to join: aiforconservation@gmail.com

Steve Haddock (haddock@mbari.org)

2023-03-23 17:59:50

👍 Sara Beery

✔️ Jon Van Oast

Akash Jaiswal (akash10987@gmail.com)

2023-03-24 08:00:33

Hi. Does anybody know about any opensource unsupervised clustering algorithm to group sounds of similar pattern together for further manual annotation?

Ben Weinstein (benweinstein2010@gmail.com)

2023-03-24 12:26:15

We were just talking about this yesterday to add to a grant proposal, post if you find something. (@Sara Beery). Can you describe what you hope to do with it, just for me to have a sense of workflow.

👍 Akash Jaiswal

Sara Beery (sara.beery@gmail.com)

2023-03-24 14:50:34

I know Tom Denton at Google has been doing some flavors of this (with query-based structure) but I don't know that they have anything open source.

👍 Akash Jaiswal

Sara Beery (sara.beery@gmail.com)

2023-03-24 14:50:41

@Holger Klinck, do you and the BirdNet team do anything along these lines?

Holger Klinck (hk829@cornell.edu)

2023-03-24 15:20:47

Yeah, a little with Tom. But nothing open source yet - first we need to generate some reasonable results :)

👍 Sara Beery

Matt Weldy (matthewjweldy@gmail.com)

2023-03-24 16:28:53

I've been using some distance based queries lately that are working fairly well. Feel free to message me and I will share some of the initial stuff I have running. There is also the ScaNN and Milvus libraries.

👍 Sara Beery, Akash Jaiswal

Akash Jaiswal (akash10987@gmail.com)

2023-03-25 03:12:13

@Ben Weinstein Currently, I am dealing with bird species identification in a subset of my recordings for a paper. And it's overwhelming (particularly for a place like Delhi which has so many bird species and lots of them have very large repertoire). I am also planning to do a fine scale annotation in those subset for some ML based work. I know Kaleidoscope (Wildlife Acoustics) has such clustering function. But don't know about any open-source for such utility. So, for my use, it can help with two things; 1) to estimate a direct count of sonotype richness in a sound file and 2) fine-scale manual annotations of those sounds pattern which can be further used for developing a deep learning model for species identification.

Vincent Christlein (vincent.christlein@fau.de)

2023-03-27 05:56:08

Maybe some of the works of Christian Bergler (e.g. Animal-spot) may be worth having a look at: https://github.com/ChristianBergler

Repositories

Followers

👍 Akash Jaiswal, Matt Weldy

Akash Jaiswal (akash10987@gmail.com)

2023-03-27 08:25:01

@Vincent Christlein Thanks for sharing.

Vincent Christlein (vincent.christlein@fau.de)

2023-03-27 10:00:11

you're welcome

Ben Williams (ben.williams.20@ucl.ac.uk)

2023-03-27 13:55:32

Interested to follow this. In the past for similar problems I've used something like YAMNet or PANNS for feature extraction, UMAP to reduce dimensions, then clustering with affinity propogation where you don't need to designate the number of clusters, in theory the clusters should group similar sounds. Sure there are more up to date/bespoke approaches like those shared above!

👍 Akash Jaiswal

Prach Sri (prach@todreamalife.com)

2023-03-24 16:16:18

Does anyone have experience in carbon economics.?

I’m totally stumped on a carbon calculation using allometric model for AGB from DBH. The Colombian national standard isn’t working and I don’t see why.

Thanks in advance!

Sara Beery (sara.beery@gmail.com)

2023-03-24 16:37:37

@David?

👍 David, Prach Sri

David (dwddao@gmail.com)

2023-03-24 22:54:06

What allometric equations are you using? Is the intention to calculate AGB for a forest carbon credit?

Prach Sri (prach@todreamalife.com)

2023-03-27 15:14:24

We did a carbon study to compare with FRELs. I’m using Chave Model Type II - Tm

ln(AGB) = a + b1ln( D )+ b2(ln( D ))2+ b3(ln( D ))3+ d ln(q)

Diameter (D) in cm and density (q) in g/cm3. The parameters for a and b are also provided in the attached paper.

Happy to review it anytime!

David (dwddao@gmail.com)

2023-03-28 12:28:09

Do you know how the FREL was calculated in Colombia, most likely through interpolations from satellites I almost assume? Did they use Chave et al (2005)? Maybe there are factors they integrated such as additionality etc into their net emissions? We have consistently met discrepancies with reported government data when using our own AI models. Many reasons are political.

Prach Sri (prach@todreamalife.com)

2023-03-29 01:29:38

The FREL estimates site Alvarez 2012. It appears they are using field sampling rather than satellite data. I’ve run the models for both Chave 2005 and Alvarez 2012 (using the provided best fit parameters) and get vastly different outputs (from each other, and the national average). I’m starting to feel the models don’t hold up…

David (dwddao@gmail.com)

2023-03-29 05:35:58

Regarding AGB you can do following satellite-based sanity check and compare the aggregared AGB from Global Forest Watch: https://data.globalforestwatch.org/maps/e4bdbe8d6d8d4e32ace7d36a4aec7b93
Spawn, 2020: https://www.nature.com/articles/s41597-020-0444-4
Santoro, 2021: https://essd.copernicus.org/articles/13/3927/2021/essd-13-3927-2021-discussion.html

Their data is available publicly and globally and from my experience, for areas > 5ha, Santoro has been surprisingly close to our field-based experiments in Ecuador: https://arxiv.org/abs/2201.11192

This could help you distill a better signal maybe!

data.globalforestwatch.org

Global Forest Watch Open Data Portal

Global Forest Watch Open Data Site

Original URL: https://data.globalforestwatch.org/maps/e4bdbe8d6d8d4e32ace7d36a4aec7b93

Nature

Harmonized global maps of above and belowground biomass carbon density in the year 2010

Scientific Data - Harmonized global maps of above and belowground biomass carbon density in the year 2010

Original URL: https://www.nature.com/articles/s41597-020-0444-4

The global forest above-ground biomass pool for 2010 estimated from high-resolution satellite observations

Abstract. The terrestrial forest carbon pool is poorly quantified, in particular in regions with low forest inventory capacity. By combining multiple satellite observations of synthetic aperture radar (SAR) backscatter around the year 2010, we generated a global, spatially explicit dataset of above-ground live biomass (AGB; dry mass) stored in forests with a spatial resolution of 1 ha. Using an extensive database of 110 897 AGB measurements from field inventory plots, we show that the spatial patterns and magnitude of AGB are well captured in our map with the exception of regional uncertainties in high-carbon-stock forests with AGB >250 Mg ha−1, where the retrieval was effectively based on a single radar observation. With a total global AGB of 522 Pg, our estimate of the terrestrial biomass pool in forests is lower than most estimates published in the literature (426–571 Pg). Nonetheless, our dataset increases knowledge on the spatial distribution of AGB compared to the Global Forest Resources Assessment (FRA) by the Food and Agriculture Organization (FAO) and highlights the impact of a country's national inventory capacity on the accuracy of the biomass statistics reported to the FRA. We also reassessed previous remote sensing AGB maps and identified major biases compared to inventory data, up to 120 % of the inventory value in dry tropical forests, in the subtropics and temperate zone. Because of the high level of detail and the overall reliability of the AGB spatial patterns, our global dataset of AGB is likely to have significant impacts on climate, carbon, and socio-economic modelling schemes and provides a crucial baseline in future carbon stock change estimates. The dataset is available at <https://doi.org/10.1594/PANGAEA.894711> (Santoro, 2018).

Original URL: https://essd.copernicus.org/articles/13/3927/2021/essd-13-3927-2021-discussion.html

arXiv.org

ReforesTree: A Dataset for Estimating Tropical Forest Carbon Stock with Deep Learning and Aerial Imagery

Forest biomass is a key influence for future climate, and the world urgently needs highly scalable financing schemes, such as carbon offsetting certifications, to protect and restore forests. Current manual forest carbon stock inventory methods of measuring single trees by hand are time, labour, and cost-intensive and have been shown to be subjective. They can lead to substantial overestimation of the carbon stock and ultimately distrust in forest financing. The potential for impact and scale of leveraging advancements in machine learning and remote sensing technologies is promising but needs to be of high quality in order to replace the current forest stock protocols for certifications. In this paper, we present ReforesTree, a benchmark dataset of forest carbon stock in six agro-forestry carbon offsetting sites in Ecuador. Furthermore, we show that a deep learning-based end-to-end model using individual tree detection from low cost RGB-only drone imagery is accurately estimating forest carbon stock within official carbon offsetting certification standards. Additionally, our baseline CNN model outperforms state-of-the-art satellite-based forest biomass and carbon stock estimates for this type of small-scale, tropical agro-forestry sites. We present this dataset to encourage machine learning research in this area to increase accountability and transparency of monitoring, verification and reporting (MVR) in carbon offsetting projects, as well as scaling global reforestation financing through accurate remote sensing.

Original URL: https://arxiv.org/abs/2201.11192

🙏 Prach Sri

Rio Akbar (riosyahakbar36@gmail.com)

2023-03-27 23:47:44

Hey Guys! I have a small dataset of 50 images of feral cats. The images are grouped into the area of the body the images focuses on (eg. Right hind leg etc.) and thus has the pattern markings for that area. I’m doing a small project on individual identification of these small cats but my supervisor has advised to not use object detection models like ResNet or YOLOv5. Instead focus on identification. Does anyone have any suggestions for this task? I’ve had a look at I3S, not sure if its suitable or not. Also not sure of any other alternatives. If anyone could help, that would be great! 🙂

Vincent Christlein (vincent.christlein@fau.de)

2023-03-28 02:55:17

I'd start fine-tuning a pre-trained model on your classes. With 50 images, you should augment a lot otherwise, you will overfit quite quickly

Rio Akbar (riosyahakbar36@gmail.com)

2023-03-28 02:58:05

Thanks Vincent, do you have a suggestion for a pre-trained model I can use? What do you think about I3S?

Vincent Christlein (vincent.christlein@fau.de)

2023-03-28 03:00:53

what are you referring to? https://github.com/daniel-brenot/I3S-Interactive-Individual-Identification-System-Desktop ? I never worked with that

daniel-brenot/I3S-Interactive-Individual-Identification-System-Desktop

A photo recognition software used to detect distinct features in a individual animal for use in conservation

Language

C++

Last updated

5 years ago

Sara Beery (sara.beery@gmail.com)

2023-03-28 11:14:52

@Jason Parham

Jacob Ayers (jgayers@ucsd.edu)

2023-03-29 00:44:31

How distinct are the markings between the individual cats? If you could quickly tell with the human eye, maybe something simple such as Hu Moments or basic PCA classification could be worth looking into.

Jason Parham (bluemellophone@gmail.com)

2023-03-31 15:41:37

For now, I would suggest looking at IBEIS https://github.com/Erotemic/ibeis

Erotemic/ibeis

image based ecological information system

Stars

Language

Python

Rio Akbar (riosyahakbar36@gmail.com)

2023-04-01 22:52:52

Thanks @Jason Parham !

Dan Morris (agentmorris@gmail.com)

2023-03-28 15:19:44

Also in doing that search I was surprised by how many unannotated data sets are out there, just orthomosaics that someone created in the process of doing a manual count survey, typically from USGS or US/state F&W groups. I did not include those, but those would be fun to index as well.

👍 Ben Weinstein, Sara Beery, Rebecca Wilks, Edward Bayes

Ronny Hänsch (rww.haensch@gmail.com)

2023-03-28 15:34:42

maybe a bit of a stretch but I would count drone/aerial images for wildlife detection as EO data. So one could add it to the GRSS EOD

Sara Beery (sara.beery@gmail.com)

2023-03-28 17:25:30

I'd say it's helpful to have a separate list for this, since it's a bit specific. Thanks Dan!

Rowan Converse (rowanconverse@unm.edu)

2023-03-29 15:43:48

This is great-- thanks for putting this together! I may have some more additions after completing the survey of the RS/ML Community of Practice group.

Dan Morris (agentmorris@gmail.com)

2023-03-29 16:23:45

Now I'm getting greedy, and I'd like to not only have a list, I'd like to get a slightly structured record for each dataset, with at least species, resolution, number and type of annotations, etc., plus a couple sample images, and - here's the stretch goal - a couple images where we've rendered an annotation onto an image (to make sure it's actually possible). It's fine if this is literally done manually (e.g. by opening the annotations in a text editor, and drawing a box on the corresponding image in Photoshop), but we can also live without this last step.

This is a bit of a pain, but if we were to divide and conquer, we could do this pretty quickly (like <1 hour), so... reply here if you're willing to sign up to do this for either a specific dataset on the list, or for N datasets that I randomly assign you.

If you're the owner of a dataset and you already have the answer, you still get all the Karma if you volunteer to provide this data for your own dataset!

If I get some takers, I'll send out a link to a spreadsheet.

🙋 Kalindi Fonda, Edward Bayes, Aakash Gupta, Josh Veitch-Michaelis

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-03-30 00:36:10

I'd like to volunteer for this. Let me know what needs to be done.

👍 Dan Morris

Zhongqi Miao (zhongqi.miao@berkeley.edu)

2023-03-30 14:37:27

I would like to volunteer as well for one or two datasets.

👍 Dan Morris

Dan Morris (agentmorris@gmail.com)

2023-03-30 18:37:17

Oh this is great, that's 5 volunteers, I'm almost off the hook for doing anything now. 🙂 I'll DM all of you if I don't have your email address, and send "assignments" by email. Thanks!

🌟 Kalindi Fonda

Dan Morris (agentmorris@gmail.com)

2023-04-12 12:51:03

OK, mission accomplished, we huddled virtually offline and gathered standardized metadata, standardized sample code, and sample annotated images for all the datasets on that list:

https://github.com/agentmorris/agentmorrispublic/blob/main/drone-datasets.md

Thanks to @Zhongqi Miao, @Aakash Gupta, @Edward Bayes, @Kalindi Fonda, and @Josh Veitch-Michaelis for participating in this experiment.

Let us know what the list is missing! And if you're the owner of any of the datasets (I'm looking at you, @Ben Weinstein), let us know if we got anything wrong.

<https://github.com/agentmorris/agentmorrispublic/blob/main/drone-datasets.md | drone-datasets.md>

Datasets with annotated wildlife in drone/aerial images Overview This is list of datasets with annotated aerial/drone/satellite imagery for wildlife surveys. We've tried to collected standardized metadata for each dataset, and to provide sample code in a similar format for each dataset that can match annotations to images and render a sample image. The goal of this exercise was to make it easier to assess whther existing datasets could be useful for training new models, e.g. if you want to find birds in images with a particular gestalt, we wanted to make it easier for you to get the gestalt of existing datasets, and have a starting point for parsing that data. Everything listed here is also either listed on LILA's <a href="https://lila.science/otherdatasets">list of other conservation datasets</a> or is on <a href="https://lila.science">LILA</a>. This list was curated by the following folks, after this list got started as a thread on the AI for Conservation Slack: • <a href="https://dmorris.net">Dan Morris</a> • <a href="https://www.linkedin.com/in/zhongqi-miao-17a50084/">Zhongqi Miao</a> • <a href="https://www.linkedin.com/in/kalindifonda">Kalindi Fonda</a> • <a href="https://www.linkedin.com/in/aakash-gupta-5ky">Aakash Gupta</a> • <a href="https://www.linkedin.com/in/jveitchmichaelis/">Josh Veitch-Michaelis</a> • <a href="https://www.linkedin.com/in/edbayes/">Ed Bayes</a> Email <mailto:agentmorris+dronedatasets@gmail.com|Dan> if anything seems off, or if you know of datasets we're missing. Data sets Improving the precision and accuracy of animal population estimates with aerial image object detection Aerial images with 4305 bounding boxes on zebra, giraffe, and elephants Eikelboom JA, Wind J, van de Ven E, Kenana LM, Schroder B, de Knegt HJ, van Langevelde F, Prins HH. Improving the precision and accuracy of animal population estimates with aerial image object detection. Methods in Ecology and Evolution. 2019 Nov;10(11):1875-87. • 5.7 GB, downloadable via http from 4TU (<a href="https://data.4tu.nl/articles/dataset/Improving_the_precision_and_accuracy_of_animal_population_estimates_with_aerial_image_object_detection/12713903/1">download link</a>) • Metadata in csv format • Categories: elephant, zebra, giraffe • Vehicle type: plane • Image information: 561 RGB images • Annotation information: 4305 boxes • Typical animal size in pixels: 50 • Code to render sample annotated image: <a href="https://github.com/agentmorris/agentmorrispublic/blob/main/aerial-drone-data-preview/preview-eikelboom-savanna.py">preview-eikelboom-savanna.py</a> <a href="https://raw.githubusercontent.com/agentmorris/agentmorrispublic/main/aerial-drone-data-preview/eikelboom_savanna_sample_image_annotated.jpg">https://raw.githubusercontent.com/agentmorris/agentmorrispublic/main/aerial-drone-data-preview/eikelboomsavannasampleimageannotated.jpg</a> UAV-derived waterfowl thermal imagery dataset 8976 bounding boxes on waterfowl in UAV-derived thermal images Hu Q, Smith J, Woldt W, Tang Z. UAV-derived waterfowl thermal imagery dataset. Mendeley Data, V4. 2021. • 4.1 GB, downloadable via http from Mendeley Data (<a href="https://data.mendeley.com/datasets/46k66mz9sz/4">download link</a>) • Metadata in csv format • Categories: waterfowl • Vehicle type: plane • Image information: 541 thermal images (unannotated RGB images included as a visual reference) • Annotation information: 8976 boxes • Typical animal size in pixels: 7 • Code to render sample annotated image: <a href="https://github.com/agentmorris/agentmorrispublic/blob/main/aerial-drone-data-preview/preview-hu-thermal.py">preview-hu-thermal.py</a> <a href="https://raw.githubusercontent.com/agentmorris/agentmorrispublic/main/aerial-drone-data-preview/hu_thermal_sample_image_annotated.jpg">https://raw.githubusercontent.com/agentmorris/agentmorrispublic/main/aerial-drone-data-preview/huthermalsampleimageannotated.jpg</a> Drones count wildlife more accurately and precisely than humans Images of 10 life-sized, replica seabird colonies containing a known number of fake birds taken from four different heights (30m, 60m, 90m and 120m) Hodgson JC, Mott R, Baylis SM, Pham TT, Wotherspoon S, Kilpatrick AD, Raja Segaran R, Reid I, Terauds A, Koh LP. Drones count wildlife more accurately and precisely than humans. Methods in Ecology and Evolution. 2018 May;9(5):1160-7. • 50 MB, downloadable via http from Dryad (<a href="https://datadryad.org/stash/dataset/doi:10.5061/dryad.rd736">download link</a>) • Metadata in csv format • Categories: fake seabirds • Vehicle type: drone • Image information: 40 RGB images • Annotation information: 1560 counts • Typical animal size in pixels: variable <a href="https://raw.githubusercontent.com/KalindiFonda/bAIo/main/aerial-drone-data-preview/drone_count_seabird_colony_60m_sample.jpg">https://raw.githubusercontent.com/KalindiFonda/bAIo/main/aerial-drone-data-preview/dronecountseabirdcolony60m_sample.jpg</a> Counting animals in aerial images with a density map estimation model 137365 point annotations on penguins in RGB UAV images Qian Y, Humphries G, Trathan P, Lowther A, Donovan C. Counting animals in aerial images with a density map estimation model [Data set]. 2023. • 300 MB, downloadable via http from Zenodo (<a href="https://zenodo.org/record/7702635#.ZChnoHZBxD8">download link</a>) • Metadata in json format (LabelBox standard) • Categories: brush-tailed penguins • Vehicle type: plane • Image information: 753 RGB images (orthorectified) • Annotation information: 137365 points • Typical animal size in pixels: 30 • Code to render sample annotated image: <a href="https://github.com/agentmorris/agentmorrispublic/blob/main/aerial-drone-data-preview/preview-qian-penguins.py">preview-qian-penguins.py</a> <a href="https://raw.githubusercontent.com/agentmorris/agentmorrispublic/main/aerial-drone-data-preview/qian_penguins_sample_image_annotated.jpg">https://raw.githubusercontent.com/agentmorris/agentmorrispublic/main/aerial-drone-data-preview/qianpenguinssampleimageannotated.jpg</a> Data from: A convolutional neural network for detecting sea turtles in drone imagery 1902 point annotations on sea turtles in drone images Gray PC, Fleishman AB, Klein DJ, McKown MW, Bezy VS, Lohmann KJ, Johnston DW. A convolutional neural network for detecting sea turtles in drone imagery. Methods in Ecology and Evolution. 2019 Mar;10(3):345-55. • 7.24 GB, downloadable via http from Zenodo (<a href="https://zenodo.org/record/5004596#.ZChnr3ZBxD8">download link</a>) • Metadata in csv format • Categories: olive ridley turtle • Vehicle type: drone • Image information: 1059 NIR images (false color NIR rendered to RGB) • Annotation information: 1902 points • Typical animal size in pixels: 10 • Code to render sample annotated image: <a href="https://github.com/agentmorris/agentmorrispublic/blob/main/aerial-drone-data-preview/preview-gray-turtles.py">preview-gray-turtles.py</a> <a href="https://raw.githubusercontent.com/agentmorris/agentmorrispublic/main/aerial-drone-data-preview/gray_turtles_sample_image_annotated.jpg">https://raw.githubusercontent.com/agentmorris/agentmorrispublic/main/aerial-drone-data-preview/grayturtlessampleimageannotated.jpg</a> The Aerial Elephant Dataset 15581 point annotations on elephants in aerial images Naude J, Joubert D. The Aerial Elephant Dataset: A New Public Benchmark for Aerial Object Detection. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops 2019 (pp. 48-55). • 16.3 GB, downloadable via http from Zenodo (<a href="https://zenodo.org/record/3234780">download link</a>) • Metadata in csv format • Categories: elephant • Vehicle type: drone • Image information: 2074 RGB images • Annotation information: 15581 points • Typical animal size in pixels: 75 • Code to render sample annotated image: <a href="https://github.com/agentmorris/agentmorrispublic/blob/main/aerial-drone-data-preview/preview-aerial-elephants.py">preview-aerial-elephants.py</a> <https://raw.githubusercontent.com/agentmorris/agentmorrispublic/main/aerial-drone-data-preview/aerialelephantssampleimageannotated.jpg|https://raw.githubusercontent.com/agentmorris/agentmorrispublic/main/aerial-drone-data-preview/…

🎉 Edward Bayes, Kalindi Fonda, Rowan Converse

💯 Aakash Gupta

🚀 Carl Boettiger

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2023-04-25 16:12:44

We are planing to also open source our data from our research project BAMBI in future including drone videos of wildlife such as deer, red deer, boar, chamois, … But this will probably still take some months. I will definetley come back to you regarding that, if you are interested 🙂

🦌 Kalindi Fonda, Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2023-04-25 20:44:27

Definitely, we will want to integrate that in our airborne work.

👍 Christoph Praschl

Thijs (thijs@q42.nl)

2023-03-30 08:16:19

I just now got back from a project in Zambia to test a system to mitigate human-elephant conflicts. The rollout was successful and we already saw it in action when we were there. You can get an impression here: https://www.linkedin.com/posts/tsuijten_techforgood-wildlifeconservation-conservation[…]106161602560-OBhv?utm_source=share&utm_medium=member_desktop

linkedin.com

Thijs Suijten on LinkedIn: #techforgood #wildlifeconservation #conservation #ai #machinelearning

𝑪𝒖𝒕𝒆𝒏𝒆𝒔𝒔 𝒂𝒍𝒆𝒓𝒕! 𝑩𝒂𝒃𝒚 𝒆𝒍𝒆𝒑𝒉𝒂𝒏𝒕𝒔 𝒊𝒏𝒄𝒐𝒎𝒊𝒏𝒈! 🐘 😍 After our project in Zimbabwe we went to Zambia: mitigate human-elephant…

Original URL: https://www.linkedin.com/posts/tsuijten_techforgood-wildlifeconservation-conservation-activity-7047177106161602560-OBhv?utm_source=share&utm_medium=member_desktop

Thijs (thijs@q42.nl)

2023-03-30 08:16:53

I really curious to get feedback from people who also have a lot of experience working with elephants.

Thijs (thijs@q42.nl)

2023-03-30 08:18:52

Currently we are using our AI-camera to detect the presence of elephants. But I would be really interested in other ways to detect elephants. For instance using audio: detecting their sounds / rumbles. Are there any experts here in that field? Detecting elephants by sound? 🎤 🐘

🐘 Suzanne Stathatos, Dan Morris, Sara Beery

😎 Jon Van Oast

Andrew Schulz (akschulz@gatech.edu)

2023-03-30 08:33:34

there is quite a lot of work on different frequencies of calls and vocalizations of elephants. I can reach out and see if they have done detection.

Patryk Neubauer (patryk.neubauer@gmail.com)

2023-03-30 08:39:36

You might want to check this out: https://elephantlisteningproject.org/

Elephant Listening Project

Elephant Listening Project Home

Explore With Us In Focus

Original URL: https://elephantlisteningproject.org/

:thumbsup_all: Andrew Schulz, Thijs, Clare Price

Thijs (thijs@q42.nl)

2023-03-30 08:48:44

Thanks @Patryk Neubauer so you happen to know anyone there?

Patryk Neubauer (patryk.neubauer@gmail.com)

2023-03-30 08:53:25

Know is perhaps a big word 😄 , but Daniela Hedwig from ELP is a mentor on a project I'm contributing to

👍 Thijs

Clare Price (theclareprice@gmail.com)

2023-03-30 19:33:19

Joyce Poole with Elephant Voices developed the Elephant Ethogram and has worked extensively on elephant acoustics!

elephantvoices.org

Welcome to ElephantVoices

ElephantVoices' Mission is to inspire wonder in the intelligence, complexity and voices of elephants and to secure a kinder future for them.

Original URL: https://www.elephantvoices.org/

elephantvoices.org

The Elephant Ethogram

Introduces the elephant ethogram: A Library of African Elephant Behavior

Original URL: https://www.elephantvoices.org/elephant-ethogram.html

:squirrel: Andrew Schulz

👍 Thijs

Michael Bunsen (notbot@gmail.com)

2023-03-30 20:09:11

Is anyone familiar with any open source mobile apps for collecting data in the field? Something akin to Avenza, or OnX but much simpler that we could modify for a specific field study. Basically looking for the ability to record locations offline, take photos, tag observations and add notes. It doesn't even need a map.

Jon Van Oast (jon@wildme.org)

2023-03-30 20:17:56

i have only played with this a little bit (personally, not professionally) but it seems to have a decent user base and track record.

https://getodk.org/

getodk.org

ODK - Collect data anywhere

Powerful offline forms to collect the data you need wherever it is. Used by leading researchers, field teams, and M&E professionals to build a better world.

Original URL: https://getodk.org/

Michael Bunsen (notbot@gmail.com)

2023-03-30 20:28:24

Hey that looks promising thanks!

Michael Bunsen (notbot@gmail.com)

2023-03-30 20:31:56

Also iNaturalist's upcoming Seek v2 app that uses React Native could be a candidate to modify

👍 Jon Van Oast, Jose Ruiz-Munoz

Catherine Villeneuve (catherine.villeneuve.9@ulaval.ca)

2023-03-31 09:34:02

The mobile version of AI2's EarthRanger (https://www.earthranger.com) can do all of this. In offline mode, it will record your location/photos/observations/notes in a local cache and once you have access to an internet connection, the app will transmit your data to a main server. It's easy to retrieve the data afterwards through their Python API, and the app is highly customizable. We're currently using ER Mobile for field census in a 100% offline setting in the Canadian High-Arctic (Nunavut). If you are interested, let me know and I can connect you the right people/share how we've adapt our EarthRanger server to our needs

earthranger.com

EarthRanger: Protecting Wildlife With Real-Time Data

EarthRanger is a software solution that aids protected area managers, ecologists, and wildlife biologists in making more informed operational decisions for wildlife conservation.

Original URL: https://www.earthranger.com

Victor Anton (victor@wildlife.ai)

2023-03-31 11:27:41

The CitSci mobile app (https://citsci.org/apps) might be worth having a look at? I can't find "how much" open source is though...

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-03-31 12:17:19

Beat officers in India use NoteCam to capture the lat/lon locations during patrolling activity. As well as to take images of waterholes, pugmarks, PIDs etc. you can attach a note to that image which can then be exported as a KML file. But this is a freemium tool.

Ephantus Kanyugi (ndungu.kanyugi@gmail.com)

2023-03-31 15:51:47

Hello, my name is Ephantus Kanyugi. I am a data management and annotation expert. In the past years I have been involved in several scientific projects on facial and body language analysis of animals, where I have been in charge of complex data management pipelines. I am also very passionate about conservation of wildlife in my home country Kenya, and have a network of connections to conservation organizations in some nature reserves. I was excited to find out about this group, where I can combine my professional interest with my passion for wilflife conservation. I would like to network, and offer my services in all tasks related to data annotation, labeling, analysis, etc. Here's a link to my LinkedIn: https://www.linkedin.com/in/ephantus-kanyugi-9a04b7137/

😎 Jason Holmberg (Wild Me), Ephantus Kanyugi, Felipe Parodi, Sara Beery, Yseult Hb

🎉 Jason Holmberg (Wild Me), Pen-Yuan Hsing, Stephanie O'Donnell

👏 Jon Van Oast, Anna Zamansky, Dan Morris, Josh Seltzer, Thomas Radinger

👋 Carly Batist, Mark Goldwater, Viktor Domazetoski, Declan

Ed Miller (ed@hypraptive.com)

2023-04-06 13:57:38

Join me tomorrow for an #AWS Heroes in 15 session on Bear Conservation with ML, Serverless and Citizen Science.: https://www.linkedin.com/video/event/urn:li:ugcPost:7038608935079043073/

linkedin.com

500 million+ members | Manage your professional identity. Build and engage with your professional network. Access knowledge, insights and opportunities.

Original URL: https://www.linkedin.com/feed/hashtag/?keywords=aws&highlightedUpdateUrns=urn%3Ali%3Aactivity%3A7049801509106040833

linkedin.com

500 million+ members | Manage your professional identity. Build and engage with your professional network. Access knowledge, insights and opportunities.

Original URL: https://www.linkedin.com/video/event/urn:li:ugcPost:7038608935079043073/

🌟 Talia Speaker, Jason Holmberg (Wild Me), Aakash Gupta

👍 Alayna Van Dervort

Ed Miller (ed@hypraptive.com)

2023-04-06 13:59:47

There's also an interview by AWS Developer Advocate, Linda Haviv, from AWS re:Invent 2022: https://youtu.be/sKGau7c53go

YouTube

Build On AWS (https://www.youtube.com/@BuildOnAWS)

Using Machine Learning to Build Bear Identification Technology | Ed Miller

Original URL: https://youtu.be/sKGau7c53go

❤️ Talia Speaker, Aakash Gupta

Silvia Zuffi (silvia@mi.imati.cnr.it)

2023-04-07 06:13:15

Hello people, does anybody know of image datasets with segmented plants and their species labels? Thanks!

:squirrel: Andrew Schulz

Jose Ruiz-Munoz (jfruizmu@unal.edu.co)

2023-04-07 12:05:06

Maybe it is not exactly what you are looking for but take a look at https://www.kaggle.com/datasets/emmarex/plantdisease

kaggle.com

PlantVillage Dataset

Dataset of diseased plant leaf images and corresponding labels

Original URL: https://www.kaggle.com/datasets/emmarex/plantdisease

Jon Van Oast (jon@wildme.org)

2023-04-07 12:20:49

plantnet is a great app/group. maybe this is useful?
https://github.com/plantnet/PlantNet-300K

plantnet/PlantNet-300K

[NeurIPS2021] A plant image dataset with high label ambiguity and a long-tailed distribution

Website

<https://doi.org/10.5281/zenodo.5645731>

Stars

Dan Morris (agentmorris@gmail.com)

2023-04-07 15:04:56

Segmented trees:
https://github.com/norlab-ulaval/PercepTreeV1

norlab-ulaval/PercepTreeV1

Implementation of Grondin et al. 2022 "Tree Detection and Diameter Estimation Based on Deep Learning". Also includes datasets and some of the pretrained models.

Stars

Language

Python

Dan Morris (agentmorris@gmail.com)

2023-04-07 15:09:16

More generally, most of these don't have segmentation labels, but... a list of image datasets that have something to do with plants:

https://lila.science/otherdatasets#images-plants

LILA BC

SciTePress - Publication Details

LILA BC - List of other conservation-related data sets

Written by

lilawp

Est. reading time

20 minutes

Original URL: https://lila.science/otherdatasets#images-plants

Silvia Zuffi (silvia@mi.imati.cnr.it)

2023-04-08 07:09:15

Thank you all for the pointers!

👍 Jon Van Oast

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2023-04-25 16:17:27

Hey Silvia, we have collected quite a huge dataset containing alpine plants from the Nationalpark Hohe Tauern (Austria). It is currently a closed dataset, but our project partner was interested in open source it. I could make contact, if you are interested in that :)

Some side information on the dataset can be found in our publications: https://doi.org/10.5220/0011607100003417 and https://doi.org/10.46354/i3m.2022.sesde.006

scitepress.org

Digital Library

Original URL: https://doi.org/10.5220/0011607100003417

cal-tek.eu

Visual Change Detection in Multi-Temporal Vegetation Transects of Alpine Plants

Due to the apparent effects of climate change on the Earth’s ecosystems, it is more important than ever to monitor flora and fauna in affected regions, e.g. mountain areas above the tree line. In the alpine ecosystem, and not just there, Vegetation plays a fundamental role and is the subject of this study. The work aims to develop algorithms for recognising small stature alpine plants from close range top view images. Ideally, automated assessment algorithms of the plant cover should objectively help scientists observe and interpret the state of the plant ecosystem over a long time series. Therefore, the aim in this respect was to derive visualisations that accurately describe plant growth and displacement (translocation). Additionally, recording changes in biodiversity was an intent. This work uses multi-temporal data comprising RGB images and multi-label masks to accomplish the aforementioned task. The evaluated methods involve mask comparison, optical flow estimation, detection of individual plants, and descriptive statistical analysis of image feature properties. Tests on the given data set show that all methods but the optical flow estimation have great potential. The mask comparison method captured plant growth and translocation most satisfactory. Individual plant detection and statistical analysis further helped to evaluate changes in biodiversity. When combined, the proposed methods give an immediate overview about relevant changes in the multi-temporal transects, which has not been done before for close-distance images of alpine plants. Due to the apparent effects of climate change on the Earth’s ecosystems, it is more important than ever to monitor flora and fauna in affected regions, e.g. mountain areas above the tree line. In the alpine ecosystem, and not just there, Vegetation plays a fundamental role and is the subject of this study. The work aims to develop algorithms for recognising small stature alpine plants from close range top view images. Ideally, automated assessment algorithms of the plant cover should objectively help scientists observe and interpret the state of the plant ecosystem over a long time series. Therefore, the aim in this respect was to derive visualisations that accurately describe plant growth and displacement (translocation). Additionally, recording changes in biodiversity was an intent. This work uses multi-temporal data comprising RGB images and multi-label masks to accomplish the aforementioned task. The evaluated methods involve mask comparison, optical flow estimation, detection of individual plants, and descriptive statistical analysis of image feature properties. Tests on the given data set show that all methods but the optical flow estimation have great potential. The mask comparison method captured plant growth and translocation most satisfactory. Individual plant detection and statistical analysis further helped to evaluate changes in biodiversity. When combined, the proposed methods give an immediate overview about relevant changes in the multi-temporal transects, which has not been done before for close-distance images of alpine plants.

Original URL: https://doi.org/10.46354/i3m.2022.sesde.006

Silvia Zuffi (silvia@mi.imati.cnr.it)

2023-04-27 16:29:45

Thank you very much! Let me talk with my collaborators, I am not sure these alpine plants are the species we need to consider.

Alayna Van Dervort (av@thebigwild.com)

2023-04-12 15:08:43

HI Everyone, Is anyone familiar with bear Den detection other than infrared? also looking for large drones used in species detection

Dimitri Korsch (korschdima@gmail.com)

2023-04-17 16:16:07

Recently, I came across the BAMBI project. They aim at species monitoring using UAVs. Unfortunately, the website is in German because it is an Austrian project, but maybe you can contact the project leads.

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2023-04-25 15:08:40

Hi Alayna,

we are using DJI M30T drones for animal detection in our project BAMBI as mentioned by Dimitri Korsch. So we can definitely talk! :)

James Farrell (jamespfarrell@gmail.com)

2023-04-17 11:11:03

Hi everyone, I have just joined, I’m interested in scaling climate finance with AI.

I have some ideas, and researching plausibility thereof while trying to catch up on what everyone is working on :)

My background is software engineering, and I’ve been residing mostly in the public blockchain space the last couple of years, where I’m co-founded Toucan.earth and KlimaDAO.finance.

The AI side is relatively new for me, so I appreciate any recommendations or exciting projects that are doing good work in the space, or any AI / Machine learning people looking to collaborate.

👋 Jaroslav Bezdek, Dimitri Korsch, Aakash Gupta, Ephantus Kanyugi, Louis Moreau, Catherine

💯 Aakash Gupta, Ruiz Rivera

Ruiz Rivera (ruiz.rivera93@gmail.com)

2023-07-10 12:14:45

Hi @James Farrell, nice to see you on here I've also done some work for Klima and I'm familiar with the work you guys have done with Toucan in tokenizing carbon credits!

I was just curious what were the types of AI/Conservation projects you were hoping to undertake that applies to your work in ReFi? Maybe there's an opportunity for collaboration with myself and the other folks in this space 👍

Aidan Dunlop (aidan.dunlop@sky.uk)

2023-04-19 10:15:24

Hi everyone! and thanks @Stephanie O'Donnell for the invite :)
I’m a software engineer working on MLOps, and currently studying a part-time master’s in AI Ethics & Society. I’m interested in how we can use AI in a responsible and transparent way.

I have a small favour to ask. For my dissertation, I am looking into the challenges AI/ML practitioners face when using open-source tools for explainable machine learning.

I am looking for participants for a (max) 10-minute anonymous survey. Your help would be much appreciated!! You can access the survey here: https://cambridge.eu.qualtrics.com/jfe/form/SV_9AKEzvsIjIZ1LXU.

If you could share the survey with other AI/ML practitioners that would be very helpful :)

Thanks ☺️

cambridge.eu.qualtrics.com

Open-source tools for explainable machine learning: a survey of ML practitioners

This survey aims to collect empirical data on the challenges faced by machine learning practitioners when using open-source tools for explainable machine learning. The survey will contribute to a master’s dissertation for the AI Ethics & Society MSt, a part-time master's programme at the University of Cambridge. Please read this participant information sheet for more information. The survey should take around 10 minutes to complete. If you answer multiple times with the same email address, the last set of answers will be taken. If you have any questions, please email <a href="mailto:amsd3@cam.ac.uk">amsd3@cam.ac.uk</a>.

Original URL: https://cambridge.eu.qualtrics.com/jfe/form/SV_9AKEzvsIjIZ1LXU

🤩 Carly Batist, Sara Beery

👋 Sara Beery, Ephantus Kanyugi

❤️ Stephanie O'Donnell, Talia Speaker, Kristina Kupferschmidt

Maciej Adamiak (adamiak.maciek@gmail.com)

2023-04-19 10:36:34

One of my colleagues made an extensive research on XAI libraries. You can contact him. Maybe he could also help you in your task. Here is his blogpost with all his findings: https://www.reasonfieldlab.com/post/a-complete-guide-on-computer-vision-xai-libraries

reasonfieldlab.com

A complete guide on Computer Vision XAI libraries

The libraries mentioned and described vary greatly. Some are general-purpose, and some of them specialize only in one type of task, algorithm, or specific data type. Currently, there is no universal library that simultaneously contains all the algorithms and metrics. When choosing a library, it is necessary to decide what features are required in your current project.This indicates that there is room for new libraries which will solve this problem. It is noticeable that there is some space to be developed by new libraries. Few libraries provide metrics that determine the degree to which we can trust explanatory algorithms. There is a lack of integration with experiment trackers, which would allow users to monitor the training process of the network. Moreover, almost all libraries are limited to the simple case of image classification. Support for the segmentation task is lacking, and support for the object detection task is severely limited. By Adam Wawrzyński

Original URL: https://www.reasonfieldlab.com/post/a-complete-guide-on-computer-vision-xai-libraries

Aidan Dunlop (aidan.dunlop@sky.uk)

2023-04-19 11:44:43

thanks @Maciej Adamiak, will do! 🙂

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-04-24 23:47:06

I hope the data from the survey (anonymized) will be shared with the participants?

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-04-24 23:52:56

Also you should add "I have not heard of this tool" as an option 🙂

Aidan Dunlop (aidan.dunlop@sky.uk)

2023-04-25 09:51:14

@Aakash Gupta I’m happy to share the anonymised data 🙂

Aidan Dunlop (aidan.dunlop@sky.uk)

2023-04-25 09:52:25

fair point, the idea was that you don’t have to click on each of the rows, so by not selecting an answer for that row it’s implied that the participant hasn’t heard of the tool

Scott Hosking (jshosking@gmail.com)

2023-04-26 04:14:30

📢 Register for the Climate Informatics 2023 x @eds_book Reproducibility Challenge by midnight AoE this Friday 28 April!

👉 Learn more and register at bit.ly/ci-2023-rc-eds!

🐦 Like, RT, and tag others on Twitter to help us promote!

https://twitter.com/Climformatics/status/1651007755270905856

Twitter

Log in to Twitter to see the latest. Join the conversation, follow accounts, see your Home Timeline, and catch up on Tweets from the people you know.

Original URL: https://twitter.com/search?q=%23CI2023

Twitter

EDS Book | fosstodon.org/@EDSbook (@eds_book) / Twitter

A computational notebook community supporting the publication of #open #collaborative #reproducible #transparent Environmental science 🌎 Tweets by @alejo_coca

Original URL: https://twitter.com/eds_book/

Twitter

Climate Informatics on Twitter

📢 Register for the #CI2023 x @eds_book Reproducibility Challenge by midnight AoE this Friday 28 April! 👉 Learn more and register at <https://t.co/IuSIa3wkrv>!

Original URL: https://twitter.com/Climformatics/status/1651007755270905856

👍 Oisin Mac Aodha, Sara Beery, Shiva Muruganandham, Dimitri Korsch, Arron Watson

Nicky Nicolson (n.nicolson@kew.org)

2023-05-09 12:14:05

Hello! I guess a lot of people here are (or work with) research software engineers, so I wanted to share the start of the 2023 Diverse RSE talks (Supporting Equity, Diversity and Inclusion within the Research Software Engineering community). Next week (May 16th) there is a talk from Dave Horsfall who is using his Software Sustainability Institute fellowship to advocate for better mental health among RSEs: https://diverse-rse.github.io/events/2023-05-16

software.ac.uk

The Software Sustainability Institute

The Software Sustainability Institute cultivates better, more sustainable, research software to enable world-class research.

Original URL: https://software.ac.uk/

DiveRSE - Supporting EDI within the RSE community

The importance of talking about mental health in the Research Software Engineering community

The home of a series of talks and related activities for promoting awareness of and supporting EDI within the Research Software Engineering community.

Original URL: https://diverse-rse.github.io/events/2023-05-16

❤️ Sara Beery, Timm Haucke, Chris Llorca, Nora Gourmelon

Cara Appel (appelc@oregonstate.edu)

2023-05-09 16:44:50

Hello, does anyone have suggestions for appropriate preprint servers and/or journals for publishing a description of a software tool related to annotation and model training? Thanks in advance!

Michael Yair (m1cha3l.ya1r@gmail.com)

2023-05-10 01:57:53

maybe -
https://zenodo.org/

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2023-05-10 02:50:40

Depending on the scope of your software maybe the MDPI Software Journal or Elseviers SoftwareX. Both are journals intended for Tool-related publications 🙂 Additionaly, maybe also a Nature Scientific Report. But all three are still scientific, so a pure software documentation won‘t workout 😄

Peter van Lunteren (peter@addaxdatascience.com)

2023-05-10 04:58:52

If the software is open source: https://joss.theoj.org

joss.theoj.org

Journal of Open Source Software

Journal of Open Source Software (JOSS) is a developer friendly, open access journal for research software packages.

ISSN

2475-9066

Original URL: https://joss.theoj.org

😎 Jason Holmberg (Wild Me)

Cara Appel (appelc@oregonstate.edu)

2023-05-10 14:00:22

thank you! yes, it is open source

Gracie Ermi (gracieermiifthen@gmail.com)

2023-05-10 12:59:42

Hi everyone! Just wanted to plug that @Carly Batist and I released another update to the Conservation Tech Directory! We are up to 836 resources in the directory, so if you haven't checked it out yet (or in a while), head on over! As always, if you know of any additional conservation tech resources that aren't in the directory yet, you can fill out our google form and we'll get them added. And if you aren't subscribed to our email alerts, head over to the website so you can be notified any time we release an update!
AND we've now reached site visitors from 100 countries!! 🎉

conservationtech.directory

1st International Workshop on Spatio-Temporal Reasoning and Learning 2022

Visit the post for more.

Original URL: https://conservationtech.directory/

🎉 Carly Batist, Talia Speaker, Pen-Yuan Hsing, Clare Price, Sara Beery, Shir Bar, Timm Haucke, Andy Viet Huynh, Dan Morris, Viktor Domazetoski, Jason Holmberg (Wild Me), Diego Calanzone, Andrzej Białaś, Ruiz Rivera

🔥 Rowan Converse, Andy Viet Huynh, Jason Holmberg (Wild Me), Andrzej Białaś

Clare Price (theclareprice@gmail.com)

2023-05-10 15:56:57

Oh very cool! I work for a professor at UBC on her “Smart Earth Project,” where we are doing a similar compilation and database production using web scrapes; would love to connect with you and chat more about what you’re up to if you ever have some spare time.

👍 Carly Batist

🙌 Carly Batist

Ruiz Rivera (ruiz.rivera93@gmail.com)

2023-07-10 12:09:53

Thanks for sharing, this would be an awesome resource!

Ayan Mukhopadhyay (ayanmukg@gmail.com)

2023-05-15 12:50:53

Hi everyone, we are organizing the data science for social good workshop at KDD this year! Check it out and submit some cool papers 🙂

https://kdd-dssg.github.io/

kdd-dssg.github.io

In this workshop, we invite the research community in artificial intelligence to submit works related to the integration of spatial and temporal reasoning with machine learning.

Original URL: https://kdd-dssg.github.io/

👍 Oisin Mac Aodha, Sara Beery, Ankita Shukla, Maia Adar, Yseult Hb, Jaanak

👍:skin_tone_4: Chris Llorca

Katie Zacarian (katie@earthspecies.org)

2023-05-17 18:00:07

We're looking for a seasoned Research Director to join our growing Earth Species Project team and lead our long-term ML research agenda into the communication systems of other species. Please share with your networks!
https://lnkd.in/gQpHizXJ

earthspecies.org

We're Hiring! Director of Research - Earth Species Project

Earth Species Project is looking for a thought leader in artificial intelligence with a track record of managing a team of researchers.

Original URL: https://lnkd.in/gQpHizXJ

😍 Sara Beery, Jon Van Oast, Stefan Schneider, Jason Holmberg (Wild Me), Lucia Gordon, Juliana Gomez Consuegra

🎉 Jon Van Oast, Dan Morris, Jason Holmberg (Wild Me)

Tarun (tarunsharma.pes@gmail.com)

2023-05-19 14:38:16

Hi everyone, I'm curious about people's experiences using vision transformers such as ViT or transformer based detection models such as DETR, DINO, grounding DINO. I'm curious whether people have found these models to outperform CNN based classification/detection models by a big margin and how straightforward or cumbersome it was to fine tune these models in practice.

👀 Ben Williams, Rebecca Wilks

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2023-05-22 15:28:36

My experience with one recently (MetaFormer) is that they work. I don't know about "big margin" because frankly we didn't test on other architectures and we wanted something quick and dirty. In this case we picked the architecture that claimed SOTA on some tasks and tried it (and it certainly works well, even with 5k classes). To a great extent it depends on dataset quality and the usual how unbiased your evaluation is (e.g. checking that your test data are balanced and one easy class isn't skewing your results)

Did it work? Yes. Was it a pain because of hardcoded choices the researchers made in their repo? Yes 😅

A strong recommendation is to use HuggingFace's model zoo, they have "reference" implementations of a lot of this stuff and I think it's probably easier to use that than hack somebody's conference submission that only works on benchmark data. Inputs and outputs are well defined and the code is mostly well-tested. They also provide training examples for most of their models here https://github.com/NielsRogge/Transformers-Tutorials/tree/master

I think another interesting avenue for ecology that is under-explored is better pre-trained models from e.g. drone observations, camera traps, etc. DINO2 had some pretty good boosts for satellite analysis that way.

👍 Rebecca Wilks

Ben Weinstein (benweinstein2010@gmail.com)

2023-05-19 19:38:35

I’m working on a talk for the day-to-day responsibilities of a machine learning researcher in biology. I’m looking for literature, testimonials, thoughtful pieces on model development and iteration. A lot of what’s out there is just fluffy 1000 word stuff on medium. Ideas? I found this helpful.

❤️ Sara Beery

Michael Yair (m1cha3l.ya1r@gmail.com)

2023-05-20 06:02:59

I can share my specific experience with agent base model I developed a while ago.
there is the ODD protocol developed a specially for such model because of the complexity to explain the agent environment interaction.
I also worked on a pipeline for protein comparison, which i didn't find any protocol for (back at that days), but for sure - there where no similarities between the processes, well aside the IDE and the Python environment 🙂.

for further details about ODD:
https://www.jasss.org/23/2/7.html

Journal of Artificial Societies and Social Simulation

The ODD Protocol for Describing Agent-Based and Other Simulation Models: A Second Update to Improve Clarity, Replication, and Structural Realism

by Volker Grimm, Steven F. Railsback, Christian E. Vincenot, Uta Berger, Cara Gallagher, Donald L. DeAngelis, Bruce Edmonds, Jiaqi Ge, Jarl Giske, Jürgen Groeneveld, Alice S.A. Johnston, Alexander Milles, Jacob Nabe-Nielsen, J. Gareth Polhill, Viktoriia Radchuk, Marie-Sophie Rohwäder, Richard A. Stillman, Jan C. Thiele and Daniel Ayllón

Original URL: https://www.jasss.org/23/2/7.html

Caleb Robinson (calebrob6@gmail.com)

2023-05-21 19:21:38

Hey Ben, I found this post by Karpathy extremely helpful -- http://karpathy.github.io/2019/04/25/recipe/

karpathy.github.io

A Recipe for Training Neural Networks

Musings of a Computer Scientist.

Original URL: http://karpathy.github.io/2019/04/25/recipe/

John Martinsson (john.martinsson@ri.se)

2023-05-22 03:14:42

The paper "Perspectives in machine learning for wildlife conservation" by Tuia et al., 2022, may be helpful: https://www.nature.com/articles/s41467-022-27980-y.

Nature

Perspectives in machine learning for wildlife conservation

Nature Communications - Animal ecologists are increasingly limited by constraints in data processing. Here, Tuia and colleagues discuss how collaboration between ecologists and data scientists can...

Original URL: https://www.nature.com/articles/s41467-022-27980-y

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2023-05-22 15:22:47

Google has a fairly technical doc about this (on top of Karpathy's essays which are usually great - as well as his Zero to Hero series)

https://github.com/google-research/tuning_playbook

google-research/tuning_playbook

A playbook for systematically maximizing the performance of deep learning models.

Stars

19672

Last updated

10 minutes ago

👍 Ben Weinstein, Caleb Robinson, Sara Beery

Sara Beery (sara.beery@gmail.com)

2023-05-30 14:26:39

https://www.nature.com/articles/s41467-023-38901-y

Nature

Deep learning enables satellite-based monitoring of large populations of terrestrial mammals across heterogeneous landscape

Nature Communications - This study presents a deep learning pipeline to automatically locate and count large herds of migratory ungulates (wildebeest and zebra) in the Serengeti-Mara ecosystem...

Original URL: https://www.nature.com/articles/s41467-023-38901-y

❤️ Stefan Schneider, Hamed Alemohammad, Jon Van Oast, Suzanne Stathatos, Declan, Yseult Hb, Alex Brace, Elie Alhajjar, Taiki Sakai - NOAA Affiliate, Malte Pedersen, Carly Batist, Timm Haucke, Andrew Schulz, Clare Price, Devis Tuia, Yongjun Song, Caterina Barrasso, Caleb Robinson, Chuck Stewart, lin xiong, Pen-Yuan Hsing, Kurran Singh, Marcus Lapeyrolerie, Lucia Gordon, Omiros Pantazis, Magali Frauendorf, Felipe Montealegre-Mora, Juliana Gomez Consuegra, Rebecca Wilks, Fadel Seydou, Rajiv Pattni, Olof Mogren, Luke Sheneman, Eric Colson

📡 Stefan Schneider, Jon Van Oast, Elie Alhajjar, Shir Bar, Rowan Converse, Malte Pedersen, Carly Batist, Timm Haucke, Caleb Robinson, Casey Youngflesh, Brandon Hays

🤩 Santiago Ruiz Guzman

Fadel Seydou (fadel.seydou@gmail.com)

2023-06-06 04:17:29

@Paul Allin Hey, check this out

Paul Allin (allinpaul@gmail.com)

2023-06-10 02:20:46

Pretty amazing

Elizabeth Campolongo (e.campolongo479@gmail.com)

2023-06-01 17:51:12

Hi Everyone!
We’d love for you to join us Aug. 14-17 for Image Datapalooza: Call for Participation.
This exciting, participant-driven event will bring together an interdisciplinary group interested in using AI/ML to extract scientific knowledge from image and video data. We anticipate seeing participants ranging from AI/ML researchers, data scientists, domain scientists, and data curators, all the way to tool developers, metadata researchers, and knowledge engineers. Participants will self-organize into small groups to work hands-on and collaboratively with self-selected targets and outcomes towards the motivations and goals of the event.

To apply to participate, please fill out the Image Datapalooza 2023 Application for Participation by the end of June 12, 2023. Funds to assist with travel expenses are available but limited, as is space.

😍 Sara Beery, Andy Viet Huynh, Jaroslav Bezdek, Diego Calanzone, Rita Pucci

😎 Jon Van Oast, Timm Haucke

🙌 Jenna Kline

Elizabeth Campolongo (e.campolongo479@gmail.com)

2023-06-12 14:53:32

Reminder: Today is the last day to apply for Image Datapalooza!

Dan Stowell (dan.stowell@naturalis.nl)

2023-06-02 06:07:16

Job at UCL (London, UK): "Lecturer/Associate Professor of Ecology and Innovative Technologies" http://tinyurl.com/bzed2kxy Deadline this Monday!

👍 Oisin Mac Aodha, Omiros Pantazis, Andrew Schulz, Yseult Hb, Jinsu Elhance

👍:skin_tone_3: Pen-Yuan Hsing

🙌 Omiros Pantazis, Jinsu Elhance

Rajiv Pattni (rajivcpattni@gmail.com)

2023-06-07 04:56:08

Hello all! Great to be here :) and thanks for adding me @Sara Beery

I'm a cofounder of Terraspect (before that I was an environmental engineer). We're trying to mitigate terrestrial ecosystem services impact from buildings during their operational life by identifying chemical, light, noise and heat mitigants.

Does anyone know of any datasets on embodied ecological impacts of building stock and causal drivers ideally in the urban global south? (just no more great crested newt data please! 🦎🙂)

Thanks!

🦎 Maia Adar, Sara Beery

Ramya (rmalu001@ucr.edu)

2023-06-07 12:21:54

https://edzt.fa.em4.oraclecloud.com/hcmUI/CandidateExperience/en/sites/CX/job/14568

Candidate Experience site

Research Intern

Develop novel algorithms that promote fairness, accountability, and inclusivity of AI systems, with a special focus on environmental sustainability applications. Advance team and company goals through end-to-end system implementation Publish research papers in reputed scientific journals, conferences, and/ or workshops. Showcase the work via presentation in invited talks, conferences and workshops, and/ or live demo sessions.

Original URL: https://edzt.fa.em4.oraclecloud.com/hcmUI/CandidateExperience/en/sites/CX/job/14568

🙌 Maia Adar

Chinmay Talegaonkar (ctalegaonkar@ucsd.edu)

2023-06-08 21:53:21

Hi, can anyone point me event based datasets collected in the wild? For different species of animals ?

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-06-08 22:14:17

Have you checked the LILA dataset.

https://lila.science/datasets

👍 Taiki Sakai - NOAA Affiliate

Dan Stowell (dan.stowell@naturalis.nl)

2023-06-09 11:45:58

Hi all. Job offer! We're now recruiting for a Project Manager, for our new EU Doctoral Network project "Bioacoustic AI":
https://www.naturalis.nl/over-ons/project-manager-for-eu-doctoral-network-project-bioacoustic-ai
Please do share this. Happy to answer any questions. Thanks!

🔈 Oisin Mac Aodha, Stephanie O'Donnell, Sara Beery, Timm Haucke, Yseult Hb, Omiros Pantazis

❤️ Carly Batist, Sara Beery, Timm Haucke, Jon Van Oast

Evangeline Corcoran (ecorcoran@turing.ac.uk)

2023-06-12 07:23:42

Hi all, I'm guest editing a special issue of Remote Sensing on remote sensing applications in biodiversity conservation and ecological modelling, which is accepting submissions until October 9th 2023. If you are interesting in contributing please feel free to message me directly or check out the special issue link (https://www.mdpi.com/journal/remotesensing/special_issues/8OQ6QPGDQ2). Thanks! 📡🌏

Remote Sensing

Remote Sensing, an international, peer-reviewed Open Access journal.

Original URL: https://www.mdpi.com/journal/remotesensing/special_issues/8OQ6QPGDQ2

📡 Ronan Wallace, Jon Van Oast, Sara Beery, Declan, Andy Viet Huynh, Katelyn Morrison, Sepand Dyanatkar

❤️ Ronan Wallace, Andy Viet Huynh

👀 Heather

Sara Beery (sara.beery@gmail.com)

2023-06-12 19:46:13

Also, if you aren't attending CVPR but you are in the Vancouver, BC area, feel free to join in and meet up as well!

💕 Jon Van Oast

👍 Thor Veen

Sara Beery (sara.beery@gmail.com)

2023-06-12 19:46:38

Also crowdsourcing locations for an informal happy hour that is good for groups!

Aniruddha Saha (anisaha1@umd.edu)

2023-06-15 09:21:46

Visited these places with large groups last time I was in Vancouver a few years ago

https://goo.gl/maps/upccLW9RKWe1jVQC6

https://goo.gl/maps/nN9UZbU7TJvngNmk9

https://goo.gl/maps/Hv8wuJcTQ8ut2c3GA

Earls was great. That would be my top recommendation. But things might have changed quite a bit over the years.

google.com

Earls Kitchen + Bar · 1185 Robson St, Vancouver, BC V6E 1B5, Canada

★★★★☆ · Restaurant

Original URL: https://goo.gl/maps/upccLW9RKWe1jVQC6

google.com

Patron Tacos & Cantina · 265 Robson St, Vancouver, BC V6B 0E7, Canada

★★★★☆ · Mexican restaurant

Original URL: https://goo.gl/maps/nN9UZbU7TJvngNmk9

google.com

The Portside Pub · 7 Alexander St, Vancouver, BC V6A 1E9, Canada

★★★★☆ · Pub

Original URL: https://goo.gl/maps/Hv8wuJcTQ8ut2c3GA

💕 Jon Van Oast

Drea Burbank (drea@savimbo.com)

2023-06-18 17:34:15

Savimbo just released an indicator-species biodiversity crediting methodology for review that allows indigenous groups to sell direct to climate markets, by calculating their environmental credits on Google Earth Engine. We are looking for developers to help us review and open-source this code. Please dm me if you want to be a technical reviewer. Its gone to Cercarbono, Verra, and Plan Vivo for certifying body adoption.

Savimbo

Savimbo — Fair-trade carbon credits

Original URL: https://www.savimbo.com

👍 Ted Schmitt, Griffin Flannery, Heather

George Darrah (george.darrah@systemiq.earth)

2023-06-19 04:50:15

Have been thinking/doing/investing at the intersection of biology and AI - here's a synthesis of the story so far: https://www.linkedin.com/posts/georgedarrah_the-key-to-natures-intelligence-is-artifi[…]440811884544-IFDm?utm_source=share&utm_medium=member_desktop

@Aamir Ahmad thanks again for inspiration re the zebras!

linkedin.com

George Darrah on LinkedIn: The key to nature’s intelligence is artificial intelligence — Systemiq…

Building enduring start-up value in an age of commoditised and rapidly evolving LLMs needs deep, proprietary and differentiated datasets. Nature is uniquely…

Original URL: https://www.linkedin.com/posts/georgedarrah_the-key-to-natures-intelligence-is-artificial-activity-7075478440811884544-IFDm?utm_source=share&utm_medium=member_desktop

👀 Stephanie O'Donnell, Jason Holmberg (Wild Me)

👍 Talia Speaker

❤️ Katie Zacarian

Levi Farrand (levi.farrand@gmail.com)

2023-06-20 07:22:14

Hi all. I am new to the group and wanted to introduce our start-up company - Deep Forestry.

We have built and autonomous drone that can autonomously fly into forests, in-between the trees, to create digital twins of entire ecosystems in a cost-effective and scalable way, at the push of a button with no drone piloting expertise required. We use three dimensional deep learning algorithms to segment and classify different objects in the forest and have a strong team of robotic developers and AI experts that are always looking for new ways to adapt our existing systems for the benefit of conservational goals. We are always looking for new pilot studies to test and improve our systems on different tasks.

If you need access to robotics and AI expertise, autonomous lightweight (or heavy lift) drones, or expansive digital twins of forest ecosystems at the push of a button, perhaps we could be a good match for collaboration. Note that we're also staring to work with hyperspectral data too if that is of interest. We are always happy to add our expertise to collaborative research projects. Note also that our first autonomous drone products are now commercially available too, so we can also provide you with a quick and easy way to conduct large-scale ecosystem surveying in projects that you may already have funded.

Email me (CEO) anytime to connect - Levi@deepforestry.com

Company introduction - https://youtu.be/xSe4M_zk4oQ

Cloud services - https://youtu.be/796pBsJuNx8

Autonomous robotics within the forest - https://youtu.be/G5wqgTNVwFI

👋 Omiros Pantazis, Elie Alhajjar, Viktor Domazetoski, Carly Batist, Yseult Hb, Ankita Shukla, Hannah Kim

👍 Yonghao Xu, lin xiong, Juliana Gomez Consuegra

👋:skin_tone_4: Chris Llorca

👍:skin_tone_4: Chris Llorca

Arvin Sun (sunbingyou1984@gmail.com)

2023-06-20 11:23:48

Hi guys,

Nice to meet you all! 👋 Thanks Sara having me here! 🫰 I’m the CEO & Founder of Traini. And a longtime champion of serial entrepreneur and innovation culture. I have been engaged in food delivery, e-commerce, new retail entrepreneurship and team management in China and the United States for more than 10 years.

1, Based in Palo Alto, before in LA.
2, Founder of Traini, An AI driven platform for dog trainers and parents. (Motion capture and PetGPT), and
•US Partner - DayDayCook ( Pre-IPO NYSE )
•Mentor - Visionary Education Foundation.
•Investor - Focus on growth-stage startups (A Round).
3, Immersed in Stanford every day, love to play basketball.
My LinkedIn: http://linkedin.com/in/arvinsun

👋 Arvin Sun

Silvia Zuffi (silvia@mi.imati.cnr.it)

2023-06-20 16:47:10

Who wants to try getting a 3D dog model from an RGB image? Try out our demo here: https://huggingface.co/spaces/runa91/bite_gradio

huggingface.co

BITE - a Hugging Face Space by runa91

Discover amazing ML apps made by the community

Original URL: https://huggingface.co/spaces/runa91/bite_gradio

😍 Sara Beery, Justin Kay, Mitchell Rogers, Viktor Domazetoski, Andrew Schulz, Timm Haucke, Elie Alhajjar

🐶 Rowan Converse, Yseult Hb, Dan Morris, Oisin Mac Aodha, Timm Haucke, Elie Alhajjar, Rebecca Wilks, Arvin Sun

Devis Tuia (devis.tuia@epfl.ch)

2023-06-21 10:08:47

👍 Oisin Mac Aodha, Silvia Zuffi, Sara Beery, Jason Holmberg (Wild Me), Anastasios Angelopoulos, Jon Van Oast

🐳 Oisin Mac Aodha, Matthias Zuerl, Katelyn Morrison, Anastasia Pagán, Sara Beery, Elizabeth Campolongo, Jason Holmberg (Wild Me), Anastasios Angelopoulos, Nico Lang

Giana Cirolia (giana@berkeley.edu)

2023-06-22 22:08:00

Hi Climate enthusiasts,

There is a VC that I know of that is rolling out a grant funding initiative, specifically focused on synthetic biology and climate.

The grants are fast grants (30 min application, 21 day turn around and up to 100k per project) with no IP taken and no reporting.

They are looking for BOTH applicants (project proposals) AND PIs to join their expert review panel.

Would anyone in this extended network be interested in being an expert reviewer on this climate initiative? Please forward widely. You can participate even if your general focus is simply climate!

Professors can both be reviewers and apply for the grants...AND many students per lab can apply.

Happy to directly connect any interested PI's with the program leads.

Warmly,
Giana Cirolia

See below for full details:

As mentioned, Manifest Grants is a 'fast grants' program. We award $25K - $100K to scientists to prototype their most ambitious synbio ideas for solving the climate crisis. We're looking for senior reviewers for the June-July period.

Expectations:

Review ~10 applications.

Review time ~ 40mins per application. Fully numeric scoring system, no write-up required!

Get paid $30 per application!

Submit the review within 8 days.

Sign an NDA to protect the confidentiality of the proposed ideas.

You can apply even if you participate in the review!

Interested? Here's my calendar if helpful.

P.S Our first iteration was with Repro Grants. We received 465 applications from all the top universities and funded 12 projects in female reproductive science. Read more (Forbes article).
Sara

—
Sara Kemppainen
Fifty Years
Linkedin / Twitter

Future

What We Learned Doing Fast Grants

We launched an abnormally fast source of emergency science funding during COVID-19. The results revealed flaws in current science funding models.

Written by

Patrick Collison, Tyler Cowen, and Patrick Hsu

Est. reading time

18 minutes

Original URL: https://future.com/what-we-learned-doing-fast-grants/

Forbes

Fifty Years Launches ‘Fast Grants’ For Female Reproductive Science To Close The Gender Research Gap

Repro Grants is a “fast grants” program, that awards $25,000 - $100,000 to ambitious research projects aiming to deepen our understanding of the female reproductive system. Women’s health attracts only a minuscule percentage of all science funding – about 11% of all NIH funding in 2019.

Original URL: https://www.forbes.com/sites/marijabutkovic/2022/11/17/fifty-years-launches-fast-grants-for-female-reproductive-science-to-close-the-gender-research-gap/?sh=2cbce6005acb

👍 Jon Van Oast

Giana Cirolia (giana@berkeley.edu)

2023-06-23 13:19:48

Thank you all to the initial interest! Please keep
Direct messaging me :)

Heather (h_peacock@ducks.ca)

2023-07-19 16:56:40

Hi @Giana Cirolia, can you provide a link for the funding applications/submissions? Are these grants only available to academics / in the US, or can NGOs apply for some projects?
Thanks!

Vaughn Shirey (vmshirey@gmail.com)

2023-06-23 14:06:29

Hi all! I'm new to the group and just wanted to throw a quick intro on the channel.

I'm a current postdoctoral fellow at the University of Southern California looking at how we can use CV to rapidly mobilize historical biodiversity data in order to reconstruct past distributions, phenologies, and morphologies. The end goal is understanding how global change processes are impacting biodiversity on the planet and to translate these data and reconstructions to on-the-ground conservation organizations looking to conduct species' assessments and influence policy. I largely do this work with butterflies but am interested in all other groups as well (especially other invertebrates!).

Looking forward to being here and learning more! BTW, you can find me on Twitter, LinkedIn, or ResearchGate if you'd like to chat or feel free to shoot me a DM. :)

Twitter

Vaughn Shirey, Ph.D. (@vmshirey) / Twitter

Artificial intelligence, conservation biology, and insects. Incoming @SmithFellows at USC; Formerly Georgetown, Fulbright :flag_fi:, Drexel. Opinions my own. :spock_hand::skin_tone_3:

Original URL: https://twitter.com/vmshirey

👋 Jon Van Oast, Yseult Hb, Vincent Christlein, Shir Bar, Lucia Gordon

👋:skin_tone_4: Chris Llorca

Philippe Hermant (philippe.hermant@entropisme.com)

2023-06-23 19:17:32

Hello, being new here I’d like to know if a summary of impacts on biodiversity exists for AI ? As there are numerous topics, does such an overview exists ? And something specific to generative AI towards biodiversity ?

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-06-23 20:11:07

Not sure if I fully understand your ask, but this paper is a great overview of the ML space within biodiversity! https://www.nature.com/articles/s41467-022-27980-y

Nature

Perspectives in machine learning for wildlife conservation

Nature Communications - Animal ecologists are increasingly limited by constraints in data processing. Here, Tuia and colleagues discuss how collaboration between ecologists and data scientists can...

Original URL: https://www.nature.com/articles/s41467-022-27980-y

🤗 Blair Costelloe

Rajiv Pattni (rajivcpattni@gmail.com)

2023-06-25 07:05:21

Likewise not sure if totally got the ask but this paper seems relevant too
https://gpai.ai/projects/responsible-ai/environment/biodiversity-and-AI-opportunities-recommendations-for-action.pdf

Philippe Hermant (philippe.hermant@entropisme.com)

2023-07-07 08:39:16

Thanks @Carly Batist @Rajiv Pattni recommendations are useful.

Sonny Burniston (sonnyburniston@yahoo.co.uk)

2023-06-24 09:34:32

Hi Everyone,
Just thought I’d briefly introduce myself. I work as a software engineer at NatureMetrics primarily working on backend systems and data pipelines. I have a keen interest in ML engineering and AI in general. I hope to learn more from everyone in this community about how this can be applied to Biodiversity. If you have any interesting opportunities or would like to collaborate please reach out! (Particularly interested in any research opportunities available)
Thanks!
Sonny 🙂

NatureMetrics

eDNA Surveys | Biodiversity Analysis & Monitoring Experts - NatureMetrics

NatureMetrics is a world-leading provider of biodiversity monitoring data, using environmental DNA (eDNA) surveys, metabarcoding and more.

Original URL: https://www.naturemetrics.com/

👋 Carly Batist, Elie Alhajjar, Jason Holmberg (Wild Me), Jon Van Oast, Sara Beery, Dan Morris

🙂 Rajiv Pattni

🧬 Dan Morris

Jacob Marks (jamarks13@gmail.com)

2023-06-26 14:27:15

Hi everyone!

My name is Jacob and I'm an ML engineer and developer evangelist at an open source cv/ml company Voxel51. Before this, I completed my PhD in physics, and I'm really excited about applications of AI for climate and conservation. Thanks @Sara Beery for letting me join this awesome group!

Three Opportunities:

I help run a virtual computer vision meetup with 4,500+ members. If anyone here would be interested in speaking at one of our events, please let me know!

I write on Medium, and I'd love to highlight some of the incredible work this community is doing. If you are using CV in climate/conservation and want me to write about it, feel free to reach out 🙂

The company I work for helps make datasets more accessible - we worked with authors of a bunch of CVPR datasets to make their datasets publicly browsable at cvpr.fiftyone.ai. If you want to make it easier for others to view and use your datasets, please reach out!

Thanks so much, and so excited to connect with the amazing people here

Voxel51

Voxel51 // Open source computer vision tools for machine learning

We enable ML engineers to build better computer vision datasets & models, more quickly. Try FiftyOne, the powerful open source computer vision toolset.

Est. reading time

17 minutes

Original URL: https://voxel51.com/

meetup.com

Computer Vision Meetups | Meetup Pro

Computer Vision Meetups is on Meetup Pro with more than 4510 members across 13 Meetups worldwide. Meetup Pro is the professional tool for organizing and communicating a network of users, partners, contributors and members.

Original URL: https://www.meetup.com/pro/computer-vision-meetups/

🙌 Peter van Lunteren, Ben Weinstein, Vincent Christlein, Katelyn Morrison, Victor Anton

👋 Peter van Lunteren, Aakash Gupta, Dimitri Korsch, Neil Kale

🙌:skin_tone_3: Pen-Yuan Hsing

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-06-28 16:31:33

🚨RFCx/Arbimon is hiring!!
Data science folks - Come work with me and an absolutely all-star group developing AI/ML techniques and models for acoustic monitoring & biodiversity conservation!
https://storage.googleapis.com/rfcx-wordpress-media/2023/06/406f13e6-rfcx-data-scientist.pdf

🔊 Suzanne Stathatos, gvanhorn, Michael Yair, Katelyn Morrison, Jason Holmberg (Wild Me)

😎 Jon Van Oast, Ștefan Istrate, Katelyn Morrison, Maddie Cusimano, Jason Holmberg (Wild Me)

❤️ Talia Speaker, Matt Weldy, Yseult Hb, Jason Holmberg (Wild Me), Prabath Gunawardane, Nicolas Arrieta Larraza

👀 Taiki Sakai - NOAA Affiliate

Ditiro Rampate (ditirorampate@gmail.com)

2023-06-29 09:17:53

📢 Seeking your support to attend Deep Learning Indaba in Accra, Ghana! 🌟

Hi everyone! 👋

I wanted to reach out and ask for your support in attending the Deep Learning Indaba in Accra, Ghana. As an Omdena chapter lead in Botswana and a member of Sisonke Biotik, an AI and healthcare community in Africa, this conference presents a fantastic opportunity for me to enhance my AI skills and contribute to the development of AI in Africa.

The Deep Learning Indaba brings together some of the brightest minds in AI and offers a platform for learning and networking. By participating in this event, I'll have the chance to connect with experts and passionate individuals, accelerating our efforts to leverage AI advancements in Africa and improve access, quality, and equity in AI development.

I kindly ask for your support by considering a donation towards my journey. Every contribution, no matter the size, will bring us closer to unlocking the potential of AI for the betterment of Africa.

Please donate here: https://gogetfunding.com/help-ditiro-attend-deep-learning-indaba-in-accra/

Additionally, I would greatly appreciate it if you could share this message in your networks to help spread the word. Together, we can make a lasting impact and create a brighter future through the power of AI.

Thank you for your support and belief in the transformative power of AI. Let's seize this opportunity and drive positive change in Africa!

Best regards,

Ditiro Rampate

Omdena | Building AI Solutions for Real-World Problems

Gaborone, Botswana Chapter - Omdena

The mission of Omdena’s Gaborone, Botswana Chapter is to run open-source AI projects to solve challenges faced by the local community.

Original URL: https://omdena.com/local-chapters/gaborone-botswana-chapter/

sisonkebiotik.africa

SisonkeBiotik - DL Indaba 2023

@ Deep Learning Indaba, Accra, Ghana | 3-9 September 2023

Original URL: https://www.sisonkebiotik.africa/events/workshops/dl-indaba-2023

❤️ Helge Rhodin

Ditiro Rampate (ditirorampate@gmail.com)

2023-08-14 06:04:13

Hello guys👋, I am still raising funds to attend the Deep Learning Indaba this September in Accra. I also have fundraising posts in twitter and LinkedIn and would appreciate if you could not only but share among your networks for better reach.

Deep Learning Indaba 2023

Indaba - Deep Learning Indaba 2023

The Deep Learning Indaba 2023 will be held in Accra, Ghana from the 3rd to the 9th of September 2023 with the theme of Yɛbɛtumi, an Akan (Twi) word meaning “we can”.

Est. reading time

2 minutes

Original URL: https://deeplearningindaba.com/2023/

Twitter

Ditiro on Twitter

📢 Seeking your support! 🌟 I need your help to attend the Deep Learning Indaba in Accra, Ghana. 🚀 Please consider donating to my fundraising campaign: <https://t.co/lhqTcFTL6k> Let's shape the future of AI together! 🌍✨ #DeepLearningIndaba #SupportEducation #HelpMeAttend

Original URL: https://twitter.com/ditiro_dt/status/1673393910557581315

linkedin.com

Ditiro Rampate on LinkedIn: Help Ditiro attend Deep Learning Indaba in Accra

🌟 Seeking Support for Deep Learning Indaba Attendance 🌍✨ I&#39;m thrilled to have the opportunity to attend the Deep Learning Indaba in Accra, Ghana! 🚀 This…

Original URL: https://www.linkedin.com/feed/update/urn:li:activity:7079161351008309248/

👏 Jon Van Oast

Luisa Orsini (l.orsini@bham.ac.uk)

2023-07-01 09:39:52

hello everyone, my name is Luisa. My research helps develop new tools for biodiversity monitoring and forecasting. I work with industry and regulators to find practical ways to halt biodiversity loss. I am also the co-lead of the Alan Turing interest group Biodiversity monitoring and Forecasting: https://www.turing.ac.uk/research/interest-groups/biodiversity-monitoring-and-forecasting

🙌 FANQI Z, Carly Batist, Sonny Burniston, Felipe Parodi, Nazanin Rezaei, Katelyn Morrison, Elie Alhajjar, Andrew Schulz, Sara Beery, Anastasios Angelopoulos, Aida Mashkouri, Juliana Gomez Consuegra, Thor Veen

👋 Carly Batist, Ștefan Istrate, Sonny Burniston, Andre Telfer, Katelyn Morrison, Elie Alhajjar, Oisin Mac Aodha, Shir Bar, Sara Beery, Anastasios Angelopoulos, Cameron Trotter, Dimitri Korsch, Cathy Atkinson, Aakash Gupta, Heather

Marundu Muturi (marundu@appsilon.com)

2023-07-03 07:32:49

👋 Hello, Good People.

I'm Marundu from Appsilon and I'm excited to share some Mbaza-related news: 📢 🗣️

There are two new Mbaza add-ons ready for you to use. Both are installable from Github and work out-of-the-box on the csv output from Mbaza:

mbaza-sequencer: Group images (sequences) together to give a combined prediction

Experiments find this improves classification accuracy by up to 5%

In the example, the delay max_delay=5, meaning images will be sequenced together until the image-to-image delay is > 5 seconds.

Why is it useful?
The first prediction in the csv is of a hare 🐇 and you will see why when you see the first image! (shown near the end)

Combining the predictions corrects this to an aardvark 💪

2. mbaza-mv-predicted: Move classified images to year / week / species folder

Both work on the .csv file output from Mbaza for Python 3.8+. The repositories also have installation and usage instructions, let us know if anything is unclear or missing in the documentation.

Attached is also a video demo using the two add-ons. A couple of things to note:

The first prediction from the Mbaza .csv file is a Hare, but the sequencer corrects this to an Aardvark after grouping predictions (images shown at end)

In the example used, an image-to-image delay to sequence images is used (< 5 seconds) but you can also specify a maximum number of images per sequence (see the documentation for all options)

If you have any questions, let me know.

appsilon.com

R Shiny | Enterprise R Shiny Dashboards | R Consulting

Appsilon delivers software development and data science solutions. Posit (RStudio) Full Service Certified Partner. Let's work together.

Original URL: https://appsilon.com/

🙌 Peter van Lunteren, Oisin Mac Aodha, Sara Beery

❤️ Antón Álvarez

Dan Watson (dan@sntech.co.uk)

2023-07-10 09:34:00

Hi everyone. Can't believe I'm only just joining this Slack but it's good to be here (thanks for the suggestion @Andrew Schulz!). Good to see some familiar faces!
I've recently been looking into underwater computer vision and have a few questions to ask around applying it to different things. Would love to chat with folks working on that type of technology.

👋 Daniel Grzenda, Sara Beery, Andrew Schulz, Justin Kay, Cameron Trotter, Jon Van Oast, Katelyn Morrison

Antonio Ferraz (antonio.a.ferraz@jpl.nasa.gov)

2023-07-10 17:56:45

We are hiring a postdoc to study and prototype innovative satellite-tag communication systems to track the movement of animals from space.

The selected candidate will work in an interdisciplinary team, including experts on the NASA-JPL Mission Designs, leaders of active and passive Radio Frequency satellite communications and animal movement ecologists.

👍 G. Andrew Fricker

Antonio Ferraz (antonio.a.ferraz@jpl.nasa.gov)

2023-07-10 17:57:04

https://www.jpl.jobs/job/R4460/Postdoc-Spaceborne-Satellite-tag-RF-Systems-to-Measure-Spatiotemporal-Patterns-of-Wildlife

JPL (Jet Propulsion Laboratory)

Postdoc: Spaceborne Satellite-tag RF Systems to Measure Spatiotemporal Patterns of Wildlife | Careers at NASA's Jet Propulsion Laboratory (JPL)

Apply for Postdoc: Spaceborne Satellite-tag RF Systems to Measure Spatiotemporal Patterns of Wildlife job with JPL in Pasadena, California, United States of America. Educational. Let's Dare Mighty Things together.

Original URL: https://www.jpl.jobs/job/R4460/Postdoc-Spaceborne-Satellite-tag-RF-Systems-to-Measure-Spatiotemporal-Patterns-of-Wildlife

🔥 Dan Watson, Sara Beery, Toryn Schafer, Ando Shah

Maddie Cusimano (maddie@earthspecies.org)

2023-07-11 01:13:13

curious if anyone will be at the animal behavior society conference this week in portland?

👋 Blair Costelloe

👍 Jon Van Oast

👏 Katie Zacarian

Blair Costelloe (blaircostelloe@gmail.com)

2023-07-11 10:15:01

See you soon at the drone workshop, actually!

😄 Maddie Cusimano

Edward Bayes (bayesbayes@gmail.com)

2023-07-12 17:55:35

Hi everyone! I'm looking to fine-tune MegaDetector and a few other models for a tiger conservation project I'm working on with the Government of Bhutan and was wondering if anyone has come across any tiger-specific camera trap datasets or models? All I've been able to find is this dataset - https://lila.science/datasets/atrw - and this model - https://github.com/KupynOrest/AmurTigerCVWC - but they use images from zoos and I'm keen to use images from the wild if possible. Thanks so much! Ed

🐅 Sara Beery, Jason Holmberg (Wild Me), Ariel Chamberlain, Anastasios Angelopoulos, Shir Bar, Cameron Trotter

🐯 Sara Beery, Jon Van Oast, Jason Holmberg (Wild Me), Dan Morris, Anastasios Angelopoulos

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2023-07-12 17:57:21

Just detection or re-ID too? We're working on multi-species big cat re-ID models and were going to include the lila.science dataset as well. Happy to collaborate. Feel free to DM directly.

🎉 Jon Van Oast, Edward Bayes

Edward Bayes (bayesbayes@gmail.com)

2023-07-12 18:00:17

Hi Jason, thanks so much for coming back to me. Just detection initially but we'd like to explore re-id in the future! I'll DM you. Thanks so much! 😊

Dan Morris (agentmorris@gmail.com)

2023-07-12 18:46:58

I can't help you with "tiger-specific datasets", but if I grab the giant .csv file that contains taxonomic information for all the camera trap data on LILA:

https://lila.science/taxonomy-mapping-for-camera-trap-data-sets/

...and I grep for "panthera tigris", I get 321 hits. Not a lot, but that's something. If you download that big .csv file and run:

cat lilaimageurlsandlabels.csv | grep -i "panthera tigris" | cut -d "," -f 2

...you should get a flat list of URLs that's easy to download, e.g.:

https://lilablobssc.blob.core.windows.net/wcs-unzipped/animals/0462/1185.jpg

Tigers are an interesting case where basically all of the candidate data comes from areas that (for good reason) have historically been hesitant to allow data publication, even images, so data sparsity is a bit fundamental to your problem. Maybe 321 wild camera trap images plus the handful of zoo images is good enough to get you started... does the organization you're working with have any labeled images, and/or do they think they have tigers amongst a pile of unlabeled images?

Also when you say "fine-tune MD", confirm that you mean "add species (or tiger/non-tiger) classes to MD", as opposed to "make MD more accurate"? There are lots of situations where fine-tuning to make MD more accurate would be worth it, but my guess is that MD is probably doing asymptotically well on tigers: they're big, they're patterned, they look like lots of other species MD saw in training, and we did have a bunch of tiger data from the Wildlife Institute of India during training.

Original URL: https://lilablobssc.blob.core.windows.net/wcs-unzipped/animals/0462/1185.jpg

❤️ Edward Bayes, Aakash Gupta

Dan Morris (agentmorris@gmail.com)

2023-07-12 18:50:18

And of course, there's always the time-tested strategy of begging Ecology Twitter for data.

😆 Edward Bayes

Shivam Shrotriya (shivam.shrotriya@gmail.com)

2023-07-13 02:58:29

Hi Edward,
Here is a fine scale model for MegaDetector (https://github.com/bhlab/SpSeg) that works really well in segregating tigers (>98% accuracy). This model was developed using datasets from Central Indian landscape, so it would work fine for Bhutan as well.

Unfortunately, training dataset isn't available for public use.

bhlab/SpSeg

Tool for species-level segregation of camera-trap images. The 'Species Segregator' is currently trained for Central Indian Landscape.

Stars

Language

Jupyter Notebook

❤️ Edward Bayes

Edward Bayes (bayesbayes@gmail.com)

2023-07-13 04:35:16

Thanks so much Dan and Shivam!

@Dan Morris - really useful feedback, thanks so much! I’ll find those 321 images in the LILA dataset! And totally hear your point re: data scarcity. They do have labeled data, yes, but to exactly your point there are sensitivities in sharing such data, particularly outside the country, so before navigating that thorny issue, we wanted to see how performant we could get a model on publicly available data. Re: fine-tune, I mean classes yes, not improving accuracy (using your super help tutorial as a start https://www.kaggle.com/code/agentmorris/fine-tuning-megadetector)

@Shivam Shrotriya - thank you so much! It sounds like this might be exactly what I was looking for. Thanks again!

kaggle.com

Fine-tuning MegaDetector

Explore and run machine learning code with Kaggle Notebooks | Using data from megadetector-tutorial

Original URL: https://www.kaggle.com/code/agentmorris/fine-tuning-megadetector

Dan Morris (agentmorris@gmail.com)

2023-07-13 19:13:40

Definitely let us know what you learn re: SpSeg, fine-tuning, etc.

The only minor thing I'll correct here - in the interest of giving credit where credit is due - is that the fine-tuning notebook you linked to isn't really mine, I just forked it temporarily to make minor copy edits that were pulled into the upstream version. The original version is here:

https://www.kaggle.com/code/evmans/train-megadetector-tutorial

It is a really nice tutorial, but I can't take credit for how nice it is. :)

❤️ Edward Bayes

Edward Bayes (bayesbayes@gmail.com)

2023-07-13 20:41:53

shall do! thanks again!

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-07-14 15:34:01

So a suggestion for such long tailed classes. Is to run an image search on google/Bing or other search engines. It will give you a large set of images, which you can simply download and use for training.

👍 Edward Bayes

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-07-14 15:37:17

I workwith the Indian government for deploying an AI platform for processing and storage of CT images. A lot of of forest officers in India are on twitter. And they keep posting wildlife images. A search on twitter should give you some datapoints.
DM me, if you would like to talk.

👍 Edward Bayes

Patrick Beukema (patrickb@allenai.org)

2023-07-14 18:40:22

We (Skylight, AI2) are hiring an MLE to help build maritime intelligence applications for ocean conservation: https://boards.greenhouse.io/thealleninstitute/jobs/5171288 #AI4Good

skylight.global

Skylight | Home

Skylight helps improve maritime transparency and protect our oceans.

Original URL: https://www.skylight.global/

boards.greenhouse.io

Sr. Machine Learning Engineer, Skylight

Seattle, WA

Original URL: https://boards.greenhouse.io/thealleninstitute/jobs/5171288

🐟 Suzanne Stathatos, Omiros Pantazis, Alan Papalia, Kristina Kupferschmidt

😎 Jon Van Oast, Katelyn Morrison, Alan Papalia, Kristina Kupferschmidt, Dan Watson

❤️ Ben Williams, Aakash Gupta, Emilio Luz-Ricca

Patrick Beukema (patrickb@allenai.org)

2023-07-14 19:52:05

Does anyone here know what the best strategy is for generating precise geolocation of every pixel in an Sentinel-2 image? I worked with VIIRS data for a previous model and NASA produces latitude and longitude arrays for the raw imagery which enables very precise geolocations for object detection. Their method is described in this technical report: https://www.star.nesdis.noaa.gov/jpss/documents/ATBD/D0001-M01-S01-004_JPSS_ATBD_VIIRS-Geolocation_B.pdf Does anyone know whether it is possible to replicate that method with the Sentinel-2 data to obtain similar lat/lon arrays for each pixel?

Hamed Alemohammad (h.alemohammad@gmail.com)

2023-07-19 12:26:42

Hi @Patrick Beukema, each S-2 scene has metadata in the GeoTIFF file which contains the coordinates of the upper left pixel of the image, and the resolution of each pixel. These two are enough to generate the coordinates of each pixel in the image. If you load the file using rioxarray in Python you already get the coordinates in the dataset array.

Patrick Beukema (patrickb@allenai.org)

2023-07-21 14:43:21

hey thanks yeah I am familiar with that method, my goal is to build the most precise coordinates possible. Do you know what the precision of that method is and if its optimal?

Hamed Alemohammad (h.alemohammad@gmail.com)

2023-07-21 19:32:00

Reading the scene metadata, as I suggested, you are basically assigning the coordinates to each pixel based on the orthorectification that the data provider did. rioxarray or any other package doesn't change the precision of the coordinates and they are as accurate as provided in the metadata.

If you are interested in correcting the alignment of scenes through time, aka mis-registration error, then the problem is something else. There are methods in the literature to mitigate this but I don't know the best ones.

Luke Sheneman (sheneman@uidaho.edu)

2023-07-17 20:50:17

While studying endangered Northern Idaho Columbian Ground Squirrels at the University of Idaho, we successfully deployed MegaDetector at the edge for near-realtime autonomous decision making workflows in the field. We're using a combination of specially instrumented enclosures, NVIDIA Jetson Nano compute, and various sensors and cameras.

🐿️ Sara Beery, Felipe Parodi, Dan Morris, Sowbaranika, Carly Batist, Vincent Christlein, Julius Roeder, Timm Haucke, Ștefan Istrate, Jason Holmberg (Wild Me), Shir Bar, Michael Bunsen, Emilio Luz-Ricca, Boyu Zhang, Edward Bayes, Rebecca Wilks

💚 Jon Van Oast, Jason Holmberg (Wild Me), Michael Bunsen

👀 Dante Wasmuht, Elizabeth Campolongo, Michael Bunsen, Oscar Schafer

👍 Olivier Gimenez, Dante Wasmuht, Maddie Cusimano, Michael Bunsen, Evan Eskew, Chris Yeh

Sara Beery (sara.beery@gmail.com)

2023-07-17 20:54:43

This is SO COOL

Dan Morris (agentmorris@gmail.com)

2023-07-18 10:55:59

Very cool indeed! Is there a two-sentence summary of what the automated decisions would be? E.g. are you capturing/tagging automatically for survey purposes?

Luke Sheneman (sheneman@uidaho.edu)

2023-07-18 12:08:42

@Dan Morris Right now the system uses the real-time output of MegaDetector to decide whether or not there is an animal in the detection chamber, and if so it continues data collection until no identifiable animal is present. We are extending that to an automated mark/recapture mechanism with the eventual goal of precision-targeted vaccine delivery in the field, down to the individual level.

👍 Dan Morris, Sara Beery, Jason Holmberg (Wild Me), Shir Bar, Boyu Zhang

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2023-07-18 13:59:14

Let me know if Wild Me (wildme.org) can be a future part of individual ID part.

👍 Sara Beery

Luke Sheneman (sheneman@uidaho.edu)

2023-07-18 14:30:47

@Jason Holmberg (Wild Me) Definitely! Looking now at your website. Individual CV ID for small mammals like deer mice and ground squirrels is challenging. I'll be in touch.

👍 Sara Beery, Jason Holmberg (Wild Me)

Dan Stowell (dan.stowell@naturalis.nl)

2023-07-18 04:46:38

Our new Doctoral Network "Bioacoustic AI" now has a website! Funded PhDs in Europe coming soon. Find out about the project: https://bioacousticai.eu/

🙌 Inês Nolasco, Sonny Burniston, Nora Gourmelon, Omiros Pantazis, Georgia Atkinson, Viktor Domazetoski, Stephanie O'Donnell, Justin Kay, Carly Batist, Yseult Hb, Subhransu Maji, Emilio Luz-Ricca, Nicolas Arrieta Larraza, Thomas Radinger

😄 Maddie Cusimano

🙌:skin_tone_4: Chris Llorca

💕 Jon Van Oast

🦇 Marius Miron

❤️ Carly Batist, Ben Williams

👍 Eelke

Arvin Sun (sunbingyou1984@gmail.com)

2023-07-19 14:21:25

A good opportunity to expose the product. Traini is looking for a number of professional service partners including pet medical, trainer, grooming, pet supplies and more.Recommend it to our users in our community by AI. We’re focusing on dogs right now. If you are DM or email to trainipet@gmail.com

:spam: Declan

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-07-19 14:32:28

Anyone else going to be at ICCB next week in Kigali? 🙋‍♀️
Let me and/or @Stephanie O'Donnell know, we’ve got a conservation tech crew started for it!

🙋 Jes Lefcourt

🎉 Jon Van Oast, Stephanie O'Donnell

Timothy Mayer (tjm0042@uah.edu)

2023-07-20 11:11:33

Hello Everyone,
I wanted to introduce myself to the channel I am Tim Mayer, Research Scientist at University of Alabama in Huntsville and the Regional Science Coordination Lead for the SERVIR Hindu Kush Himalaya region as part of SERVIR global's NASA Science Coordination Office (SCO). SERVIR is a joint initiative of NASA and USAID and we are working around the globe with local partners to develop geospatial solutions and applications to address environmental challenges. I want to thank @Ben Weinstein for connecting me to this group!

Within NASA SCO our team has a dedicated TensorFlow Working Group (TFWG) which focuses on capacity building, application development, and knowledge sharing around Deep Learning approaches. I will be sure to share future talks and updates coming from our group here.

Again really excited to join this slack channel!
Tim

sites.google.com

TensorFlow Working Group

To add the TFWG Calendar Click Here

Original URL: https://sites.google.com/uah.edu/2020-tensorflow-technical-exch/home

👋 Sara Beery, Elie Alhajjar, Avi Sundaresan, Alan Papalia, Carly Batist, Omiros Pantazis, Ben Weinstein, Jon Van Oast, Anastasios Angelopoulos, Dan Morris, Jason Holmberg (Wild Me), Andrew Schulz, Timm Haucke, Maia Adar, Jonathan Roberts, Declan, Lindsey Dukles, aruna, Nicolas Arrieta Larraza, Carl Boettiger

Patrick Beukema (patrickb@allenai.org)

2023-07-21 14:44:26

What do folks use for annotating geospatial data for computer vision (object detection)? commercial, bespoke, in house, etc? our needs are to label objects in Sentinel-1 or Sentinel-2 imagery. Scenes are large, we need to look through many channels often.

👀 Mikey Tabak, Carl Boettiger

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-07-21 16:20:27

Perhaps Smarter-labelme https://github.com/robot-perception-group/smarter-labelme can help? It is very useful for tracking objects over time for quick annotation, but you can also use it for an arbitrary set of images. Let me know if you find it useful and we (@Eric Price or I) can help you with any questions about that. The only issue could be multiple channels (other than RGB).

robot-perception-group/smarter-labelme

Video / Image Annotation (Polygon, Semantic mask, Classification) with Python and Re3

Stars

Language

Python

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-07-21 16:21:36

I recently posted about it on Linkedin -- https://www.linkedin.com/posts/ahmadaamir_robotics-labeling-annotate-activity-7086729440344322048-qAud?utm_source=share&utm_medium=member_desktop

linkedin.com

Aamir Ahmad on LinkedIn: #robotics #labeling #annotate #intelligent #opensource #ias2023…

Labeling videos and images is a laborious and boring task! It is also prone to errors. For computer scientists, especially in #robotics, #labeling such data…

Original URL: https://www.linkedin.com/posts/ahmadaamir_robotics-labeling-annotate-activity-7086729440344322048-qAud?utm_source=share&utm_medium=member_desktop

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-07-21 16:28:19

Here's a comprehensive list of annotation tools that you can use with their github links:
https://www.thinkevolveconsulting.com/list-of-open-source-annotation-tools-for-machine-learning-research/

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-07-21 16:29:55

Have you tried using the segment-geospatial or segment-Anything-eo packages. I have tried it with drone imagery and it seems to give a good output. Maybe it could be trained for Sentinel images.

Patrick Beukema (patrickb@allenai.org)

2023-07-21 17:02:51

Thanks so much! I don’t mind if it’s commercial, paid for product. But these are a great reference, thanks!

Roni Choudhury (roni.choudhury@kitware.com)

2023-07-21 18:15:45

i work for kitware inc. and am a technical lead for DIVE, an open source video/image annotation platform: https://kitware.github.io/dive/

kitware.github.io

Home - DIVE

Documentation for DIVE

Original URL: https://kitware.github.io/dive/

Roni Choudhury (roni.choudhury@kitware.com)

2023-07-21 18:15:59

others in my team have expertise in geospatial data as well

Roni Choudhury (roni.choudhury@kitware.com)

2023-07-21 18:16:48

our software is generally open source, and we partner with people to develop customizations needed for your specific workflow

Roni Choudhury (roni.choudhury@kitware.com)

2023-07-21 18:17:08

if you're interested in exploring whether we can help you, @Patrick Beukema, please do reach out via DM

Dan Morris (agentmorris@gmail.com)

2023-07-23 13:19:58

@Aakash Gupta Great list! One to add: https://github.com/mfl28/BoundingBoxEditor. It's fast and simple, and the developer has been very responsive. I'm trying to work out a workflow for doing a small number of offline annotations, and FWIW my "top two" right now are LabelMe (the offline one) and BoundingBoxEditor. But I wasn't aware of DIVE, definitely going to take a look at that too!

❤️ Roni Choudhury

👍 Aakash Gupta

Mikey Tabak (tabakma@gmail.com)

2023-07-27 07:52:17

I use CVAT, which has a free online interface. I have not checked out these other tools, but I will be looking at them next time I need to annotate.

Matthias Zuerl (matthias.zuerl@fau.de)

2023-07-24 09:30:39

Hey! I am currently conducting a literature research on video-based re-ID methods and I am specifically interested in video-based re-ID datasets for any non-human species. My colleagues and I recently published such dataset for polar bears ( https://doi.org/10.3390/ani13050801 ). Tbh, I am not able to find anything similar. Did I miss something? Do you know any video-based re-ID dataset for animals?

(I know that e.g. in the Amur Tiger dataset ( https://doi.org/10.1145/3394171.3413569 ) one can find video clips of the animals, but the way it is annotated it's not usable "out-of-the-box" for video-based re-ID. This statement is true for some other datasets which contain videos. But I am looking for something which is designed in the same way as the human benchmark datasets, e.g. Mars, iLIDS-VID ... --> sequences as samples, including movement, sorted by ID)

Any help is greatly appreciated! 🙂

🐻‍❄️ Dan Morris

👍 Mitchell Rogers

Sara Beery (sara.beery@gmail.com)

2023-07-24 09:31:25

@Peter Kulits @Chuck Stewart @Jason Holmberg (Wild Me)

🙏 Matthias Zuerl

🎉 Jon Van Oast

Peter Kulits (peterkulits@gmail.com)

2023-07-24 15:32:59

Sure, people have been using video for animal re-id for a while, though unfortunately most of it isn't public. Here's a handful of papers that release video re-id datasets that I think have the information you're looking for:

Towards Self-Supervision for Video Identification of Individual Holstein-Friesian Cattle: The Cows2021 Dataset

Visual Localisation and Individual Identification of Holstein Friesian Cattle via Deep Learning

Classification and Re-Identification of Fruit Fly Individuals Across Days With Convolutional Neural Networks

Re-Identification of Zebrafish using Metric Learning

One public, one by request: Deep-learning based identification, tracking, pose estimation, and behavior classification of interacting primates and mice in complex environments

👍 Mitchell Rogers, Malte Pedersen, Than Hitt

Malte Pedersen (mape@create.aau.dk)

2023-07-25 03:22:38

Tracking is basically frame-by-frame re-identification and you have the ID's in the ground truth files. So if you find a lack of variation in re-id video datasets of animals for benchmarking purposes maybe an option could be to use tracking sequences as a supplement? Just a quick idea.

3D-ZeF - stereo tracking dataset with zebrafish

BrackishMOT - tracking dataset from brackish water

Tracking of ants

Neha Hulkund (nhulkund@mit.edu)

2023-07-24 09:40:45

Has anyone worked with or knows of tabular data for conservation? I'm hoping to look at different applications of this, but wasn't sure if it was widely used or not?

👀 Sara Beery

Carl Boettiger (cboettig@berkeley.edu)

2023-08-02 11:51:32

yes, lots of conservation-related data is still distributed in tabular formats! e.g. the largest occurrence records, GBIF is distributed as parquet, & eBird is distributed as compressed csv. What kind of conservation data did you have in mind?

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2023-07-25 12:12:59

Hello everyone,

With my original background in ecology/conservation, I'm working on an open science project to give motion-sensing wildlife camera traps binocular vision. The goal is for camera traps to capture spatial information along with wildlife photos. This data has the potential to greatly ease estimating wildlife populations.

I've been working with some engineers from the Gathering for Open Science Hardware, and one of them has devised a 3D-printed stereo lens attachment for camera traps, while not adding to the power envelope. We hope this will be easier than modifying camera traps or building them from scratch to provide 3D vision. I know some of those here are working on stereovision algorithms for wildlife observations, and being able to directly and efficiently obtain stereo images could be helpful.

To that end, we've started a crowdfunding campaign on Experiment.com, and thanks to generous contributors we're already at 92% of our funding goal but with only 9 days left!

I'm posting this here to request your support and spread the word. Any contribution, even just $1 will help! This project may also be of interest as demonstrating a different way of funding open science. If successful, our design of the stereo lens for camera traps and any data obtained will be published as open data and open source hardware.

Let me know if you have any questions/comments, and thank you in advance for you help!

Experiment - Moving Science Forward

Helping conservation biologists with 3D glasses for wildlife camera traps

Camera traps photos tell you what animals are in an area, but figuring out how many is tricky. E.g. two foxes in a photo doesn't mean there are only two; or a camera might see the same fox twice. But, knowing how many is key to answering questions like "Is the population of a species declining?"...

Original URL: https://experiment.com/projects/helping-conservation-biologists-with-3d-glasses-for-wildlife-camera-traps

👍 Peter van Lunteren, Timm Haucke, Aida Mashkouri

😎 Timm Haucke, Dan Morris, Andrew Schulz, Mikey Tabak

Dan Stowell (dan.stowell@naturalis.nl)

2023-07-28 11:12:58

Hi folks. Method request! Do you know any work that shows how to apply machine learning to GPS tracks data, e.g. collar tags? (Any species. I'd like to find some examples to use in a course. It has to involve machine learning...)

Sara Beery (sara.beery@gmail.com)

2023-07-28 11:14:26

@Catherine Villeneuve

🙏 Dan Stowell

Oisin Mac Aodha (macaodha@caltech.edu)

2023-07-28 11:18:27

https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/2041-210X.12926

👍 Dan Stowell

Dan Stowell (dan.stowell@naturalis.nl)

2023-07-28 11:35:28

Oh I know that first-author! Nice example

Toryn Schafer (tschafer@tamu.edu)

2023-08-01 17:53:03

I'd be curious to see if other examples pop up. I would say machine learning is not popular yet with movement ecologists. Here is another example though:
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0235750

Also, if you consider MAXENT to be ML, it is frequently used with telemetry data to estimate resource selection.

journals.plos.org

Machine learning for modeling animal movement

Animal movement drives important ecological processes such as migration and the spread of infectious disease. Current approaches to modeling animal tracking data focus on parametric models used to understand environmental effects on movement behavior and to fill in missing tracking data. Machine Learning and Deep learning algorithms are powerful and flexible predictive modeling tools but have rarely been applied to animal movement data. In this study we present a general framework for predicting animal movement that is a combination of two steps: first predicting movement behavioral states and second predicting the animal’s velocity. We specify this framework at the individual level as well as for collective movement. We use Random Forests, Neural and Recurrent Neural Networks to compare performance predicting one step ahead as well as long range simulations. We compare results against a custom constructed Stochastic Differential Equation (SDE) model. We apply this approach to high resolution ant movement data. We found that the individual level Machine Learning and Deep Learning methods outperformed the SDE model for one step ahead prediction. The SDE model did comparatively better at simulating long range movement behaviour. Of the Machine Learning and Deep Learning models the Long Short Term Memory (LSTM) individual level model did best at long range simulations. We also applied the Random Forest and LSTM individual level models to model gull migratory movement to demonstrate the generalizability of this framework. Machine Learning and deep learning models are easier to specify compared to traditional parametric movement models which can have restrictive assumptions. However, machine learning and deep learning models are less interpretable than parametric movement models. The type of model used should be determined by the goal of the study, if the goal is prediction, our study provides evidence that machine learning and deep learning models could be useful tools.

Original URL: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0235750

Timothy Mayer (tjm0042@uah.edu)

2023-07-28 13:10:08

For all those planning to attend AGU 2023, and looking for an exciting session please checkout out and submit an abstract to GC018 - Applications of Machine Learning and Deep Learning to address environmental challenges and Sustainable Development Goals session. Abstracts due on 8/2 :)

AGU - AGU23

Applications of Machine Learning and Deep Learning to address environmental challenges and Sustainable Development Goals

Over the past decade, there has been a significant rise in the use of Machine Learning (ML), Deep Learning (DL), and Artificial Intelligence (AI) through the application of Earth observations, including groundbreaking efforts through foundational model development. These technologies have been used to address environmental challenges, such as monitoring air quality, improving precision agriculture, and providing near real time deforestation alerts. This session aims to showcase innovative research and applications of these technologies in data-sparse regions. The goal is to address environmental development challenges while promoting open data access, cloud computing, capacity-building efforts, and data democracy. The session intends to bring together experts, non-profits, and applied scientists to promote collaboration in leveraging these technologies to create positive impacts in developing regions and across the globe.

Original URL: https://agu.confex.com/agu/fm23/prelim.cgi/Session/189389

Michael Bunsen (notbot@gmail.com)

2023-07-28 14:39:13

😎 Jason Holmberg (Wild Me), Shir Bar, Carly Batist, Sara Beery

👍 Jon Van Oast, Timothy Mayer, Dan Morris, Maddie Cusimano

🤩 Carly Batist, Thomas Radinger

Dan Morris (agentmorris@gmail.com)

2023-07-31 10:34:55

New dataset on LILA, thanks to the amazing work being done at The Cacophony Project to manage invasive predators in New Zealand:

https://lila.science/datasets/new-zealand-wildlife-thermal-imaging/

This is LILA's first thermal camera trap dataset... I feel like even just a couple years ago, thermal cameras were a concept one might dream of deploying if you had $20,000 and a best friend in the CIA. Now they're becoming quite practical (in part thanks to The Cacophony Project's sibling organization, 2040), and thermal images open up a whole new interesting world of AI. As much as I like to throw MegaDetector at everything, detection is basically a non-issue with thermal sensors. Species classification is... different than it is with optical sensors. You have some information in the pixels, but at least as much information in the movement trajectory. So, for AI folks who want to try something that there aren't already 1000000 papers about, take a look at this dataset!

Example video:

https://storage.googleapis.com/public-datasets-lila/nz-thermal/videos/1486055.mp4

@Giampaolo Ferraro @matthew hellicar

👏 Malte Pedersen, Fagner Cunha, Dante Wasmuht, Cameron Trotter, Elie Alhajjar, Timm Haucke, Jason Holmberg (Wild Me), Alan Papalia, Juliana Gomez Consuegra, Aakash Gupta, Felipe Montealegre-Mora, Sara Beery, David Will

👍 Justin Kay, Valentin Gabeff, Benjamin Kellenberger, Elie Alhajjar, Jonathan Roberts, Timm Haucke, Peter van Lunteren, Jason Holmberg (Wild Me), Matthias Zuerl, Vincent Christlein, Thor Veen, Paul Melki

🙌 Carly Batist, Elie Alhajjar, Timm Haucke, Jason Holmberg (Wild Me)

🔥 Oisin Mac Aodha, Elie Alhajjar, Elizabeth Campolongo, Shir Bar, Timm Haucke, Rowan Converse, Peter van Lunteren, Jason Holmberg (Wild Me), Taiki Sakai - NOAA Affiliate, Ghazi Randhawa, Maddie Cusimano, Carl Boettiger, Josh Veitch-Michaelis

😎 Jon Van Oast, Jason Holmberg (Wild Me), Michael Bunsen

:flag_nz: Mitchell Rogers

👏:skin_tone_3: Pen-Yuan Hsing

Timothy Keitt (tkeitt@utexas.edu)

2023-07-31 18:11:26

Hi Folks, I am Timothy Keitt from the University of Texas at Austin. I just joined the slack. We're interested in automated monitoring of biodiversity. I'll be at ESA (Mon-Wed) if anyone wants to chat in person. All the best.

sites.cns.utexas.edu

Ecology in the age of automation

Keitt TH, Abelson E. Ecology in the age of automation. Science [Internet]. 373 (6557) :858-859.

Original URL: https://sites.cns.utexas.edu/keittlab/publications/ecology-age-automation

👋 Sara Beery

Jacob Marks (jamarks13@gmail.com)

2023-08-02 14:29:48

Hey everyone!

My name is Jacob and I'm an ML Engineer and Developer Evangelist at Voxel51.

I run an industry spotlight blog series where I focus on how computer vision is being utilized in different industries. So far I have done Agriculture, Manufacturing, and Healthcare is coming out in 2 weeks).

Next up on my list is Climate and Conservation!!

Would love to chat with anyone working in the space, both in academia and industry. If you are open to speaking about your work, feel free to reach out to me either here on Slack, via email at jacob@voxel51.com, or on LinkedIn — would love to learn about what you're working on!!

Best,
Jacob

😎 Jon Van Oast, Jason Holmberg (Wild Me), Sara Beery, Aakash Gupta, Nora Gourmelon, Amara McCune

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-08-03 07:12:58

Very happy to announce our latest work in the direction of inferring animal behavior automatically from drone videos. Not only this, we also release a never-seen-before kind of dataset of zebras (both plains and Grévy's) recorded in their natural environment in Kenya using two synchronized drones.

Code: https://github.com/robot-perception-group/animal-behaviour-inference
Paper: https://www.biorxiv.org/content/10.1101/2023.07.31.551177v1
Dataset: https://keeper.mpdl.mpg.de/d/a9822e000aff4b5391e1/
Video: https://youtu.be/Zu-t0JJsz5o

Feel free to reach out to me about it.

bioRxiv

A Framework for Fast, Large-scale, Semi-Automatic Inference of Animal Behavior from Monocular Videos

An automatic, quick, accurate, and scalable method for animal behavior inference using only videos of animals offers unprecedented opportunities to understand complex biological phenomena and answer challenging ecological questions. The advent of sophisticated machine learning techniques now allows the development and implementation of such a method. However, apart from developing a network model that infers animal behavior from video inputs, the key challenge is to obtain sufficient labeled (annotated) data to successfully train that network - a laborious task that needs to be repeated for every species and/or animal system. Here, we propose solutions for both problems, i) a novel methodology for rapidly generating large amounts of annotated data of animals from videos and ii) using it to reliably train deep neural network models to infer the different behavioral states of every animal in each frame of the video. Our method's workflow is bootstrapped with a relatively small amount of manually-labeled video frames. We develop and implement this novel method by building upon the open-source tool Smarter-LabelMe, leveraging deep convolutional visual detection and tracking in combination with our behavior inference model to quickly produce large amounts of reliable training data. We demonstrate the effectiveness of our method on aerial videos of plains and Gr&evy's Zebras (Equus quagga and Equus grevyi). We fully open-source the code (<https://github.com/robot-perception-group/animal-behaviour-inference>) of our method as well as provide large amounts of accurately-annotated video datasets (<https://keeper.mpdl.mpg.de/d/a9822e000aff4b5391e1/>) of zebra behavior using our method. A video abstract of this paper is available at (<https://youtu.be/Zu-t0JJsz5o>). ### Competing Interest Statement The authors have declared no competing interest.

Original URL: https://www.biorxiv.org/content/10.1101/2023.07.31.551177v1

KEEPER

Detection_and_Behaviour_Classification_Zebras_2022

Share link for Detection_and_Behaviour_Classification_Zebras_2022.

Original URL: https://keeper.mpdl.mpg.de/d/a9822e000aff4b5391e1/

YouTube

Aamir Ahmad (https://www.youtube.com/@KgpianAsimov)

Animal Behavior Inference Framework based on Smarter-Labelme

Original URL: https://youtu.be/Zu-t0JJsz5o

robot-perception-group/animal-behaviour-inference

Animal Behavior Inference Framework based on Smarter-Labelme

Language

Python

Last updated

6 days ago

🙌 Jonathan Roberts, Ștefan Istrate, Matthias Zuerl, Sara Beery, Shir Bar, Enis Berk Çoban, Felipe Parodi, Yseult Hb, Timm Haucke, Toryn Schafer, Dan Morris, Aakash Gupta, Juliana Gomez Consuegra, Anastasios Angelopoulos, mimi, Felipe Montealegre-Mora, Jacob Marks, Lingchao Mao, Aleksis Pirinen

:zebra_face: Blair Costelloe, Cameron Trotter, Taiki Sakai - NOAA Affiliate, Mitchell Rogers, Andrew Schulz, Jacob Marks

👏 Katie Zacarian, Steve Haddock, Urs

👍 Hsun-Yi Hsieh, Jon Van Oast, Omiros Pantazis, Valentin Gabeff

🙏 Anastasios Angelopoulos

🎉 Pen-Yuan Hsing

Blair Costelloe (blaircostelloe@gmail.com)

2023-08-03 07:29:49

Awesome! Dan Rubenstein was telling me about this in Anchorage a couple of weeks ago. Do you think it can work in nadir footage of zebras, or is the oblique viewpoint necessary?

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-08-03 07:47:03

Thanks. The network as is will most likely need a small bit of fine tuning, but that is what this paper is all about.. with a small number of manual annotations you can kick start the whole process and have the network tuned for your dataset. If you want, you could send us a few pictures to try out quickly.

Blair Costelloe (blaircostelloe@gmail.com)

2023-08-03 07:54:17

If it's not too much trouble it would be really interesting to know how it works OOTB with my data. I'll send a few example frames to your email. Thanks!

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-08-03 07:54:46

No prob! Looking forward to it.

Dan Morris (agentmorris@gmail.com)

2023-08-03 12:55:09

Really neat dataset! I'm able to click through to some of the individual images, but if I try to download the dataset (or one of the top-level folders), I get "size too large". I don't know how Keeper works... do you have a recommendation for downloading the dataset?

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-08-03 14:12:02

Thanks! There is a server configuration limitation apparently (as the folders are about ~100GB). I am trying to get it resolved through the MPI admins. In the meantime, I will figure out another way to share it with you directly.

Dan Morris (agentmorris@gmail.com)

2023-08-03 14:50:22

Thanks! No rush to share with me directly... I'm just wearing my dataset-list-making hat, and part of that is making sure datasets I add to various lists are downloadable. I'll be your first download beta-tester when you get it all worked out.

👍 Aamir Ahmad, Jason Holmberg (Wild Me)

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2023-08-03 15:24:39

I would love to get a copy too. Thank you!

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-08-30 10:40:05

Hi @Dan Morris and @Jason Holmberg (Wild Me).. we solved this download issue with a workaround. There is now a zipped version of each folder in the root folder. You can download the zips individually. Downloading the folder was not working apparently because of some buffer size limit issue, but now you do not need to download the folder. However, I leave the folders also there so people can browse through individual files if they want. Let me know if there is still some issue.

Dan Morris (agentmorris@gmail.com)

2023-09-01 21:21:32

Thanks! This is straightforward to work with; I confirmed I could read the annotations and render boxes on a sample image, then added to the running list of aerial/drone wildlife datasets:

https://github.com/agentmorris/agentmorrispublic/blob/main/drone-datasets.md#large-scale-semi-automatic-inferen[…]mal-behavior-from-monocular-videos

While I was there, I also added another one from the backlog, from Koger et al, ~40k boxes on ungulates and geladas in drone images:

https://github.com/agentmorris/agentmorrispublic/blob/main/drone-datasets.md#quantifying-the-movement-behaviour[…]s-using-drones-and-computer-vision

😎 Jason Holmberg (Wild Me)

:zebra_face: Blair Costelloe, Aamir Ahmad

👍 Aamir Ahmad

Carl Boettiger (cboettig@berkeley.edu)

2023-08-09 13:51:05

Hi friends -- what's the state of the art these days for camera trap metadata? Do people use schema.org or EML, or other custom formats?

Jinsu Elhance (jelhance@gmail.com)

2023-08-09 16:49:37

Hi Carl! Just pinging @Nathaniel Rindlaub for any insights?

👋 Carl Boettiger

Dan Morris (agentmorris@gmail.com)

2023-08-09 18:20:38

I'm aware of two standards... the Camera Trap Metadata Standard (CTMS):

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5267527/

...was used by eMammal; I'm not aware of current systems that use this as a native format.

If I were putting my eggs in one basket, it would be the newer Camtrap DP standard:

https://tdwg.github.io/camtrap-dp/

...which (according to the page, I've not personally verified this) is exported by Agouti and Trapper, and supported as an input format for GBIF.

Approximately 7000 people, approximately 3000 of whom are on this Slack, wrote a paper about it:

https://ecoevorxiv.org/repository/view/5593/

Given that many of those authors were also authors on the original CTMS standard, I think it's safe to declare the CTMS deprecated.

PubMed Central (PMC)

An Open Standard for Camera Trap Data

Original URL: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5267527/

tdwg.github.io

Camtrap DP

Data exchange format for camera trap data

Original URL: https://tdwg.github.io/camtrap-dp/

🙌 Carl Boettiger

💚 Carl Boettiger

Riley Knoedler (mknoedler@west-inc.com)

2023-08-09 15:57:29

Hi folks, I'm Riley Knoedler, I heard about the slack at CVPR and I'm very excited to join the community! I'm a Data Scientist at Western EcoSystems Technology, and most of my team's work focuses on developing machine learning solutions for monitoring wildlife interactions with renewable energy facilities.

👏 Jon Van Oast, Jacob Marks

Ben Weinstein (benweinstein2010@gmail.com)

2023-08-09 16:38:11

Welcome to the community, what would you say the largest challenge your team faces? Model development? annotations? We’ve done some wind farm bird monitoring work, just curious.

Riley Knoedler (mknoedler@west-inc.com)

2023-08-10 10:51:54

Probably sufficient annotations and distinguishing near misses from collisions. What kind of challenges has your team faced?

Ben Weinstein (benweinstein2010@gmail.com)

2023-08-10 14:02:15

We are on surveys before active energy development, so we have to cover large areas to look for birds in proposed wind areas. Large data sizes and very long species list.

Riley Knoedler (mknoedler@west-inc.com)

2023-08-10 14:18:24

That's really interesting, what are your survey methods? Are you using cameratraps looking up at the sky or drones or something else?

Jinsu Elhance (jelhance@gmail.com)

2023-08-09 16:51:07

Hey all, curious if anyone has come across a tool that allows you to view very high resolution true color imagery and visually estimate fractional cover as an input to a remote sensing model that maybe uses coarser imagery?

Sara Beery (sara.beery@gmail.com)

2023-08-09 16:58:53

@Dan Morris

Dan Morris (agentmorris@gmail.com)

2023-08-09 18:15:18

I don't know of any tools that do this, though you could certainly use existing high-resolution (let's say 1m) tree cover maps as training data for a system you'll eventually run with coarser (10m?) imagery. Consider the Chesapeake Land Cover dataset:

https://lila.science/datasets/chesapeakelandcover

...which is really an older version of:

https://www.chesapeakeconservancy.org/conservation-innovation-center/high-resolution-data/lulc-data-project-2022/

...or the NOAA Coastal Change Analysis Program's high-resolution map:

https://catalog.data.gov/dataset/coastal-change-analysis-program-c-cap-high-resolution-land-cover-and-change-data

That's a lot of geographic bias, but it's a place to start. I'm not aware of other large (i.e., larger than one city) high-resolution land cover or tree cover datasets, though I don't claim that I would know of others.

LILA BC

Chesapeake Land Cover - LILA BC

Overview This dataset contains high-resolution aerial imagery from the USDA NAIP program [1], high-resolution land cover labels from the Chesapeake Conservancy [2], low-resolution land cover labels from the USGS NLCD 2011 dataset [3], low-resolution multi-spectral imagery from Landsat 8 [4], and high-resolution building footprint masks from Microsoft Bing [5], formatted to accelerate machine learning research...

Written by

lilawp

Est. reading time

6 minutes

Original URL: https://lila.science/datasets/chesapeakelandcover

Chesapeake Conservancy

Chesapeake Bay Program Land Use/Land Cover Data Project

Est. reading time

4 minutes

Original URL: https://www.chesapeakeconservancy.org/conservation-innovation-center/high-resolution-data/lulc-data-project-2022/

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-08-10 10:14:40

check out the Segment-Anything-model. you can model it to annotate the tree-area or building foot-print and then pipe it to your model with lower resolution imagery

Ruben Remelgado (ruben.remelgado@gmail.com)

2023-08-11 05:40:05

https://www.nature.com/articles/s41586-018-0411-9

A worthwhile read on this topic. You can (and should) replace modelling components, but the workflow is still valid.

Nature

Global land change from 1982 to 2016

Nature - Satellite data for the period 1982–2016 reveal changes in land use and land cover at global and regional scales that reflect patterns of land change indicative of a human-dominated...

Original URL: https://www.nature.com/articles/s41586-018-0411-9

Riley Knoedler (mknoedler@west-inc.com)

2023-08-10 14:33:15

Has anyone ever leveraged the computer vision models used by iNaturalist or eBird to process large amounts of data locally? I'm trying to figure out if that is a possible / permissible use case.

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-08-10 14:53:19

@gvanhorn ?

Oisin Mac Aodha (macaodha@caltech.edu)

2023-08-10 16:19:01

There is a good performing model training on the iNat2021 competition dataset now available on HuggingFace. Worth checking out, there is also an online demo on the right of this page:
https://huggingface.co/timm/eva02_large_patch14_clip_336.merged2b_ft_inat21

huggingface.co

timm/eva02_large_patch14_clip_336.merged2b_ft_inat21 · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Original URL: https://huggingface.co/timm/eva02_large_patch14_clip_336.merged2b_ft_inat21

👍 Riley Knoedler, Dan Morris, Sara Beery, Emilio Luz-Ricca

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-08-11 16:30:07

Hi all, we are now back in the Hungarian Steppe to deploy and test our autonomous blimps for biodiversity monitoring (here Przewalski's horses in particular). Feel free to follow updates regarding this on my LinkedIn page (https://www.linkedin.com/posts/ahmadaamir_autonoumous-blimp-day1-activity-7095860403892559872-AZJj?utm_source=share&utm_medium=member_desktop)

linkedin.com

Aamir Ahmad on LinkedIn: #autonoumous #blimp #day1 #przewalski #biodiversity #monitoring

Flight Robotics and Perception Group is back to the Hungarian Steppe! And this time we are here with our #autonoumous #blimp . #day1 of the mission was spent…

Original URL: https://www.linkedin.com/posts/ahmadaamir_autonoumous-blimp-day1-activity-7095860403892559872-AZJj?utm_source=share&utm_medium=member_desktop

👀 Pen-Yuan Hsing, Yseult Hb, Alasdair Davies

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-08-13 12:04:38

https://arxiv.org/abs/2201.11192

arXiv.org

ReforesTree: A Dataset for Estimating Tropical Forest Carbon Stock with Deep Learning and Aerial Imagery

Original URL: https://arxiv.org/abs/2201.11192

👍 Aamir Ahmad, Sara Beery, Timm Haucke, Björn Lütjens

Caleb Robinson (calebrob6@gmail.com)

2023-08-13 14:28:54

PyTorch Dataset available in TorchGeo -- https://torchgeo.readthedocs.io/en/latest/api/datasets.html#reforestree

🙌 Carly Batist

Emily Lines (erl27@cam.ac.uk)

2023-08-14 06:09:08

Hmmm... this is a nice idea but given the known high levels of uncertainty of applying general allometric equations (particularly in the tropics, and particularly for equations without height as an input), I hope others would also search out e.g. TLS or destructively sampled ground truth datasets, which are often available. It would also be great to see these datasets attempt to compute the uncertainties in their biomass measurements.

I do find it disappointing that the title discusses 'tropical forest carbon stock' when this is very specifically a small scale dataset of regional tropical agro-forestry. Monitoring agro-forestry for NBS payments is a very worthy aim in itself, and I think it would help the community progress if manuscripts/datasets were more specific about their scope...

👍 Björn Lütjens, stefano puliti

Björn Lütjens (bjoern.luetjens@gmail.com)

2023-08-26 12:54:56

Hi emily, I was part of the paper and unfortunately it's too late to rename but I agree that agroforestry would have been a more appropriate title! good point thank you

Ben Weinstein (benweinstein2010@gmail.com)

2023-11-06 16:36:40

Following up here, were all trees intended to be annotated? Or just ground truth trees with biomass? https://github.com/gyrrei/ReforesTree/issues/5

#5 Confirm annotation overlap? Are all trees annotated?

I am the mantainer of the DeepForest (<a href="https://deepforest.readthedocs.io/en/latest/">https://deepforest.readthedocs.io/en/latest/</a>) tree detection backbone. I'd like to add these annotations as pretraining for our algorithm. I downloaded the dataset, loaded mapping/final_dataset.csv and overlaid them on the matching names in the /tiles directory. I can't tell if the annotations were correctly read in. In orange are the ReForestTree ground truth, in blue are DeepForest predictions, they can be ignored, except to show that there are unlabeled trees in the images that lack annotations? Were all trees intended to be annotated? Let me know if I'm misunderstanding how to use the dataset. <a href="https://user-images.githubusercontent.com/1208492/280852410-5618dc42-e0cc-451a-89cf-7b75b9bd276d.png">Screenshot 2023-11-06 at 1 33 40 PM</a>

Patrick Beukema (patrickb@allenai.org)

2023-08-14 11:05:32

Hi all, we hosted David Rolnick a few weeks ago in our Environmental AI series. He delivered an inspired talk on applying ML for climate action. Among his many incisive remarks: “AI for Good” doesn’t mean just adding new “good” applications of AI. It means shaping all applications of AI to be better for society.
Check it out!
https://www.youtube.com/watch?v=rs4MpjNxOLQ

YouTube

Allen Institute for AI (https://www.youtube.com/@allenai)

Machine Learning in Climate Action

Original URL: https://www.youtube.com/watch?v=rs4MpjNxOLQ

❤️ Caleb Robinson, Katelyn Morrison, Ditiro Rampate, Juliana Gomez Consuegra, Aakash Gupta, Jason Holmberg (Wild Me), Yseult Hb, Emilio Luz-Ricca, Michael Bunsen

🎉 Michael Bunsen

Andrew Schulz (akschulz@gatech.edu)

2023-08-16 10:05:21

Hi All! After a long year of writing, editing, and proofing with the amazing @Suzanne Stathatos and more! We have our perspectives piece out on conservation tools (this is not particularly for people in this group, but I think everyone can learn something from the paper)...This paper more serves as a starter guide on some of the new techniques, tools, and technologies used for conservation. We include a vocab list of some of the common words thrown around as well as a case study highlighting some of the excellent work shared throughout this group. Hope you enjoy it! Check out the paper here: https://doi.org/10.1098/rsif.2023.0232

🎉 Avi Sundaresan, Elie Alhajjar, Elizabeth Bondi-Kelly, Sara Beery, Devis Tuia, Rowan Converse, Yuanqi Du, Timm Haucke, Omiros Pantazis, Taiki Sakai - NOAA Affiliate, Nico Lang, Michael Procko, Jon Van Oast, Shir Bar, Anastasios Angelopoulos, Sam Lapp, Yseult Hb, Dan Morris, Jason Holmberg (Wild Me), Emilio Luz-Ricca

❤️ Suzanne Stathatos, Maddie Cusimano, Taiki Sakai - NOAA Affiliate, Eric Greenlee, Anastasios Angelopoulos, Justine Boulent, Pranav Khandelwal, Elie Alhajjar, Alison Ketz, Pen-Yuan Hsing, Jason Holmberg (Wild Me), Talia Speaker

👍 Emily Lines, Valentin Gabeff, Anastasios Angelopoulos, Casey Youngflesh, Elie Alhajjar, Aamir Ahmad

charlotte (deshchang@gmail.com)

2023-08-16 14:58:02

Hi everyone! I’m Charlotte Chang, and it’s great to be here! I’m an Assistant Professor at Pomona College, and my work has used natural language processing to look at conservation communications on social media. I’d love to chat with anyone who is interested in NLP and conservation (communications). I’ve also used bioacoustics and CV models in my (undergraduate) teaching.

👋 Devis Tuia, Toryn Schafer, Ankita Shukla, Andrew Schulz, Sara Beery, Omiros Pantazis, Abhay, Dan Morris, Emerson de Lemmus, Valentin Gabeff, Emilio Luz-Ricca, Talia Speaker, Lauren Harrell

👍 Bistra Dilkina, Aakash Gupta, Emerson de Lemmus, Jason Holmberg (Wild Me)

👏 Katie Zacarian, Emerson de Lemmus, Tom Wye (Fishial.ai), Jason Holmberg (Wild Me)

👍:skin_tone_3: Pen-Yuan Hsing

🙌 Maia Adar, Jason Holmberg (Wild Me)

Taiki Sakai - NOAA Affiliate (taiki.sakai@noaa.gov)

2023-08-16 15:12:14

Hi Charlotte! I don't do any NLP work, but I went to Harvey Mudd and work in bioacoustics 🐋

💯 charlotte

❤️ charlotte

Abhay (abhaykash12@gmail.com)

2023-08-16 17:40:09

Hi Charlotte! My background is primarily in NLP and I've done some work in the AI x Conservation space (cv). Happy to chat more if I you think i'd be a useful resource! :)

💯 charlotte

👋 charlotte

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-08-21 04:03:34

Hi Guys, just wanted to update on our fantastic field trip!

Our airship was able to visually track and autonomously follow Przewalski's horses roaming free in the Steppe. In the video below, we show how the blimp 'fights' the wind to keep the horses centered in its field of view. We are a step closer to do long term monitoring with autonomous robots. Some more info in the post here: https://www.linkedin.com/posts/ahmadaamir_hungarian-steppe-airship-activity-7099298534696312832-EC-k?utm_source=share&utm_medium=member_desktop . Congrats @Eric Price and @Pranav Khandelwal for an awesome teamwork this summer!

linkedin.com

Aamir Ahmad on LinkedIn: #hungarian #steppe #airship #autonomously #biodiversity #monitoring…

What a fantastic field trip! Thanks to the excellent work of my team (Eric Price and Pranav Khandelwal and others back home) and colleagues from the Hungarian…

Original URL: https://www.linkedin.com/posts/ahmadaamir_hungarian-steppe-airship-activity-7099298534696312832-EC-k?utm_source=share&utm_medium=member_desktop

👍 Eric Price, Jason Holmberg (Wild Me), Urs, Aakash Gupta, Sara Beery, Timm Haucke, Emilio Luz-Ricca, Yuru Jia, Rebecca Wilks

🙌 Blair Costelloe, Jason Holmberg (Wild Me), Timm Haucke, Yseult Hb

Andrew Schulz (akschulz@gatech.edu)

2023-08-22 03:41:26

Hi All - I did not know where to put this, but I wanted to highlight an incredible ecology researcher Dr. Gabriela Palomo. Specifically wanted to link everyone to her website (https://gabspalomo.github.io/). Why am I putting this in the AI for Conservation slack, well I know many of us use drones, camera traps, GIS, etc. and she has a bunch of incredible silhouettes (https://gabspalomo.github.io/silhouettes.html) that are under a Creative Commons Attribution-NonCommercial 3.0 Unported license (see example)! Happy science-ing!

😍 Leonardo Viotti, Yseult Hb, Jason Holmberg (Wild Me), Timm Haucke, Juni Rechter, Björn Lütjens, Talia Speaker

👍 mimi, Jason Holmberg (Wild Me), Rowan Converse, Timm Haucke, Juni Rechter

Pietro Perona (perona@caltech.edu)

2023-08-22 14:31:16

A take-home message I get from the Kerner-Nakelembe presentation is that (a) crop yields is a crucial issue, (b) people are unclear on how to impact yields (the factors are clear: land fractioning, use of fertilizers, use of well selected seeds etc. but affecting those factors requires changes in society, in the economy etc.). I could not tell if it is possible to estimate crop yields (and thus track progress) from satellite images.

Eric Price (eric.price@ifr.uni-stuttgart.de)

2023-08-23 02:06:24

From what I understand, the relation between sat image and yield is highly crop specific, so some reference ground data is necessary to calibrate the estimation for a specific type of crop, but there's quite a lot of work on that topic: https://scholar.google.com/scholar?q=related:2qpIflWO5X8J:scholar.google.com/&scioq=estimating+crop+yields+from+satellite+images&hl=en&as_sdt=0,5

Maddie Cusimano (maddie@earthspecies.org)

2023-08-23 11:39:50

Hey everyone, Earth Species Project is partnering with the Footprint Coalition to provide grants for research at the intersection of ML and interspecies communication, including applications to conservation. This program will issue fast-grants of up to $10,000 to researchers to do work in this domain, covering anything from dataset design to processing for machine learning, field studies of interspecies communication or developing new ways to investigate non-human signals.

A detailed description of this funding program and instructions for applying can be found on the Experiment.com website here. Students, postdocs and researchers outside of universities are all encouraged to apply.

The deadline is currently Sept 22, with rolling admissions. Please get in touch if you have questions 😊

👀 Yseult Hb, Ankita Shukla, Talia Speaker

😎 Jason Holmberg (Wild Me)

❤️ Marius Miron

Atriya Sen (atriya@atriyasen.com)

2023-08-24 16:59:20

Hello folks, I have a research assistantship position available for an MS or PhD student (new admission) in computer science at the University of New Orleans.

The position will be initially funded by a recent National Science Foundation award, details of which (including an abstract of the project) may be found here:
NSF Award Search: Award # 2246032 - CRII: III: Explainable Artificial Intelligence for Biodiversity Science & Conservation

I'm relatively flexible about the research agenda though.

The position comes with full tuition coverage, benefits, and a monthly stipend of $2000. Please email me at asen@uno.edu with the following documents in a single PDF file:

Academic transcripts,

Half-page statement of interest.

The position is also advertised here: https://www.uno.edu/academics/cos/computer-science/open-positions-in-computer-science

Katelyn Morrison (kcmorris@andrew.cmu.edu)

2023-08-24 17:11:06

Hi is it okay if I share this in the Explainable AI slack channel that I am a part of??

Katelyn Morrison (kcmorris@andrew.cmu.edu)

2023-08-24 17:11:46

I'm currently a PhD student but super interested in the project topic. Would love to chat sometime about what you are working on 🙂

❤️ Sara Beery

Atriya Sen (atriya@atriyasen.com)

2023-08-24 17:35:10

Please do share; thanks!

Devis Tuia (devis.tuia@epfl.ch)

2023-08-25 02:46:57

Is there an xAI channel? 😮

Katelyn Morrison (kcmorris@andrew.cmu.edu)

2023-08-25 09:56:47

There is @Devis Tuia https://join.slack.com/t/explainableaiworld/shared_invite/zt-21xy5r0wm-N4XjItBxoCsExFDsWFdvVw

👀 Atriya Sen

Irina Tolkova (itolkova@g.harvard.edu)

2023-08-28 09:19:33

Hi everyone! Two upcoming postdoctoral fellowship application deadlines:

1. The first is the Christopher W. Clark Postdoctoral Fellowship in Conservation Bioacoustics at the K. Lisa Yang Center for Conservation Bioacoustics at the Cornell Lab (revised deadline October 15). The selected candidate will be the inaugural recipient of this three-year fellowship. Candidates will propose innovative and independently developed bioacoustics applications in service of conservation, honoring the legacy of the fellowship’s namesake and program’s founding director, Christopher Clark. While Chris Clark primarily focused on baleen whales, scientists at the Yang Center apply bioacoustic methods for studying marine mammals, fishes, birds, primates, frogs, bats, elephants, insects, anthropogenic sounds, and more in ecosystems around the world! Yang Center staff collect and synthesize data, develop hardware and software tools, and work with partners and students around the world to facilitate the use of bioacoustics in conservation. We encourage proposals with bold, potentially transformative research potential, and especially welcome projects that leverage the tools or research developed at the Center to have an impact on biodiversity conservation (e.g., tool development, knowledge generation, implementation etc.).

In addition to salary support and annual raises, applicants can anticipate approximately $10K/year in research support, with the possibility of additional funds through collaboration with ongoing projects at the Center. For the full posting and to apply, please visit: https://academicjobsonline.org/ajo/jobs/24693.

In the cover letter, applicants should suggest one or more members of the Center as a potential postdoctoral advisor. This person is not required to be a direct research collaborator, but will provide administrative support, professional guidance, and a connection into the Yang Center and Lab of Ornithology. Any member of the Center can serve as a secondary or co-mentor. Primary advisors can include: Dena Clink, Ben Gottesman, Daniela Hedwig, Holger Klinck, Aaron Rice, Larissa Sugai, Laurel Symes, and Connor Wood. Please reach out to us with questions!

2. The second upcoming fellowship is the Edward W. Rose postdoctoral fellowship (deadline September 9). This fellowship supports postdoctoral research at the Cornell Lab of Ornithology. We encourage you to learn more about this fellowship and to share the information with your colleagues and research networks. As part of the Cornell Lab of Ornithology, the K. Lisa Yang Center for Conservation Bioacoustics can host Rose postdoctoral fellows who are conducting research that is synergistic with the research conducted by the Yang Center, even if fellows are working on taxa other than birds.

Candidates are welcome to apply to both positions, while noting that the Yang Center Clark Fellowship is particularly focused on bioacoustics and conservation, and that the Rose Fellowship supports positions across the Lab of Ornithology, is open to topics beyond bioacoustics, and can have a focus on basic or applied research.

If your research advances bioacoustics research and conservation, we encourage you to contact us and apply to either or both programs!

Cornell Chronicle

New $1.5M fellowship honors bioacoustics pioneer | Cornell Chronicle

A new $1.5 million gift from philanthropist K. Lisa Yang ’74 has established the Christopher W. Clark Postdoctoral Fellowship in Conservation Bioacoustics in honor of Clark, the retired director of the bioacoustics program at the Cornell Lab of Ornithology.

Original URL: https://news.cornell.edu/stories/2021/12/new-15m-fellowship-honors-bioacoustics-pioneer

academicjobsonline.org

Cornell University, Lab of Ornithology

Full service online faculty recruitment and application management system for academic institutions worldwide. We offer unique solutions tailored for academic communities.

Original URL: https://academicjobsonline.org/ajo/jobs/24693

Birds, Cornell Lab of Ornithology

Rose Postdoctoral Fellowship: Annual Competition

Quick Links Cornell Lab programs Lab researchers The Cornell Lab of Ornithology invites applications for our Edward W. Rose Postdoctoral Fellowships. These competitive postdoctoral fellowship awards support innovative, independent research by early career scholars of exceptional promise.

Original URL: https://www.birds.cornell.edu/home/rose-postdoctoral-fellowship-annual-competition/

👍 Holger Klinck, Subhransu Maji, Ian Ingram

😍 Sara Beery, Andrew Schulz

🐋 Taiki Sakai - NOAA Affiliate

slackbot

2023-08-28 23:55:54

This message was deleted.

:spam: Atul Ingle, Declan

Justin Kay (justinkay92@gmail.com)

2023-08-29 01:17:42

Hey @Arvin Sun this is a bit spammy and not relevant to this community. Can you please take the post down?

Sara Beery (sara.beery@gmail.com)

2023-08-29 01:33:15

I deleted

👍 Justin Kay, Jason Holmberg (Wild Me), Katelyn Morrison, Declan, Shir Bar

❓ Jon Van Oast

Sara Beery (sara.beery@gmail.com)

2023-08-29 01:33:33

Thanks @Justin Kay for the heads up

👍 Justin Kay, Jason Holmberg (Wild Me), Katelyn Morrison, Shir Bar

Casey Colson (caseycolson24@gmail.com)

2023-08-29 23:24:58

Hi all - by your judgment, what proportion of papers/projects cited in this slack channel refer terrestrial species vs. ocean species? Just curious. Anecdotal guesses are fine. Tx.

Cameron Trotter (cater@bas.ac.uk)

2023-08-30 04:39:52

Unsure I'd be able to give a ball park figure, but marine-focused research makes up a relatively small proportion of work in this domain.

Anecdotally, I am in the process of writing up a literature review of CV applications to benthic environments, and when I discuss future direction, a lot of the papers I talk about focus on terrestrial data rather than from other marine environments. This is mostly due to the higher abundance of terrestrial data allowing for greater research scope.

It's relatively cheap to set up some land-based camera traps and leave them in an area for a few months, but marine data can be expensive to obtain, requires a large human effort, and often needs specialist equipment, especially at depth.

👍 Eric Colson

Levi Cai (levi@levicai.com)

2023-08-30 11:02:18

Agreed with above. In terms of number of labelled images that can be used to train detectors for instance, terrestial has several ordered of magnitude more publicly available labelled data.

Levi Cai (levi@levicai.com)

2023-08-30 11:03:42

And even then, quite a bit of the available ocean data is surface-based imagery of animals like whales, turtles, and sharks/rays.

👍 Eric Colson

Patrick Beukema (patrickb@allenai.org)

2023-08-30 11:57:08

Fascinating discussion for those of us working in the marine space. We get asked all the time “how many fisheries are unsustainably fished” even just a simple count of “how many fish are there” is incredibly challenging. There is an older article from the Atlantic on the topic: http://webcache.googleusercontent.com/search?q=cache:https://www.theatlantic.com/science/archive/2016/10/how-many-fish-are-in-the-sea/502937/ There is a quote in there which stuck with me:

“1970s with John Shepherd, a fisheries management specialist at England’s University of Southampton: Counting fish is like counting trees, but the trees are invisible and constantly on the move.”

This topic is really critical to us, and we would like to do a better job at answering these types of questions, and more broadly linking our work (Skylight) to these metrics where possible. I spoke to Chris Wilcox recently about this question who has done a ton of work in this space. He’s not in this slack group, but he would be a great person to contact, he has an extremely nuanced and pragmatic view, the one number he cited that stuck with me is 60-70% of the fisheries they have studied are unsustainably harvested (i.e. will eventually collapse). (I am not an expert in this space, but trying to get up to speed).

@Cameron Trotter I am very interested in that lit review, if I can help you in any way, let me know.

scholar.google.com.au

Chris Wilcox

Marine Ecologist, CSIRO Marine and Atmospheric Research - Cited by 25,420 - conservation biology - fisheries - statistics - movement - incentives

Original URL: https://scholar.google.com.au/citations?user=V-o9ZkIAAAAJ&hl=en

🙏 Eric Colson

Cameron Trotter (cater@bas.ac.uk)

2023-08-30 12:07:17

@Patrick Beukema more than happy to send it your way when I'm able. Would be good to keep in touch if you're working in the space 🙂

Patrick Beukema (patrickb@allenai.org)

2023-08-30 12:15:33

yeah that sounds great. maybe we could even set up a talk or something? A lot of people on our team would be interested. Its quite a complex space to model/get good data on with all the different dynamics, human behavior, stats, a lot of noise.

Elizabeth Campolongo (e.campolongo479@gmail.com)

2023-08-31 14:53:51

Have you checked out fathomnet for undersea images?

🐟 Malte Pedersen, Eric Colson

Cameron Trotter (cater@bas.ac.uk)

2023-09-01 04:49:48

@Elizabeth Campolongo I have yet to make use of it, though it is on my todo list. I work with Antarctic benthic imagery however, which is an area that even by marine data standards can be somewhat sparse, so I am unsure if it would help in my specific situation, though it may be a good resource for those who work in more explored waters. I'd be interested to hear from those who have used it in the past for their work, it could be an extremely valuable resource!

Sara Beery (sara.beery@gmail.com)

2023-08-30 21:55:05

https://twitter.com/CstGhosh/status/1695350270958490069

X (formerly Twitter)

shreya ghosh on X

🌍:zebra_face:We are excited to announce ACM SIGSPATIAL International Workshop on AI-driven Spatio-temporal Data Analysis for Wildlife Conservation (GeoWildLife 2023)! <https://t.co/sYOzq2VMgE> Deadline: 26th September 2023 #SIGSPATIAL23 #WildlifeConservation #AI @BDilkina @SIGSPATIAL_GIS

Original URL: https://twitter.com/CstGhosh/status/1695350270958490069

👍 Justin Kay, Jason Holmberg (Wild Me), Eric Colson, Omiros Pantazis, Oisin Mac Aodha, Shir Bar, Elizabeth Campolongo

💕 Jon Van Oast, Katelyn Morrison, Caleb Robinson

🌏 Rowan Converse

🙏 Patrick Beukema, Alex Brace

Paul Allin (allinpaul@gmail.com)

2023-09-05 10:32:39

Hi everyone, I’m looking at using ML to count wildlife in an open system and thinking about ways of reducing the amount of data I need to process. Does anyone here know if I can take two images say 1 day apart and subtract the two from each other to only be left with the not stationary objects? hope this makes sense…

Maximilian Schall (Maximilian.Schall@hpi.de)

2023-09-05 10:41:33

Hey Paul,

Given that the camera is stationary: the keyword here would be background subtraction. We had some experiments with videos, where it works quite well. It depends on the scenario and your requirements if it works good enough with two images.

Devis Tuia (devis.tuia@epfl.ch)

2023-09-05 10:42:38

hey Paul, it seems like a reasonable assumption at first, but there are some caveats.. you need to be able to fly twice (seems that you do) and you need to aqcuire the images athte same time of the day (to avoid shadows or illumination strong changes). You also want to avoid fyling on a sunny day vs a cloudy one (you are in South Africa right?). Finally, you also will need very precise co-registration of every pixel, otherwise you will detect a lot of borders as objects.

😲 Rita Pucci

Devis Tuia (devis.tuia@epfl.ch)

2023-09-05 10:55:37

on a stationary camera this is ok-ish (as Maximillian says), but you seem to eb working with drones, right?

Quentin Bateux (quentin.bateux@yale.edu)

2023-09-05 11:24:52

To add on the previous suggestions: if you have small uncontrolled camera motion, you may need a preprocessing step where you perform 'image alignment' before running the background subtraction process.

👍 Sara Beery

Paul Allin (allinpaul@gmail.com)

2023-09-05 11:47:34

Thanks for the swift responses, yes this would be from a drone and I was thinking that with real-time white balance I could correct the lighting for each image. But I do understand that slight variations in height will result in different pixel sizes but currently they are around 5cm so slight difference in height won’t have a big impact on resolution

Dan Morris (agentmorris@gmail.com)

2023-09-05 17:59:03

I'm interested whether drone folks know of anything off the shelf for this... though Devis is definitely "drone folks". 🙂 So the fact that he didn't say "use package X that everyone knows about" suggests it's not an off-the-shelf thing. And it seems like even if you have GPS on the drone, to even try background subtraction, you need registration that's tighter than what you'll get from an off-the-shelf georeferencing tool. Although I say that with no basis in reality, I've never tried.

But assuming you will have to be in the business of writing code, and assuming you want to conduct said business in Python, I would personally start with OpenCV's RANSAC implementation (RANSAC is a common algorithm for aligning images based on keypoints):

https://docs.opencv.org/3.4/d1/de0/tutorial_py_feature_homography.html

Good tutorial here:

https://learnopencv.com/image-alignment-feature-based-using-opencv-c-python/

Even once you have aligned images, background subtraction is a game of heuristics: this is hard even for camera traps, which are literally bolted to a tree. But if your objects are large and salient relative to random other things that change between images (shadows, moving grass, things getting wet, etc.), doable.

Definitely let this Slack know what you learn!

👆 Quentin Bateux

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-09-05 18:08:45

Although I have never tried background subtraction with drone images, my gut feeling is that the precise alignment is going to be very difficult, especially with off the shelf drones. If you have access to high frequency telemetry (self-built drones), it might be of some help. If you are flying off-the-shelf, perhaps anafi is good option to go with. They are cheap and come with very 'accessible' telemetry and programmable control methods -- so you can program paths to fly repeatedly, for example.

also, going back to the original motivation -- would be great to know the actual context/scale of data reduction necessity. Perhaps there a completely alternative solution may emerge.

Jan Kees (jankees.schakel@sensingclues.org)

2023-09-06 02:30:23

hi Paul, it does matter what type of wildlife you want to count. Are you looking at elephant and rhinos (as a must), and at lion, cheetah and other (smaller to much smaller) species as a nice to have?

Paul Allin (allinpaul@gmail.com)

2023-09-06 10:23:25

I definitely intend to explore this further based on the responses so far but not putting all my eggs in one basket. The drone I will be using is a custom built fixed-wing with very accurate on-board gps but perhaps even that is not sufficient. Seems like someone has to go out and try 🙂
@Jan Kees I am looking at ‘large’ mammalian herbivores, i.e. impala and bigger

👍 Jan Kees

Thijs (thijs@q42.nl)

2023-09-06 04:02:25

For those interested, here is a (technical) writeup of a project we have done in the Carpathian mountains to try and keep bears 🐻 out of villages and prevent human-wildlife conflicts. In many ways this is similar to human-elephant-conflicts, but there were some distinct (technical) challenges we had to deal with, for example Romania is not as sunny ☀️ as Africa :)

https://engineering.q42.nl/ai-bear-repeller/

Q42 Engineering

Using AI to keep bears, humans and livestock safe

The 7-step plan for doing something that has never been done before...

Written by

Thomas Broch

Filed under

Artificial Intelligence, Creative tech

Original URL: https://engineering.q42.nl/ai-bear-repeller/

👍 Martin Marzidovsek, Nicolas Arrieta Larraza, Omiros Pantazis, Timm Haucke, Anastasios Angelopoulos, Paul Melki, Prabath Gunawardane, Dan Morris, Alasdair Davies

❤️ Avi Sundaresan, Yseult Hb, Justine Boulent, Timm Haucke, Leonardo Viotti, Marius Miron, Regina Eckert, Anjana Sengupta, Jon Van Oast, Anastasios Angelopoulos, Paul Melki, Chuck Stewart, Steve Haddock, Aakash Gupta

🐻 Anastasia Pagán, Rowan Converse, Anastasios Angelopoulos, Valentin Gabeff, Paul Melki, Varshani Brabaharan, Talia Speaker

Thijs (thijs@q42.nl)

2023-09-06 04:03:55

Just look at one of the beautiful images our system captures, this one is from this morning ❤️

🤯 Jon Van Oast, Anastasia Pagán, Anastasios Angelopoulos, Varshani Brabaharan

Nicolas Arrieta Larraza (n.arrieta.larraza@gmail.com)

2023-09-06 04:34:42

Awesome work! 😊

❤️ Thijs

Ștefan Istrate (stefan.istrate@gmail.com)

2023-09-06 07:23:50

Well done! 👏

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-09-08 07:31:55

This is great! Through NASSCOM and TAIM We are organizing an online roundtable on Realtime monitoring of Wildlife, would you be interested. I have dropped you a DM pls check.

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-09-06 18:28:55

Does anyone have a go-to deep learning/neural network infographic or figure that does a really good job of giving a VERY high-level summary geared towards a non-ML audience?

Patrick Beukema (patrickb@allenai.org)

2023-09-06 19:18:17

this is my go to: https://playground.tensorflow.org/#activation=tanh&batchSize=10&dataset=circle&regDataset=[…]se&problem=classification&initZero=false&hideText=false not sure if the right level though.

playground.tensorflow.org

Tensorflow — Neural Network Playground

Tinker with a real neural network right here in your browser.

Original URL: https://playground.tensorflow.org/#activation=tanh&batchSize=10&dataset=circle&regDataset=reg-plane&learningRate=0.03&regularizationRate=0&noise=0&networkShape=4,2&seed=0.14028&showTestData=false&discretize=false&percTrainData=50&x=true&y=true&xTimesY=false&xSquared=false&ySquared=false&cosX=false&sinX=false&cosY=false&sinY=false&collectStats=false&problem=classification&initZero=false&hideText=false

👍 Carly Batist

Gracie Ermi (gracieermiifthen@gmail.com)

2023-09-06 19:31:09

I've turned this into a gif in the past: the visualization right at this point in this code.org video is pretty good. You might be looking for something more explanatory of the nuts and bolts, but I feel like it visualizes the idea of model training in a helpful way: https://youtu.be/KHbwOetbmbs?t=126

YouTube

Code.org (https://www.youtube.com/@codeorg)

What is Machine Learning?

Original URL: https://youtu.be/KHbwOetbmbs?t=126

👍 Carly Batist

Roni Choudhury (roni.choudhury@kitware.com)

2023-09-08 11:29:52

i don't know if this is the right level/tone: https://www.linkedin.com/posts/bettymohler_post-ugcPost-7018185045085409280-YW_J/

linkedin.com

Betty Mohler on LinkedIn: post

This is a very nice cheat sheet- I especially like the last page.

Original URL: https://www.linkedin.com/posts/bettymohler_post-ugcPost-7018185045085409280-YW_J/

👍 Carly Batist

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2023-09-08 12:42:54

I recently had an email from a colleague wanting to introduce megadetector into their camera trap framework, and I know how we did it in 2021, but I disconnected a bit this year from the species classification matters to analyze the data and write papers, and everything moves so quickly in the AI arena that I wanted to make sure I give them the latest advice. Are there any recent developments on the UI front for run megadetector on large datasets?

Bistra Dilkina (dilkina@usc.edu)

2023-09-08 14:48:46

Following, as I similarly have a collaborator organization interested in using it with a UI

Peter van Lunteren (peter@addaxdatascience.com)

2023-09-08 15:19:42

I created a GUI to deploy MegaDetector and train custom species detectors: https://github.com/PetervanLunteren/EcoAssist

PetervanLunteren/EcoAssist

Train and deploy custom YOLOv5 species classifiers for camera trap images based around the MegaDetector model

Stars

Language

Python

❤️ Tiziana Gelmi Candusso, Sepand Dyanatkar

Dan Morris (agentmorris@gmail.com)

2023-09-08 16:22:40

EcoAssist (which Peter referred to) is the most common GUI for running MegaDetector locally. CamTrap Detector (https://camtrap.net/detector) is also quite polished.

I track all GUIs and other non-traditional ways of running MegaDetector here:

https://github.com/agentmorris/MegaDetector/blob/main/megadetector.md#is-there-a-gui

We also have a 100.000% success rate at helping ecologists get MD running at the command line :), which is sometimes still the best option, depending on the project.

Always feel free to point people who have questions to cameratraps@lila.science (that's me and Siyu), we love hearing from users and can usually quickly route people to the right tools given their specific project requirements.

💚 Carl Boettiger, Carly Batist, Toryn Schafer

❤️ Tiziana Gelmi Candusso

👍 Bistra Dilkina, Cameron Trotter

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2023-09-08 22:33:45

Thank you Dan! I will point them towards the gui and eventually the option of sending you guys an email, I was hesitant about sending them directly to the latter, didnt want to overload you all.

Sara Beery (sara.beery@gmail.com)

2023-09-08 12:54:43

On our end it's still the same public GitHub repo

Sara Beery (sara.beery@gmail.com)

2023-09-08 12:55:31

There are other tools people have developed around it, but since I just run direct from the code I can't speak to the user experience. This would be a good question to ask in the AI for Conservation slack

💯 Tiziana Gelmi Candusso

Vardaan Pahuja (vardaanpahuja@gmail.com)

2023-09-11 17:21:07

Hi everyone, I am a Ph.D. student at The Ohio State University, advised by Prof. Yu Su. I am working on developing species classification models for camera traps using auxiliary information such as taxonomy, location coordinates, etc. I am currently looking to use the lila.science datasets (https://lila.science/datasets) to benchmark my approach. Is there any public benchmark or leaderboard for lila.science datasets? I am aware of the kaggle leaderboard for iwildcam datasets but it doesn't have much information on the methodology of the submission in most cases. Thanks.

Dan Morris (agentmorris@gmail.com)

2023-09-11 17:55:54

To my knowledge, no one has trained a Very Big Model on LILA datasets, so there's not a uniform benchmark. A couple notes that may be useful, though:

I'm increasingly encouraging everyone to think of "all the camera trap data on LILA" as one dataset. It's clearly evolved as a collection of individual datasets, but because it's been mapped to a common taxonomy and because there is a Big Giant .csv file that represents every camera trap image on LILA (both at https://lila.science/taxonomy-mapping-for-camera-trap-data-sets/), you can treat it as one big dataset.

All of the camera trap data on LILA (except for thermal or stereo data) is also available as a Hugging Face dataset (https://huggingface.co/datasets/society-ethics/lila_camera_traps), based on that same Big Giant .csv file, which also may facilitate benchmarking.

I have been trying to keep track of papers that report numbers on individual LILA datasets; look for the "LILA" tag here: https://agentmorris.github.io/camera-trap-ml-survey/#papers-with-summaries... and tell me what I'm missing!

👍 Vardaan Pahuja, Sara Beery, Valentin Gabeff, Fagner Cunha

Dan Morris (agentmorris@gmail.com)

2023-09-11 17:59:50

Also note that there is no precise location information attached to any data on LILA, though where possible, the Big Giant .csv file indicates country/continent/state.

👍 Vardaan Pahuja

Vardaan Pahuja (vardaanpahuja@gmail.com)

2023-09-11 18:01:57

Sounds good, thanks a lot for these resources. I will reach out to you if I come across any papers apart from those mentioned in your repo.

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-09-12 00:49:47

I have executed three projects, two of which are deployed and running on the field. Please refer to the following research in reverse chronological order:

Hakunamatata - This was a hackathon on Driven Data where the winning model used a fine-trained EfficientNet model on Serengeti dataset. I believe this was around 2019-20

DeCaTron - We have deployed a two-stage model (MDv5 + Shifted Window classifier) in one of our projects in India.

ReWilding Europe - This project was centered on an ambitious initiative to reintroduce wildlife into the forests of Europe. We utilized the Yolov7 backbone to train and subsequently deploy a model targeting 30 distinct species.

A primary challenge you may encounter pertains to training for the long-tail distribution, especially in scenarios where data is limited. Additionally, it's crucial to consider how your pipeline addresses and tracks false positives.

👍 Vardaan Pahuja

Devis Tuia (devis.tuia@epfl.ch)

2023-09-14 08:43:26

🚨 Special Issue alert! With @Sara Beery, @Blair Costelloe and @Ruth Oliver we invite you to submit to our new special issue on AI and ecology on Methods in Ecology and Evolution!

Deadline for short white papers on December 1st!

👍 gvanhorn, Oisin Mac Aodha, Cameron Trotter, Lukas Picek, Omiros Pantazis, Benjamin Hoffman, Shir Bar, Vardaan Pahuja, Gustavo Perez, Timm Haucke, Andrew Schulz, Robin Zbinden, Sara Keen, Alexander Merdian-Tarko

🎉 Sara Beery, Katelyn Morrison, Justin Kay, Lukas Picek, Jon Van Oast, Evan Eskew, Taiki Sakai - NOAA Affiliate, Timm Haucke, Robin Zbinden, Valentin Gabeff

❤️ Rowan Converse, Jon Van Oast, Mitchell Rogers, Andrew Schulz, Robin Zbinden

Devis Tuia (devis.tuia@epfl.ch)

2023-09-14 08:45:53

More info here: https://besjournals.onlinelibrary.wiley.com/hub/journal/2041210x/call-for-papers/conservation-ecology-ai

Hamed Alemohammad (h.alemohammad@gmail.com)

2023-09-14 09:46:07

Hi All, we are hiring for two positions at Clark University Center for Geospatial Analytics: 1) Program Director and 2) Geospatial Software Engineer.
Please share these with those who might be interested:
https://bit.ly/ClarkCGAPositions

👍 Sara Beery

Dan Stowell (dan.stowell@naturalis.nl)

2023-09-15 10:22:30

Hi all. We have 10 funded PhD positions to offer on "Bioacoustic AI"! See https://bioacousticai.eu/ and click on "Apply for a PhD position" to link directly to the currently-open opportunities

Bioacoustic AI

Understanding animal sounds with machine learning - Bioacoustic AI

Animals make sounds that convey so much information. How can we use this to help monitor and protect wildlife? In our research programme we want to make sense of animal sounds using AI.

Original URL: https://bioacousticai.eu/

‼️ gvanhorn, Devis Tuia, Sara Beery, Sonny Burniston, Suzanne Stathatos, Oisin Mac Aodha, Carly Batist, Jason Holmberg (Wild Me), Toryn Schafer, Angela Szesciorka, Mikey Tabak, Ben Weinstein, Maddie Cusimano

🎵 Sara Beery, Sonny Burniston, Dan Morris, Jason Holmberg (Wild Me)

❤️ Carly Batist, Jon Van Oast, Jason Holmberg (Wild Me), Viktor Domazetoski, Talia Speaker, Amara McCune, Nicolas Arrieta Larraza, Maddie Cusimano

🐋 Taiki Sakai - NOAA Affiliate, Alexander Merdian-Tarko

🎯 John Martinsson

Amara McCune (amaramccune@gmail.com)

2023-09-16 20:11:18

This looks incredible, thanks for sharing! I was wondering if you know of any opportunities like this that are in the US and open to candidates with a PhD already?

Dan Stowell (dan.stowell@naturalis.nl)

2023-09-19 06:01:12

Hi Amara - I'm no expert on the US scene I'm afraid, but I'm sure there are others here who can give tips!

Dan Stowell (dan.stowell@naturalis.nl)

2023-09-19 07:32:35

Here's a postdoc option at Cornell: https://academicjobsonline.org/ajo/jobs/24693

academicjobsonline.org

Cornell University, Lab of Ornithology

Full service online faculty recruitment and application management system for academic institutions worldwide. We offer unique solutions tailored for academic communities.

Original URL: https://academicjobsonline.org/ajo/jobs/24693

Amara McCune (amaramccune@gmail.com)

2023-09-19 13:15:28

Thanks!

Olof Mogren (olof.mogren@ri.se)

2023-09-18 09:18:00

Hi! On Thursday, @Nico Lang from Copenhagen University is speaking on global vegetation monitoring with probabilistic deep learning in our seminar series RISE Learning Machines. The seminar is open to all on Zoom (and in person in Lund, Sweden), but requires registration! 15:00 CET.
https://www.ri.se/en/learningmachinesseminars/nico-lang-global-vegetation-monitoring-with-probabilistic-deep-learning

RISE

Nico Lang: Global vegetation monitoring with probabilistic deep learning | RISE

At RISE Learning Machines Seminar on September 21 2023, we have the pleasure to listen to Nico Lang, University of Copenhagen give his talk: Global vegetation monitoring with probabilistic deep learning. This talk will present recent research results for global vegetation height mapping exploiting ESA’s Sentinel-2 optical imagery and NASA’s GEDI full waveform LIDAR.

Original URL: https://www.ri.se/en/learningmachinesseminars/nico-lang-global-vegetation-monitoring-with-probabilistic-deep-learning

❤️ Justin Kay, Oisin Mac Aodha, Ben Weinstein, gvanhorn, Kristina Kupferschmidt, Suzanne Stathatos, Sara Beery, Avi Sundaresan, Catarina Silva, Aleksis Pirinen

📆 Shir Bar, Aleksis Pirinen

🌲 Dan Morris, Aleksis Pirinen

Vardaan Pahuja (vardaanpahuja@gmail.com)

2023-09-18 15:33:18

Hi everyone, I have a question regarding the ENA24-detection dataset https://lila.science/datasets/ena24detection available in lila.science. The associated publication "Dynamic Programming Selection of Object Proposals for Sequence-Level Animal Species Classification in the Wild. IEEE Transactions on Circuits and Systems for Video Technology, 2019" is not available in IEEE Xplore or any bibliographic database. There is only a citation in google scholar (the article is missing). I tried to contact the author (Hayder Yousif) but the email is non-existent. If anyone has access to this paper, please let me know. Thanks!

Dan Morris (agentmorris@gmail.com)

2023-09-18 20:00:41

I don't have the paper, and the email address I have for Hayder is his Missouri email (presumably that's the one that no longer works), but Roland Kays is a co-author on the paper and IIRC one of Hayder's co-advisors, so Roland can probably help. I'll DM you with Roland's email address.

Vardaan Pahuja (vardaanpahuja@gmail.com)

2023-09-18 20:01:18

Sounds good, thanks!

Jacob Marks (jamarks13@gmail.com)

2023-09-18 15:53:28

Hey everyone!
☕🏔️📡🧊
One of the AI for Conservation community members, @Nora Gourmelon, published the CaFFe benchmark dataset for tracking glacial calving fronts with SAR imagery. Nora was kind enough to tell me about her research so that I could write a popular science article about it, which was published this morning 🙂

If you're interested glacial calving, computer vision for climate, and supporting your fellow AI for Conservation community members, check out the blog post! https://medium.com/voxel51/caffe-calving-fronts-and-where-to-find-them-1ff57520da45

Medium

CaFFe: Calving Fronts and Where to Find Them

Calving might be the most important natural phenomenon you’ve never heard of. It may sound like an exercise you missed on leg day, but…

Reading time

9 min read

Original URL: https://medium.com/voxel51/caffe-calving-fronts-and-where-to-find-them-1ff57520da45

👏 Vincent Christlein, gvanhorn, Amara McCune, Sara Beery, Cameron Trotter, Leopoldo André Dutra Lusquino Filho, Alexander Merdian-Tarko

😎 Jon Van Oast

👏:skin_tone_4: Chris Llorca

❤️ Nora Gourmelon

Ben Weinstein (benweinstein2010@gmail.com)

2023-09-19 06:09:21

I'm sitting in a great talk that I think would be helpful to anyone studying museum specimens and want to mark body regions and extract morphological locations. https://github.com/EchanHe/PhenoLearn . The developer is here at our session.

EchanHe/PhenoLearn

A tool for phenotyping biodiversity images

Stars

Language

Python

👀 Elizabeth Campolongo, Yseult Hb, Elie Alhajjar, Shir Bar

❤️ Sara Beery, Rita Pucci

😎 Jon Van Oast, Shir Bar

Sara Beery (sara.beery@gmail.com)

2023-09-19 12:43:24

New NSF Global Center on AI and Biodiversity Change!! Led by @Tanya Berger-Wolf with @Justin Kitzes @David Rolnick Graham Taylor Kaitlin Gaynor Marta Jarzyna Laura Pollock @Oisin Mac Aodha @Devis Tuia @Tilo Burghardt Bernd Meyer and me. Very honored to be a part of this!

Read more: https://twitter.com/sarameghanbeery/status/1704170649281802530?t=KZavdjAKMPESV0E1HQiozw&s=19

X (formerly Twitter)

Sara Beery on X

I'm honored to be a part of the leadership team for the new Global Center in AI and Biodiversity Change (ABC), one of seven NSF-funded Global Centers in Climate Change and Clean Energy. Read more about our research goals within the center at @MIT_CSAIL: <https://t.co/2hdROyVdET>

Original URL: https://twitter.com/sarameghanbeery/status/1704170649281802530?t=KZavdjAKMPESV0E1HQiozw&s=19

👍 Vardaan Pahuja, Ben Weinstein, Paul Allin, Elie Alhajjar, Nico Lang, Vincent Christlein, Jacob Marks, Anastasia Pagán, gvanhorn, Dan Morris, Olof Mogren, FANQI Z, Ruben Remelgado, Subhransu Maji, Risa Shinoda, Alex Brace, Cameron Trotter, Leopoldo André Dutra Lusquino Filho, Ted Schmitt, Hemal Naik, Robin Zbinden, Maddie Cusimano, Sepand Dyanatkar, Benjamin Tremoulheac, Valentin Gabeff, charlotte, Rebecca Wilks

😎 Jon Van Oast, Ștefan Istrate, Elie Alhajjar, Oisin Mac Aodha, Anastasia Pagán, Casey Youngflesh

❤️ Justin Kay, Bistra Dilkina, Leonardo Viotti, Sowbaranika, Devis Tuia, Rowan Converse, Malte Pedersen, David Rolnick, Declan, Stefan Schneider, Mélisande Teng, Elie Alhajjar, Arjun Subramonian (they/them), Gustavo Perez, Amara McCune, Oisin Mac Aodha, Jacob Marks, Elizabeth Campolongo, Anastasia Pagán, Viktor Domazetoski, Talia Speaker, Shir Bar, Risa Shinoda, aruna, Alex Brace, Lukas Picek, Omiros Pantazis, Andrew Schulz, Ted Schmitt, Robin Zbinden, Valentin Ștefan, Olivier Dietrich, Ronan Wallace

🙌 Shir Bar, Carly Batist, Jenna Kline, Evan Eskew, Omiros Pantazis, Carl Boettiger, Ruth Oliver, Yseult Hb, Ronan Wallace

🎉 Matt Weldy, Ruth Oliver, Shir Bar, charlotte, Urs, Alexander Merdian-Tarko

Bistra Dilkina (dilkina@usc.edu)

2023-09-19 12:45:39

This is just amazing news!!! Looking forward to what this initiative brings to our field!

❤️ Sara Beery

😍 Sara Beery, David Rolnick

Arjun Subramonian (they/them) (arjun.subramonian@gmail.com)

2023-09-19 13:08:03

Congrats 💜

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-09-20 07:56:18

Really excited to announce a new white paper from Rainforest Connection (RFCx): Harnessing the Power of Sound & AI to track Global Biodiversity Framework (GBF) Targets.

The paper explores the power of ecoacoustics and AI to monitor biodiversity and track progress towards GBF targets using case studies from around the world. 🔊🌳

Read it here ➡️ https://rfcx.org/publications/harnessing-the-power-of-sound-and-ai-to-track-global-biodiversity-framework-gbf-targets

👍 gvanhorn, Justin Kay, Dan Morris, Georgia Atkinson, Leopoldo André Dutra Lusquino Filho, Viktor Domazetoski, Dante Wasmuht, Yseult Hb, Marconi Campos

🔈 Suzanne Stathatos, Rowan Converse, Amee Assad, Marconi Campos

🎉 Eric Greenlee, Gracie Ermi, Talia Speaker, charlotte, Marconi Campos

😎 Jon Van Oast, Marconi Campos

Leopoldo André Dutra Lusquino Filho (leopoldo.lusquino@unesp.br)

2023-09-20 10:35:32

Hello everyone,
We are in the process of establishing a Green AI center in Brazil dedicated to climate research, named GRAIN. This center comprises three of Brazil's leading universities: the Federal University of Rio de Janeiro, São Paulo State University, and the Federal Fluminense University. Currently, GRAIN brings together 12 faculty members from various disciplines, including Computer Science, Electrical Engineering, Environmental Sciences, Meteorology, and Health, as well as over 60 undergraduate students and 20 PhD students.
The primary focus of our research group will be a large-scale project aimed at developing techniques for predicting extreme weather events, both forecasting and nowcasting. Additionally, we will work on the creation of Digital Twins for monitoring hydrodynamic reservoirs using Green ML, TinyML, Deep Learning, Weightless Neural Networks, Hyperdimensional Computing, IoT, Data Fusion, and Physics-informed Neural Networks.
For this project, we are fortunate to have the support of the Brazilian Navy, the São Paulo State Watershed Committee, and the AI Hubs in the states of Rio de Janeiro and São Paulo. This is a long-term project, and we are actively seeking international collaborations.
If you are affiliated with a research center whose work aligns with ours and are interested in establishing collaborations, please do not hesitate to reach out to me at leopoldo.lusquino@unesp.br.

❤️ Jon Van Oast, Sara Beery, Hemal Naik, Yseult Hb, Philippe Hermant, Carly Batist

🌎 Sara Beery

Peter van Lunteren (peter@addaxdatascience.com)

2023-09-21 06:44:04

Hi everyone 👋

I would like to introduce my recently founded company, Addax Data Science. 🚀 At Addax, we have the simple mission of providing tools that enable ecologists to spend less time on boring tasks and more time on conservation. Think statistics, software, automation, and custom identification models for bird vocalizations or camera trap images. Visit the website for more information: https://addaxdatascience.com/

Do you have any questions or want to know more? Don't hesitate to get in touch!

Addax Data Science

Addax Data Science - Customised coding solutions for ecologists

Statistics and software for ecologists. Addax offers customised coding solutions for more efficient data analysis and intelligent workflow automation.

Original URL: https://addaxdatascience.com/

🙌 Carly Batist, Sara Beery, Dan Morris, Timm Haucke, Yseult Hb, Marconi Campos

Jan Kees (jankees.schakel@sensingclues.org)

2023-09-27 07:04:11

congrats Peter 🙂

😁 Peter van Lunteren

Sara Beery (sara.beery@gmail.com)

2023-09-21 07:26:01

Who is going to ICCV? Or who is local in Paris?

I'd love to do another bird walk!

🐦 Oisin Mac Aodha, Risa Shinoda, Katelyn Morrison, Subhransu Maji, Shir Bar, Caleb Robinson, Yseult Hb, Joakim Bruslund Haurum, Peter Kulits

Sara Beery (sara.beery@gmail.com)

2023-09-21 07:29:29

I've heard reports of several research collaborations getting started at the CVPR bird walk, which is awesome 🎊

Oisin Mac Aodha (macaodha@caltech.edu)

2023-09-21 07:30:17

I'm going and would love to join!

❤️ Sara Beery

Risa Shinoda (shinoda.lisa.47z@gmail.com)

2023-09-21 08:09:53

I’ll be there and I’d love to join in! 🐥

❤️ Sara Beery

Devis Tuia (devis.tuia@epfl.ch)

2023-09-21 08:31:28

unfortunately not 😞

Sara Beery (sara.beery@gmail.com)

2023-09-21 08:32:20

@Devis Tuia as if you would wake up early 😂😂

Devis Tuia (devis.tuia@epfl.ch)

2023-09-21 08:35:11

it’s in my DNA. But I was annoucning that sadly I wont’ be in Paris at all

😞 Sara Beery, Oisin Mac Aodha

Jonathan Roberts (jdr53@cam.ac.uk)

2023-09-21 08:55:42

I’m also going and would be keen to join 🙂

❤️ Sara Beery

Diego Marcos (diego.marcos.gonzalez@gmail.com)

2023-09-21 11:15:05

Working with the Pl@ntNet people I should be proposing a plant walk, but I'd be happy to join for the birds 😅

😍 Sara Beery

Caleb Robinson (calebrob6@gmail.com)

2023-09-21 12:40:52

I'll be there and would love to come too!

😍 Sara Beery

Sara Beery (sara.beery@gmail.com)

2023-09-21 12:42:12

We can look at plants too!!!!

😄 Diego Marcos

Thomas Radinger (thomasrad@protonmail.com)

2023-09-22 03:07:54

Local in Paris and would love to join as well for bids and plants :)

❤️ Sara Beery

😍 Sara Beery

Sara Beery (sara.beery@gmail.com)

2023-09-27 18:45:41

https://twitter.com/sarameghanbeery/status/1707164338769875380?t=UVWoUdHgoeoT_krxtERseQ&s=19

X (formerly Twitter)

Sara Beery on X

📢📢📢 Calling all computer vision bird nerds!! We're holding another #CVBirdWalk at #ICCV2023 next week! Meet up at 7am CET on Thursday 10/5 in front of the convention center, we'll carpool to Bois de Boulogne and loop around the lake 🐦🦢🦆 All are welcome!!

Original URL: https://twitter.com/sarameghanbeery/status/1707164338769875380?t=UVWoUdHgoeoT_krxtERseQ&s=19

😍 Jon Van Oast, Oisin Mac Aodha

🥐 Suzanne Stathatos

🦢 Jonathan Roberts

🙌 Thomas Radinger

👍 Thor Veen

Sara Beery (sara.beery@gmail.com)

2023-10-04 10:16:56

https://twitter.com/sarameghanbeery/status/1709547363012948263?t=0j3QXp5lmzmHsqLf6tGHKQ&s=19

X (formerly Twitter)

Sara Beery on X

Excited to see all of you tomorrow morning! We'll meet at 7am outside of the entrance gates to the convention center and then carpool to Bois de Boulogne. If you want to meet us there, we will be by the lake next to Pavilion Royale until 7:30ish <https://t.co/gv9WiLBST2>

Original URL: https://twitter.com/sarameghanbeery/status/1709547363012948263?t=0j3QXp5lmzmHsqLf6tGHKQ&s=19

🥖 Jon Van Oast, Frederic

❤️ Jon Van Oast

🪶 Shir Bar

🥐 Diego Marcos

Barbie D (barbara.raven42@gmail.com)

2023-09-21 11:20:54

https://www.smithsonianmag.com/science-nature/four-amazing-impacts-of-this-ai-powered-bird-migration-tracker-180982932/

This is so cool!! Thought I’d share this article about BirdCast
(1) birdCast can help target city wide lights-out campaigns to correspond with high migration time
(2) New York city audubon uses it to predict number of dead birds they’re gonna see and plan accordingly
(3) birds become “trapped” in the 9/11 tribute lights, and volunteers monitor that and turns it off when too many birds are trapped
(4) you can see speed, direction, and altitude of migrations!!
(5) it can warn farmers when to be on high alert for avian flu

Smithsonian Magazine

Four Amazing Impacts of This A.I.-Powered Bird Migration Tracker

A reimagined tool called BirdCast is helping birders, scientists and even farmers

Original URL: https://www.smithsonianmag.com/science-nature/four-amazing-impacts-of-this-ai-powered-bird-migration-tracker-180982932/

😮 Jon Van Oast, Eric Greenlee

👍 gvanhorn, Sara Beery, Dan Morris, Viktor Domazetoski

Vardaan Pahuja (vardaanpahuja@gmail.com)

2023-09-24 01:16:18

Hi everyone, I have a question regarding the MegaDetector detections available for lila.science. Some datasets (e.g. Orinoquía Camera Traps) have multiple animals annotated per image in some cases. Corresponding to these, the MegaDetector also provides multiple detections with high confidence. Is there a way to may the detections to respective categories? The MegaDetector categories are just animal, person, and vehicle, so they are not useful in this case.

LILA BC

Orinoquía Camera Traps - LILA BC

Overview This data set contains 104,782 images collected from a 50-camera-trap array deployed from January to July 2020 within the private natural reserves El Rey Zamuro (31 km2) and Las Unamas (40 km2), located in the Meta department in the Orinoquía region in central Colombia. We deployed cameras using a stratified random sampling design across...

Written by

lilawp

Est. reading time

3 minutes

Original URL: https://lila.science/datasets/orinoquia-camera-traps/

Dan Morris (agentmorris@gmail.com)

2023-09-24 20:58:39

There's not a way to do this reliably, sorry. If you are, for example, combining those MegaDetector boxes with image-level labels to train a multiclass detector, you will likely have to either (a) ignore images with multiple species or (b) manually assign labels to boxes for those images. FWIW, outside of the African datasets, multi-species images are extremely rare on LILA, so you probably don't lose a lot of information if you discard multi-species images in, for example, Orinoquia Camera Traps, and/or it would be super-duper fast to manually assign labels for the relatively small number of boxes this affects.

If you are working with the Snapshot Safari data (where multi-species images are common enough to possibly want a solution), you could run a classifier trained on cropped animals on each box in each of those images, and I think 99.9% of the time, even if the classifier isn't great, you would be able to assign boxes to labels (e.g. if you had zebras and warthogs in the same image, you only really need to rely on the classifier to assign a higher probability to zebra than warthog or vice-versa to help you assign those boxes). But even there, I think it would be faster to do it manually.

Dan Morris (agentmorris@gmail.com)

2023-09-24 21:00:15

Actually if you are training a multiclass detector, the most elegant solution is probably to leave those out, train an initial detector, use that detector to assign labels to that small set of boxes, manually clean up what should now be a very small number of mistakes, and re-train.

Vardaan Pahuja (vardaanpahuja@gmail.com)

2023-09-24 21:36:34

Sounds good, thanks for the information!

Olof Mogren (olof.mogren@ri.se)

2023-09-25 02:20:46

Amazing seminar last Thursday with @Nico Lang in our Learning Machines series. Thanks @Sara Beery, @Ben Weinstein, @Alexander MathisEPFL and others for attending in-person and contributing to a great event.

We have created a new youtube playlist with all our environment-related talks. Check it out!

https://www.youtube.com/playlist?list=PLqLiVcF3GKy0-jZFGg-VqLzh51LqCfduN

November 2, Pria Donti is visiting our seminar! More info: https://ri.se/lm-sem

YouTube

AI for the Environment - Learning Machines Seminars

Our popular RISE Learning Machines Seminars take place every Thursday at 15:00 CET, with invited speakers from all over the world, covering recent advancemen...

Original URL: https://www.youtube.com/playlist?list=PLqLiVcF3GKy0-jZFGg-VqLzh51LqCfduN

RISE

Learning Machines Seminars – Upcoming Seminars | RISE

Here you find all the upcoming Learning Machines Seminars. Update your calendar with upcoming seminar dates of your interest!

Original URL: https://ri.se/lm-sem

👍 Oisin Mac Aodha, Valentin Gabeff, Shir Bar, Ștefan Istrate, Paul Melki, gvanhorn, Elie Alhajjar, Declan, Carly Batist, Ben Weinstein, Olivier Dietrich, Dario Prifti, Alexander Merdian-Tarko, stefano puliti

🙌 John Martinsson, Justin Kay, Elie Alhajjar, Suzanne Stathatos

❤️ Sara Beery, Oskar Åström, Nico Lang

Matt Weldy (matthewjweldy@gmail.com)

2023-09-25 14:41:57

I am currently preparing for my PhD comprehensive exams and I am digging through a wide range of Ecology and Methods literature; however, I wanted to reach out to see if anyone had any reading (book, paper, other) recommendations relevant to the interests of this community (AI in conservation? In other words what reading lists have you/would you recommended to PhD students interested in ecology, AI, and analysis?

Here is a simple google drive sheet if anyone has any recommendations.
https://docs.google.com/spreadsheets/d/1F7dE7UUYZcU4Rx6tsg4QTgoUcYTqYMh_Kugt0jnzWqA/edit?usp=sharing

Thank you!

😍 Sara Beery, Nanticha Ocharoenchai (Lyn)

🙏 Sara Beery

🎉 Jon Van Oast, Shir Bar

Peter van Lunteren (peter@addaxdatascience.com)

2023-09-25 14:46:16

Sorry, because I’m on my phone I can’t edit your google docs file. I thought this book was an interesting read: AI in the Wild - Sustainability in the Age of Artificial Intelligence - By Peter Dauvergne

👍 Matt Weldy

Matt Weldy (matthewjweldy@gmail.com)

2023-09-25 14:51:56

Thanks! Feel free to add here and I'll move them to the sheet.

Titus (titus@colossal.com)

2023-09-25 16:02:43

We recently released a review on AI for elephant monitoring https://arxiv.org/abs/2306.13803

arXiv.org

Elephants and Algorithms: A Review of the Current and Future Role of AI in Elephant Monitoring

Artificial intelligence (AI) and machine learning (ML) present revolutionary opportunities to enhance our understanding of animal behavior and conservation strategies. Using elephants, a crucial species in Africa's protected areas, as our focal point, we delve into the role of AI and ML in their conservation. Given the increasing amounts of data gathered from a variety of sensors like cameras, microphones, geophones, drones, and satellites, the challenge lies in managing and interpreting this vast data. New AI and ML techniques offer solutions to streamline this process, helping us extract vital information that might otherwise be overlooked. This paper focuses on the different AI-driven monitoring methods and their potential for improving elephant conservation. Collaborative efforts between AI experts and ecological researchers are essential in leveraging these innovative technologies for enhanced wildlife conservation, setting a precedent for numerous other species.

Original URL: https://arxiv.org/abs/2306.13803

👍 Matt Weldy

Vardaan Pahuja (vardaanpahuja@gmail.com)

2023-09-25 21:41:47

Hi, I am unable to download the annotations for Missouri camera traps in LILA https://lila.science/datasets/missouricameratraps. The annotations link hosted by LILA is non-functional https://lilaannex.blob.core.windows.net/lila-annex/missouri_camera_traps_set1.zip Any help is appreciated.

LILA BC

Missouri Camera Traps - LILA BC

Overview This data set contains approximately 25,000 camera trap images representing 20 species (for example, the most common labels are red deer, mouflon, and white-tailed deer). Images within each sequence share the same species label (even though the animal may not have been recorded in all the images in the sequence). Around 900 bounding boxes...

Written by

lilawp

Est. reading time

3 minutes

Original URL: https://lila.science/datasets/missouricameratraps

Dan Morris (agentmorris@gmail.com)

2023-09-25 23:20:59

Good catch! Should be fixed now (new link). The issue was... no, actually the issue is really boring, but it should have only impacted this particular file. Let me know if anything looks off, the metadata for this dataset has been... complicated (see the note on the page about some bounding boxes that aren't quite right).

🎉 Jon Van Oast, Sara Beery

Vardaan Pahuja (vardaanpahuja@gmail.com)

2023-09-25 23:31:45

thanks, it works now!

Vardaan Pahuja (vardaanpahuja@gmail.com)

2023-09-28 12:26:52

The page says that bounding boxes provided by the authors of Missouri camera traps dataset aren't accurate. How about the megadetector results available on LILA? Are those reliable to be used instead?

Dan Morris (agentmorris@gmail.com)

2023-09-29 18:37:10

It's not that the bounding boxes aren't accurate, rather that for 79 images (there is a list on the page), only one bounding box is available, when in fact there are multiple animals in the image. So if you are using the dataset to train a detector, I would leave out those 79 images. If you are cropping the boxes and using those crops to train a classifier, they're all fine.

But yes, I also expect that the MDv5 results for this dataset are good too.

👍 Vardaan Pahuja

Vardaan Pahuja (vardaanpahuja@gmail.com)

2023-10-02 00:03:38

Hi, while processing the LILA datasets, I found that for most datasets, the annotation images and the images available for download correspond to slightly different sets of images. Just want to make sure there is no processing issue at my end.

Dan Morris (agentmorris@gmail.com)

2023-10-02 14:18:15

The only difference you should see is that images with humans are in the metadata, but not in the actual image folders. That would also include images labeled, e.g. "vehicle", and in some cases "domestic dog" or even "horse". When posting datasets to LILA, we avoid posting images that even might have people in them.

If you see a significant number of images in the metadata that are not labeled human and are not present in the cloud bucket(s) or zipfile(s), like missing images of lions or bobcats or whatever, please let me know at info@lila.science. Thanks for checking everything carefully!

👍 Vardaan Pahuja

Drea Burbank (drea@savimbo.com)

2023-09-26 22:37:07

We might have the world’s first certified biodiversity credit. We could really use some open-source coders to help with this. isbm.savimbo.com

https://www.savimbo.com/biodiversity

Savimbo

Our biodiversity credit — Savimbo

Savimbo uses a indicator species biodiversity methodology to calculate biodiversity credits. Direct from smallfarmers in the Colombian Amazon, these credits were designed for indigenous peoples, who guard 80% of the conserved biodiversity on earth. so they can access biodiversity markets fairly, and

Original URL: https://www.savimbo.com/biodiversity

👍 Jose Ruiz-Munoz

👍:skin_tone_4: Chris Llorca

😎 Jon Van Oast

Casey Clifton (caseyclifton@proton.me)

2023-11-26 02:28:05

Hi Drea, this is really interesting and something ive been trying to research lately. Can you point me at any papers discussing the extend to which indicator species can proxy broader biodiversity and ecosystem health?

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-09-28 09:19:18

Cool use of camera-traps for non-wildlife insights! https://x.com/jamiealison30/status/1706953803474387324?s=20

X (formerly Twitter)

Jamie Alison on X

🚨New preprint!🚨 Wildlife cameras capture more than just #wildlife. This #deeplearning model extracts sunshine, snow and hail from millions of images. The data help to explain #ecology on camera. <https://t.co/4PSMOi6A4D> @TokeHoye @HjalteMann @EkoLogIt @ProteaSandals @annebeejay

Original URL: https://x.com/jamiealison30/status/1706953803474387324?s=20

👍 Justin Kay, mimi, Dan Morris, Eric Greenlee, Ștefan Istrate, Shir Bar, Sara Beery, Valentin Gabeff, Aakash Gupta

❤️ Jamie Alison

Justin Kay (justinkay92@gmail.com)

2023-09-28 09:21:06

@Catherine Breen

❤️ Sara Beery

Catherine Breen (cbreen@uw.edu)

2023-10-18 23:08:04

I love this!! I am catching up on all my missed slacks and this looks awesome. Thank you for sharing Carly 🙂 !! And for tagging me Justin 🙂

👍 Carly Batist

stefano puliti (stefano.puliti@nibio.no)

2023-09-28 14:57:31

For those of you interested in 🌲 and lasers 🚨 here is a new ML-ready benchmark dataset we recently published. It's called FOR-instance and is composed of manually segmented trees in 3D forest point cloud scenes. IT can be used for instance (trees) and semantic (tree components) segmentation.

data paper: https://arxiv.org/abs/2309.01279
data: https://zenodo.org/record/8287792

Zenodo

FOR-instance: a UAV laser scanning benchmark dataset for semantic and instance segmentation of individual trees

The challenge of accurately segmenting individual trees from laser scanning data hinders the assessment of crucial tree parameters necessary for effective forest management, impacting many downstream applications. While dense laser scanning offers detailed 3D representations, automating the segmentation of trees and their structures from point clouds remains difficult. The lack of suitable benchmark datasets and reliance on small datasets have limited method development. The emergence of deep learning models exacerbates the need for standardized benchmarks. Addressing these gaps, the FOR-instance data represent a novel benchmarking dataset to enhance forest measurement using dense airborne laser scanning data, aiding researchers in advancing segmentation methods for forested 3D scenes. In this repository, users will find forest laser scanning point clouds from unamnned aerial vehicle (using Riegl sensors) that are manually segmented according to the individual trees (1130 trees) and semantic classes. The point clouds are subdivided into five data collections representing different forests in Norway, the Czech Republic, Austria, New Zealand, and Australia. These data are meant to be used either for developement of new methods (using the dev data) or for testing of exisitng methods (test data). The data splits are provided in the data_split_metadata.csv file. A full description of the FOR-instance data can be found at <http://arxiv.org/abs/2309.01279>

Original URL: https://zenodo.org/record/8287792

🙌 Joe Ferdinando, Suzanne Stathatos, Ben Weinstein, Emily Lines, Enis Berk Çoban, Nico Lang, Dan Morris, Emilio Luz-Ricca, Antonio Ferraz

👍 Caterina Barrasso

stefano puliti (stefano.puliti@nibio.no)

2023-09-28 14:58:36

[ MESSAGE TEXT EMPTY ]

Dan Morris (agentmorris@gmail.com)

2023-10-02 14:35:12

New dataset on LILA, courtesy of USGS:

https://lila.science/datasets/izembek-lagoon-waterfowl/

This dataset contains over 500k points (with species-level labels) on waterfowl in aerial images from Alaska. It's a subset of a larger dataset that's also public (https://alaska.usgs.gov/products/data.php?dataid=484), and usually I don't host datasets on LILA that already exist somewhere else, but in this case (a) the original dataset is massive, very difficult to download from ScienceBase, and contains ~95% empty images, and (b) a bunch of work went into aligning the original metadata with the images and converting to a standard format, and I didn't want everyone else who works with this dataset to have to repeat that work.

And I have a specific request for anyone who wants to tinker with this dataset!

tl;dr: it would be great if someone could train YOLOv8x on this data and compare to the YOLOv5x6 model we've already trained.

Long version...

We've trained a YOLOv5 (specifically YOLOv5x6) detector on this data that USGS and US Fish & Wildlife are happy with, so the project is in a good stable state (https://github.com/agentmorris/usgs-geese). But the decision to use YOLOv5 rather than the newer, better YOLOv8 was mostly based on the fact that YOLOv5x6 has a 1280-pixel input size, and there isn't (yet) a 1280-pixel version of YOLOv8. But there's not really a reason not to break everything into 640px patches (instead of 1280px) and use YOLOv8x. It would require no new code to train YOLOv8 now, because all the training was done through the YOLOv5 CLI (which is almost identical to the YOLOv8 CLI); it would just be some environment setup and job management and GPU time. So, if someone really wants to get your detector on, let me know.

Everyone place your bets on this thread about whether using a newer model that will require a little more inference time due to the smaller patch size (YOLOv8x is like 10% faster per patch than YOLOv5x6, but will require ~4x as many patches) will work better, or whether it won't really make a difference. You can see a bunch of sample patches and sample results at the GitHub link above to formulate your opinion.

👍 Aamir Ahmad, Ben Weinstein, Vincent Christlein, Devis Tuia, Fadel Seydou, Sam Lapp

🦆 Rowan Converse, Sara Beery, Emilio Luz-Ricca

‼️ gvanhorn

😎 Jon Van Oast

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-10-03 03:54:55

we could give YOLOv8 a try on this... just need a bit of time (after IROS this week and our next field trip.. we are back in the office on 24th Oct)

Dan Morris (agentmorris@gmail.com)

2023-10-03 09:12:15

SG, drop me an email when you're back (agentmorris@gmail.com), and I'll send some more information.

👍 Aamir Ahmad

Dan Morris (agentmorris@gmail.com)

2023-10-02 17:09:10

There was a thread here a few months ago about bounding box annotation tools; @Aakash Gupta maintains a useful list here.

Since then I tried a bunch of tools for the specific scenario where an AI model generates boxes that are mostly sensible, but need a bunch of cleanup. I found that most tools - especially Label Studio, which seems to be the big-and-generally-quite-good gorilla in this space - did not make this super-easy unless you were completely bought in to the tool's built-in ML pipeline, but labelme came the closest, because it's fast and simple, with a simple file format, and the code is simple enough to be easy to modify, easy enough that it's efficient to make disposable versions of the tool for specific tasks.

I ended up forking labelme to add some minor features that end up being a win for my specific scenario, like keyboard adjustment of box boundaries, and keyboard control of box selection. Also the ability to make absurdly large, bright boxes, because when everything is working well, you're just banging the "next" button as fast as you can and it's really helpful to have super-salient boxes.

Everything I did is a total hack and crashes randomly and I don't think this fork will be useful to anyone, but I am curious about the features that other folks doing similar work found were necessary for AI-assisted annotation that other tools were missing. I.e., the previous thread was really about which tools are "good", now I'm curious about what features are missing in existing annotation tools for everyone's conservation dataset tasks.

FWIW I found BoundingBoxEditor to be the best tool for validating and previewing YOLO-formatted boxes, as a final consistency check before training with a framework that expects YOLO-formatted annotations.

👍 Aakash Gupta, Akash Jaiswal

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-10-03 03:53:41

Perhaps https://github.com/robot-perception-group/smarter-labelme?search=1 could be of help.. we built it on top of Labelme. We further adapted it for behavior annotations here https://github.com/robot-perception-group/animal-behaviour-inference (perhaps that thread you meant was about out smarterlabelme tool when we released it). A short summary of what we do in this tool is tracking (predict-detect loop like (but not exactly) a Kalman filter) the bounding boxes. Both predicter and detector are DNNs. A human in the loop fixes small deviations if any. We see almost 10-fold decrease in time spent annotating/correcting.

Gracie Ermi (gracieermiifthen@gmail.com)

2023-10-02 17:13:42

The company I work for, Impact Observatory, has started offering the ability to more easily download our free annual land use and land cover maps clipped to any area of interest for any year from 2017-2022.

Our global maps were already open access, but now you can order them through our store and get your desired data emailed to you already clipped to your area of interest along with various metrics about your data. Thought this might be helpful or interesting to some folks in this group! https://x.com/ImpactObserv/status/1702684441271804004?s=20

X (formerly Twitter)

Impact Observatory on X

It's important to us that our open land cover data is easy to access. Now, anyone can download our Historic Annual Land Cover Maps directly through IO Monitor. Get a map and pre-calculated metrics clipped to your area of interest starting today. <https://t.co/akfTHQkkP6>

Original URL: https://x.com/ImpactObserv/status/1702684441271804004?s=20

👍 Dan Morris, Carly Batist

👀 Elizabeth Campolongo

🎉 Jon Van Oast, Carly Batist

Dan Morris (agentmorris@gmail.com)

2023-10-02 17:18:34

I just placed a test order for a small area around where I live to try out the tool; the process was very slick!

Gracie Ermi (gracieermiifthen@gmail.com)

2023-10-02 17:33:43

Glad to hear it @Dan Morris!

Paul Allin (allinpaul@gmail.com)

2023-10-25 06:08:05

Cool product! Is there some way of saving the polygon or getting multiple years in one order?

Gracie Ermi (gracieermiifthen@gmail.com)

2023-11-03 14:48:51

Thanks, @Paul Allin! Using the same custom polygon for multiple orders is a much-requested feature, and one that we definitely want to offer soon!

Chris Yeh (chrisyeh96@gmail.com)

2023-10-02 17:26:47

🚨 Last call for submissions to the 2023 NeurIPS Workshop on Computational Sustainability!
https://www.compsust.net/compsust-2023/

- Submission Deadline: Oct 3, 2023 (AOE)
- Notification of Acceptance: Oct 21, 2023
- Workshop: December 15th, 2023; New Orleans, Louisiana

We hope to see some submissions from the AI for Conservation community!

Feel free to message me or contact neurips-workshop2023@compsust.net with any questions.

Katelyn Morrison (https://aiforconservation.slack.com/team/U04A34D3673)

Hi everyone, I am helping organize the <a href="https://www.compsust.net/compsust-2023/">NeurIPS 2023 workshop on Computational Sustainability</a>. Computational sustainability (CompSust) is an interdisciplinary research area that uses computational methods to help address the 17 United Nations Sustainable Development Goals (UN SDGs), including but not limited to hunger and poverty reduction, infrastructure development, and environmental conservation. While we encourage all submissions related to computational sustainability, this year’s workshop will have a special focus on promises and pitfalls on the pathway from theory to development. In particular, we encourage submissions with negative results that highlight challenges and gaps in existing methods. We <a href="https://openreview.net/group?id=NeurIPS.cc/2023/Workshop/CompSust">invite submissions</a> of 4 pages, with an emphasis on topics related to deploying computational sustainability methods. Reviews will be double-blind. The workshop does not publish proceedings, and submissions are non-archival. Submission to this workshop does not preclude future publication. Submissions of work which has been previously published, including papers accepted to the main NeurIPS 2023 conference are allowed this year. Important Dates: - Submission Deadline: Oct 3, 2023 (AOE) - Notification of Acceptance: Oct 21, 2023 - Workshop: December 15th, 2023; New Orleans, Louisiana We hope to see some submissions from the AI for Conservation community! Feel free to message me or contact <a href="mailto:neurips-workshop2023@compsust.net">neurips-workshop2023@compsust.net</a> with any questions.

🙌 Patrick Beukema

Dan Morris (agentmorris@gmail.com)

2023-10-02 17:34:09

The negative results focus is a great idea. Negative ML results in the literal sense of low accuracy are often a little tricky because it can be impossible to say whether the results were fundamental or just related to implementation issues. But negative results where the ML worked (in the sense of high accuracy), but the model still didn't end up being useful (because the test domain surprised you, or you built a model you thought the user wanted but it wasn't actually what they wanted, or the user thought 91% accuracy was good enough but it turns out they needed 94% accuracy, or whatever)... those are super-interesting! Excited to see what submissions you get in the "negative results" department.

👍 Chris Yeh, Cameron Trotter, Yuanqi Du, Vardaan Pahuja, Sara Beery, Ștefan Istrate, Paul Melki, Katelyn Morrison, Justin Kay, Sepand Dyanatkar

❤️ Suzanne Stathatos, Sara Beery, Katelyn Morrison

Sara Beery (sara.beery@gmail.com)

2023-10-03 09:50:26

Amazing slide from @Caleb Robinsons talk at the HADR workshop 😂😂😅😅😅

👍 Jonathan Roberts, Olivier Dietrich

👏 Esther Rolf, Devis Tuia, Dan Morris, Mark Goldwater, Taiki Sakai - NOAA Affiliate, Emilio Luz-Ricca, Katelyn Morrison

🙃 Chris Yeh

Sako Arts (sako@fruitpunch.ai)

2023-10-03 10:22:10

FruitPunch AI for Turtles Challenge coming up!
🌊🐢 Dive into the world of Sea Turtle Conservation! 🐢🌊

We're thrilled to announce the 𝐀𝐈 𝐟𝐨𝐫 𝐓𝐮𝐫𝐭𝐥𝐞 𝐂𝐡𝐚𝐥𝐥𝐞𝐧𝐠𝐞! 🐢🤖

Our goal is to develop cutting-edge computer vision software that can recognize and distinguish individual turtles through automated identification. Current turtle monitoring methods are intrusive, time-consuming, and expensive, posing challenges to organizations dedicated to protecting these endangered creatures.

How it Works:
Sea turtles have unique facial scales, much like human fingerprints. AI can identify these distinct characteristics, helping researchers associate them with individual turtles. In this Challenge, we partner up with @ Sea Turtle Conservation Bonaire, an amazing organization committed to protecting these precious animals and their habitat.

In this challenge, we aim to:

Explore Siamese and Triplet networks for turtle identification.

Investigate state-of-the-art techniques like 𝐀𝐫𝐜𝐅𝐚𝐜𝐞.

Develop data pipelines for processing training and testing sets.

Create an easy-to-use GUI in collaboration with researchers for efficient turtle identification.

We're seeking passionate individuals from various backgrounds, including AI/ML enthusiasts, software developers, biologists/conservationists interested in AI, and computer scientists.
We expect a commitment of ~8 hours per week. You will be joining weekly meetings, masterclasses, and presentations.

Join us in making a real impact on turtle conservation in Bonaire and contribute to preserving these incredible creatures! 🌊🐢

Learn more about the challenge on our website:
👉 https://www.fruitpunch.ai/challenges/ai-for-turtles

fruitpunch.ai

AI Challenge aiming to automate the identification process of sea turtles

Join to put your AI skills to the test and help conservationists in their efforts to protect the turtles!

Original URL: https://www.fruitpunch.ai/challenges/ai-for-turtles

🎉 Carly Batist, Dan Morris, Cameron Trotter, Jon Van Oast, Katelyn Morrison, Sara Beery, Ronan Wallace

🐢 Cameron Trotter, Yseult Hb, Elizabeth Campolongo, Gracie Ermi, Mitchell Rogers, Katelyn Morrison, Sako Arts, Sara Beery, Ronan Wallace

Elizabeth Campolongo (e.campolongo479@gmail.com)

2023-10-03 11:25:43

Do the datasets for these challenges get published somewhere afterwards? I couldn't find a link to data on past challenges on the website.

Katelyn Morrison (kcmorris@andrew.cmu.edu)

2023-10-04 00:39:26

Have you chatted with the ML engineers at WildMe yet? @Jason Holmberg (Wild Me) and his team may have some good insights on this challenge 🙃

Sako Arts (sako@fruitpunch.ai)

2023-10-04 09:07:39

Hi Elizabeth, it depends on the Challenge Owner who provide the data if we can publish their data. If we can we always publish it on our platform but for many cases we sadly cannot. In this case we have 3 datasets, two of those are from the Caribbeans, they will only be made available to challenge participants for now, the last dataset is a public data set from South Africa.

Sako Arts (sako@fruitpunch.ai)

2023-10-04 09:16:43

@Katelyn Morrison we have most definitely been in touch, mostly regarding our animal classification cases, not yet on an identification case, though I'd be eager to collaborate. I see they already have a Internet of Turtles project

❤️ Katelyn Morrison

Sako Arts (sako@fruitpunch.ai)

2023-10-04 09:17:39

@Jason Holmberg (Wild Me) I read you use a SIFT based method. I'd be interested to hear about your experiences with this algorithm!

❤️ Katelyn Morrison

Urs (urs.waldmann@uni-konstanz.de)

2023-10-05 04:00:14

Hi Sako, what kind of annotations does the public dataset from South Africa contain? Can you post a link to this public dataset, please? Thanks, Urs

Sako Arts (sako@fruitpunch.ai)

2023-10-05 04:23:17

Hi Urs, You can find the data and the explanation of the labels here:
https://zindi.africa/competitions/turtle-recall-conservation-challenge/data

Urs (urs.waldmann@uni-konstanz.de)

2023-10-05 04:31:25

Tnhanks, @Sako Arts

Casey Clifton (caseyclifton@proton.me)

2023-11-26 02:23:44

Hey @Sako Arts, I've been approached to advise a sea turtle citizen science startup on AI methods for individual ID of sea turtles. Wondering if you have any suggestions on open source models/techniques I should be looking at? Cheers 🙂

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2023-11-26 12:31:31

@Casey Clifton ArcFace loss + EfficientNet is our state of the art for general re-ID. Turtles are on our roadmap (data from iot.wildbook.org), but it might be Q1 2024 before we have time to train.

https://github.com/WildMeOrg/wbia-plugin-miew-id

WildMeOrg/wbia-plugin-miew-id

Stars

Language

Python

🙌 Casey Clifton

Casey Clifton (caseyclifton@proton.me)

2023-11-26 18:35:00

Thanks @Jason Holmberg (Wild Me)!

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-10-04 18:52:29

Anyone else going to the GEOBON Monitoring Biodiversity for Action conference in Montreal next week?! 😃

❤️ Jon Van Oast, Timm Haucke, Sara Beery

👋 Timm Haucke, Benjamin Kellenberger, Ruben Remelgado, Justine Boulent

Sara Beery (sara.beery@gmail.com)

2023-10-05 07:46:01

@Timm Haucke will be representing our group there!

🎉 Carly Batist

Mélisande Teng (tengmeli@mila.quebec)

2023-10-06 11:46:07

I will also be attending!

👋 Timm Haucke, Carly Batist

🎉 Carly Batist, Justine Boulent

Peter van Lunteren (peter@addaxdatascience.com)

2023-10-11 11:34:55

👋 Hi, I’d like to ask for some advice about training a single-class species detector. I’m training a YOLOv8 classification model to be used in conjunction with MegaDetector 5. At the moment, I’m only interested in one class: lion. The problem with one class is that the built-in validation of YOLOv8 doesn’t work, since all images are always classified as the only possibility, namely lion. Does anyone have experience with this? It seems to me that the YOLOv8 architecture doesn’t really support one-class training. Do you recommend adding a second class non-lion with representative other animals from the ecosystem (e.g. giraffe, zebra, elephant, etc.)? Or is it better to create custom test and validation functions that incorporate some kind of confidence threshold? Thanks!

Titus (titus@colossal.com)

2023-10-11 11:36:09

Yes, you have to have an alternative option like non-lion. Otherwise your attempting to train lion vs NULL. That’s the easiest route at least.

🙏 Peter van Lunteren

Sara Beery (sara.beery@gmail.com)

2023-10-11 11:39:38

Is the idea to classify crops from megadetector or to train a separate lion-only detector? If it's the second, and you have bbox labels, then it should use the background not in the lion boxes as the negative class.

👍 Titus

🙏 Peter van Lunteren

Peter van Lunteren (peter@addaxdatascience.com)

2023-10-11 13:40:01

Great 🙂 Yes, it will classify crops from megadetector. I'll go and find some non-lion crops to add a second class!

Different scenario: if I wanted to train a classifier for lion and zebra , do you recommend to add a third class other? Or could I in this case use the confidence values to decide which animals are not lion nor zebra ? For example, if the crops from megadetector don't get a classification for either lion or zebra above the threshold value X, then it'll be a classified as other.

Dan Morris (agentmorris@gmail.com)

2023-10-12 11:23:04

This is one of those questions where I think you get a free PhD in computer science if you can provide a universal answer. :) I'm assuming you have labels for a bunch of other species too, e.g. you're working with Snapshot Safari data, or more generally, it's likely that whatever your training data is, it's not only lion and zebra that were labeled. Even if you're only interested in lions and zebras, I'm 71% confident that you want to include labels for at least the other common classes as well, but you probably don't want to use so many training examples that they overwhelm lions or zebras. E.g. if you're in an ecosystem where wildebeest and elephants are also common, you could (a) lump them into an "other" class or (b) give them their own classes, and maybe lump the less common animals into "other". I'm 71% confident your performance on lions and zebras will go up if you choose (b), but I'm not sure. And it's hard to say how far down the list of classes you want to go before you start lumping things into "other". If you have 10 species that are as common as lions, I think you want them all to be their own classes, but I'm not sure. Let us know what you learn!

👀 Steve Haddock

🙏 Peter van Lunteren

😀 Alan Stenhouse

Chuck Stewart (cvstewart@gmail.com)

2023-10-12 11:29:49

❤️ I really like this answer.

➕ Sara Beery

Peter van Lunteren (peter@addaxdatascience.com)

2023-10-13 07:54:04

I was hoping for one of those answers with which you'll get a free PhD... 😁 Thanks guys! That makes a lot of sense 🚀

Patrick Beukema (patrickb@allenai.org)

2023-10-11 17:00:22

Hey all, We came across a strange artifact when we were evaluating one of our newer models in a sentinel-2 image. It looks like confetti separation of the RGB channels (shown in the TCI imagery below). We don’t know what this is — and a quick G search did not yield anything relevant. Any ideas? Has anyone seen this before?

Caleb Robinson (calebrob6@gmail.com)

2023-10-11 17:21:33

Here's the scene ID https://planetarycomputer.microsoft.com/api/stac/v1/collections/sentinel-2-l2a/items/S2A_MSIL2A_20231011T025631_R032_T49NDJ_20231011T101112

Caleb Robinson (calebrob6@gmail.com)

2023-10-11 17:21:48

And a link to look around on the PC https://planetarycomputer.microsoft.com/explore?c=110.6857%2C7.9889&z=9.99&v=2&d=sentinel-2-[…]Natural+color&s=false%3A%3A100%3A%3Atrue&sr=desc&ae=0

planetarycomputer.microsoft.com

Microsoft Planetary Computer

Supporting sustainability decision-making with the power of the cloud

Original URL: https://planetarycomputer.microsoft.com/explore?c=110.6857%2C7.9889&z=9.99&v=2&d=sentinel-2-l2a&m=Most+recent+%28any+cloud+cover%29&r=Natural+color&s=false%3A%3A100%3A%3Atrue&sr=desc&ae=0

Patrick Beukema (patrickb@allenai.org)

2023-10-11 17:22:42

there is separation there as well

Patrick Beukema (patrickb@allenai.org)

2023-10-11 17:22:47

https://planetarycomputer.microsoft.com/explore?c=111.0984%2C7.9519&z=13.54&v=2&d=sentinel-2[…]Natural+color&s=false%3A%3A100%3A%3Atrue&sr=desc&ae=0

planetarycomputer.microsoft.com

Microsoft Planetary Computer

Supporting sustainability decision-making with the power of the cloud

Original URL: https://planetarycomputer.microsoft.com/explore?c=111.0984%2C7.9519&z=13.54&v=2&d=sentinel-2-l2a&m=Most+recent+%28any+cloud+cover%29&r=Natural+color&s=false%3A%3A100%3A%3Atrue&sr=desc&ae=0

Patrick Beukema (patrickb@allenai.org)

2023-10-11 17:22:51

nice viewer by the way

❤️ Caleb Robinson

Caleb Robinson (calebrob6@gmail.com)

2023-10-11 17:29:55

1000% agree, but I can't take any credit for it:)

Patrick Beukema (patrickb@allenai.org)

2023-10-11 17:31:09

yeah its super nice — responsive, fast, does exactly what you think it should do with zero need to read documentation

Nathan Jacobs (jacobsn@wustl.edu)

2023-10-11 17:55:52

That looks like the combination of specular reflections and different acquisition times / angles. Some additional info: https://gis.stackexchange.com/questions/332634/water-reflectance-effect-on-sentinel-2-images

Geographic Information Systems Stack Exchange

Water reflectance effect on Sentinel 2 images

I'd like to know what is causing the effect on the water depicted on the following screen (sentinel tile id: S2B_MSIL1C_20190805T144739_N0208_R139_T20PPA_20190805T162922 located near Saint Lucia). ...

Original URL: https://gis.stackexchange.com/questions/332634/water-reflectance-effect-on-sentinel-2-images

🙌 Patrick Beukema, Dan Morris

Patrick Beukema (patrickb@allenai.org)

2023-10-11 18:25:25

wow. That makes total sense. Thank you for responding.

Patrick Beukema (patrickb@allenai.org)

2023-10-11 18:48:27

@Nathan Jacobs your explanation makes sense but also I would expect to see this effect more often. To date I have only seen it once, and we look at a significant amount of S2 TCI imagery.

Nathan Jacobs (jacobsn@wustl.edu)

2023-10-11 23:01:40

maybe it's a scale issue... the reflections need to be large enough to not get averaged out w/in a single pixel... honestly not sure

Patrick Beukema (patrickb@allenai.org)

2023-10-11 23:34:38

thank you!

Devis Tuia (devis.tuia@epfl.ch)

2023-10-12 02:34:54

I remember back in the days Quickbird had similar isues because the IR band was not acquired simulataneously to the RGB. SO you could see moving objects by making a false color composite

✔️ Patrick Beukema

Patrick Beukema (patrickb@allenai.org)

2023-10-12 11:44:55

Appreciate all the responses. For the moment we are just going to tell our users that this is a known anomaly with the cause likely attributed to reflectance as Nathan mentioned. If we see this more often, and if it has an impact on our precision, then we will revisit and remove this effect — perhaps via a conventional noise filtering technique (erosion, dilation).

Kostas Papafitsoros (k.papafitsoros@qmul.ac.uk)

2023-10-12 05:56:24

Exciting PhD opportunity!
I am advertising a fully funded PhD studentship at the Queen Mary University of London under the project "Data-driven Image Processing Methods with Applications to Wildlife Conservation" (deadline: December 1st). If you have a strong mathematical/computer science background, good coding skills, are interested in AI and you would like to put those for the good of wildlife conservation with a special focus on sea turtles, then this project is for you!
👉 More information here: bit.ly/46s8CFB (or drop me a direct message here)

❤️ Jaroslav Bezdek, Lukas Picek, Oisin Mac Aodha, Emily Lines, Yseult Hb, Suzanne Stathatos, Valentin Gabeff, Sara Beery, Katelyn Morrison, Akshay Paruchuri, Alexander Merdian-Tarko, Alex Brace

🎉 Lukas Picek, Dan Morris, Sara Beery, Katelyn Morrison, Jaka Cikač

🐢 Cameron Trotter, Robin Zbinden, Emilio Luz-Ricca, Sara Beery, Katelyn Morrison, Malte Pedersen, Jaka Cikač

Benjamin Hoffman (benjaminsshoffman@gmail.com)

2023-10-16 14:59:35

Hi everyone, I’m writing to share some tool-building work @Maddie Cusimano and I did which might be useful for other folks working with long audio recordings: https://github.com/earthspecies/voxaboxen.

This is a framework to train a sound event detection model directly on audio plus Raven selection tables. It differs from most other sound event detection approaches that we know, in that it uses an object detection framework (similar to YOLO) to box events, rather than a segmentation or classification framework. This approach is useful for labeling data at a fine temporal scale, which we wanted for looking at interactive vocal behavior of animals.

We talk more about our approach in this blog post: https://www.earthspecies.org/blog/voxaboxen-new-tool-to-support-annotation-of-large-audio-files. Happy to answer any questions, hear comments, and receive feedback! We can also help you to run this on your data.

earthspecies.org

Voxaboxen: New Tool Supports Annotation of Large Audio Files - Earth Species Project

Original URL: https://www.earthspecies.org/blog/voxaboxen-new-tool-to-support-annotation-of-large-audio-files

earthspecies/voxaboxen

Stars

Language

Python

🎉 Jon Van Oast, Maddie Cusimano, Elizabeth Campolongo, Dan Morris, gvanhorn, Enis Berk Çoban, John Martinsson, Yseult Hb, Dylan Van Bramer (she/her)

❤️ Nicolas Arrieta Larraza, Enis Berk Çoban, Aniruddha Saha, L, Marius Miron

👍 Steve Murphy, Eelke, Vincent Christlein, Cameron Trotter, Valentin Gabeff, Alan Stenhouse

Dan Morris (agentmorris@gmail.com)

2023-10-16 21:28:33

How many other people immediately went to YouTube after reading this post and searched for "what sound does a meerkat make?". I did. It was worth it. Meerkats just got more adorable, which is saying something.

💚 Maddie Cusimano, Mitchell Rogers, Shir Bar, Benjamin Hoffman

😀 Alan Stenhouse

Nicolas Arrieta Larraza (n.arrieta.larraza@gmail.com)

2023-10-17 02:52:23

Great work! I really enjoyed reading through the article 👍😊

😄 Benjamin Hoffman

😊 Maddie Cusimano

Louisa van Zeeland (cepstrum@gmail.com)

2023-10-17 15:41:35

Hi all, I want to pass along Tom Mustill's call for contribution to his upcoming work MESH x How to Speak Whale. They aim "to bring the voices of life in the ocean to new ears, to help artists discover new creative terrains and forge links through science and art to ocean conservation". All contributors will be fully credited. I must add that Tom's Whale Song Bath at the British Library this past summer was absolutely brilliant and I'm sure this event will be as well. Please see attached flyer for more information and his contact info 🐋

😎 Jon Van Oast, Justin Kay, Jason Holmberg (Wild Me)

🐋 Dan Morris, Alex Brace, Cameron Trotter

Patrick Beukema (patrickb@allenai.org)

2023-10-18 12:00:04

Hi all, I am writing to share several remote sensing computer vision models that we recently created and open sourced.

https://github.com/allenai/vessel-detection-sentinels

https://github.com/allenai/vessel-detection-viirs

These models support real time streaming computer vision (vessel detection) from NASA’s and ESA’s publicly available satellite imagery (VIIRS, Sentinel-1, and Sentinel-2) in our platform (Skylight: https://www.skylight.global/). Many of these models were built by or in close collaboration with Prior (AI2's CV division, and especially Favyen Bastani who is incredibly talented at geospatial AI). Our team has benefitted massively from all of the expertise and insights from users in this community and we are grateful for feedback about these models (engineering, modeling, documentation, or otherwise).

skylight.global

Skylight | Home

Skylight helps improve maritime transparency and protect our oceans.

Original URL: https://www.skylight.global/

allenai/vessel-detection-viirs

Stars

Language

Python

😍 Sara Beery, Gracie Ermi, Elie Alhajjar, Taiki Sakai - NOAA Affiliate, Henry Herzog, Jason Holmberg (Wild Me), Carl Boettiger

😎 Jon Van Oast, Shir Bar, Jason Holmberg (Wild Me)

👏 Jaka Cikač

🙌 Alan Stenhouse

Taiki Sakai - NOAA Affiliate (taiki.sakai@noaa.gov)

2023-10-18 13:25:15

This looks awesome! Our group has some projects trying to understand how vessel traffic noise affects marine mammal presence, but we are only using AIS data. We're always concerned about how many vessels we might miss because they don't have AIS

🙌 Patrick Beukema

Steve Haddock (haddock@mbari.org)

2023-10-18 14:42:07

Echoing the congrats. 🌏
We have used VIIRS to detect bioluminescent milky seas, and are working toward ML methods to automatically flag them in the data streams. Your project looks like a helpful launch point.
https://www.nature.com/articles/s41598-021-94823-z

Nature

Honing in on bioluminescent milky seas from space

Scientific Reports - Honing in on bioluminescent milky seas from space

Original URL: https://www.nature.com/articles/s41598-021-94823-z

🙌 Patrick Beukema

Patrick Beukema (patrickb@allenai.org)

2023-10-18 14:59:02

Very cool project! (and brings me back to swimming in bioluminescent bays in Puerto Rico with my wife.

FWIW There are downloading scripts (for creating datasets) NASA in there as well you may find useful. We use this service in a real time platform, and we actively maintain/monitor this code — if you notice any issues feel free to open an issue, and if you want any more details we would be happy to talk — we think VIIRS can be a very powerful technology, and there is even a third satellite that was recently launched (NOAA-21) that will enable us to have 3 passes globally every night. (I gave a talk on this model recently at the NOAA-AI conference: https://noaaai2023.sched.com/event/1SA4O/live-demo-shedding-light-on-shadowed-waters-geospatial-computer-visions-role-in-tackling-illegal-fishing)

https://github.com/allenai/vessel-detection-viirs/blob/d345c61725f117049577ee359b160d9cb1b9f61a/src/gen_obj_detection_dataset.py#L121-L133

noaaai2023.sched.com

5th NOAA Workshop on Leveraging AI in Environmental Sciences: Live Demo: Shedding Light on Shadowed Wa...

View more about this event at 5th NOAA Workshop on Leveraging AI in Environmental Sciences

Original URL: https://noaaai2023.sched.com/event/1SA4O/live-demo-shedding-light-on-shadowed-waters-geospatial-computer-visions-role-in-tackling-illegal-fishing

<https://github.com/allenai/vessel-detection-viirs/blob/d345c61725f117049577ee359b160d9cb1b9f61a/src/gen_obj_detection_dataset.py | gen_obj_detection_dataset.py>

<pre><code> if not os.path.exists(dnb_path): with open(dnb_path, "w+b") as fh: utils.download_url(dnb_url, TOKEN, fh) geo_path = os.path.join(image_dir, geo_url.split("/")[-1]) if not os.path.exists(geo_path): with open(geo_path, "w+b") as fh: utils.download_url(geo_url, TOKEN, fh) phys_path = os.path.join(image_dir, phys_url.split("/")[-1]) if not os.path.exists(phys_path): with open(phys_path, "w+b") as fh: utils.download_url(phys_url, TOKEN, fh) </code></pre>

👀 Steve Haddock

Dan Morris (agentmorris@gmail.com)

2023-10-18 22:28:48

There were a couple threads recently about training classifiers on MegaDetector crops... I mentioned a few months ago that I posted MD results for all the camera trap datasets on LILA; I'm posting a minor update to those. Partially because it's possibly useful and partially because the mountain was there, I ran the repeat detection elimination process on all of those results, which is basically a semi-automated process that removes a lot (but not all) (maybe not even most) of the false detections on rocks and sticks that happen a zillion times in a row. If you are training classifiers on LILA data using MD crops, the RDE results are the ones you want to use:

https://lila.science/megadetector-results-for-camera-trap-datasets/

Unrelated to the RDE, I also cleaned up some ambiguity about exactly what the base paths were in each of the results files.

Thanks again to @Doantam Phan who recently added a visualization of lots of small tiles to the RDE code, without which it wouldn't be possible to do this at this scale.

Let me know if these are useful, and/or if anything looks goofy in the files.

👍 Valentin Gabeff, Carly Batist, Sara Beery, Alan Papalia, Jon Van Oast, Timm Haucke, Emilio Luz-Ricca, Jason Holmberg (Wild Me), Sepand Dyanatkar, Alan Stenhouse

🙌 Elizabeth Campolongo, Mitch Fennell, Jason Holmberg (Wild Me)

🙏 Thor Veen, Peter van Lunteren, Jason Holmberg (Wild Me)

Sarah (thom1253@umn.edu)

2023-10-26 14:28:59

Hi all -- I'm a lurker here, but maybe its time to say something 🙂. I run a huge cam trap effort for the state of Idaho (tens of millions of images). I'd really love to talk to other people in similar situations & hear about any tools or tricks you have found that help you manage your flow of pics. We use the Megadetector and Timelapse. In addition, I've built a few R tools and scripts over the years for various things (renaming pics, QA/QC/ etc). I'm interested in pooling efforts/ideas/wishlists or just providing emotional support for each other 🙂 . Specifically, I am interested in a regionally-specific classifier (ID, WA, MT, CO, OR, WY?, UT?). I'm also clueless about hardware that might help us (right now, field staff copy SD cards to 2TB ext hard drives, those get mailed to me-- then I run MD etc). I'd like to disperse some of these activities to not-me, but need to build easy-to-use tools for that to happen. I have lots of ideas, little time, mediocre computing skills.. anyone interested in talking? reach out to me!

👋 Sara Beery, Elie Alhajjar, Omiros Pantazis, Boyu Zhang, Michael Bunsen, Talia Speaker, Enis Berk Çoban, Aakash Gupta, Toryn Schafer, David

🦌 Dan Morris, David

😀 Michael Bunsen

🙌 Carly Batist

👍 Luke Sheneman, Steve Murphy

👏 Alan Stenhouse

Michael Procko (xprockox@gmail.com)

2023-10-26 14:56:17

Hey Sarah, myself and some prior colleagues spent a lot of time developing a workflow to rename, blur, and bin photos using MegaDetector, and I have recently been delving into training regionally-specific classifiers for Washington, British Columbia, Alberta (though only broad classification into carnivore vs. non-carnivore at the moment). Not sure if I can be of much help, but I’d be happy to chat if you have any specific questions!

😎 Jason Holmberg (Wild Me), Sara Beery

❤️ Tiziana Gelmi Candusso

Cara Appel (appelc@oregonstate.edu)

2023-10-26 15:06:47

Hi Sarah (also hi Michael!), I am working on similar projects in Oregon and would be happy to chat as well

👋 Michael Procko, Michael Bunsen

😎 Jason Holmberg (Wild Me), Sara Beery

Sarah (thom1253@umn.edu)

2023-10-26 15:13:27

Yes to both of you! I'll reach out to you each to schedule a shortish call. thanks for responding

👍:skin_tone_2: Cara Appel

👍 Michael Procko, Jason Holmberg (Wild Me), Michael Bunsen

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2023-10-26 17:40:35

WildMe.org here in Oregon. Please include me in your discussions. We take imagery to the individual ID level where possible and would love to work with regional states on AI for aerial surveys too

👍 Sarah

Michael Bunsen (notbot@gmail.com)

2023-10-27 00:42:50

Hi all! I am in Oregon as well and would love to join the conversation. I am part of an international group focused on automating insect monitoring. Some regional support would be very welcome 🙂

👍 Sarah

👍:skin_tone_2: Cara Appel

Michael Bunsen (notbot@gmail.com)

2023-10-27 00:46:23

@Cara Appel I will be at OSU this weekend for the PNW Lepidopterists Workshop. They are going to show off a new collection in the newly remodeled Cordley Hall! https://osac.oregonstate.edu/events/44th-annual-pacific-northwest-lepidopterists-workshop

😍 Sara Beery

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2023-10-27 11:19:51

Hi Sarah, Our workflow here in Toronto uses megadetector and timelapse too!

😎 Sarah

Cara Appel (appelc@oregonstate.edu)

2023-10-27 14:09:37

@Michael Bunsen Sounds so interesting! I will be out of town. Do you know Michael Getz at OSU? Looks like he's not on this channel, but yet another Michael in the AI world and he does pollinator work.

Michael Bunsen (notbot@gmail.com)

2023-10-27 14:37:12

@Cara Appel I don't know that Michael but I am intrigued! I appears that he is a leader in the field of truffle dog training as well 😃

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-10-30 01:18:19

Hi Sarah - I have worked on creating a workflow and species level classification for Wildling Europe and another project in Telangana, India. In India we are working with local state governments, to setup their machine annotation pipelines for the millions of data points that they gather through regular monitoring.
Happy to have a chat and discuss any potential collaborations.

😎 Jason Holmberg (Wild Me), Michael Bunsen

👍 Sarah, Prabath Gunawardane

Prabath Gunawardane (prabathg@gmail.com)

2023-10-31 00:30:31

Hi Sarah,
We do something similar in SF Bay Area with Felidae Conservation Fund Wildepod Project (https://felidaefund.org/projects/community/wilde-pod)
We have about 1.6 million images (and counting) collected from a couple of hundred camera traps scattered around bay area. A small team of volunteer engineers (including me and @Abhay) built the wildepod.org site which allows the field volunteers to upload the image sets, which are then processed by MegaDetector and then annotated for species etc by a group of other volunteers through a custom interface. It's a Django site hosted on App Engine and I'm happy to share details with you based on your interest.

We've been meaning to implement automated species classification as well but so far this year keeping abreast of front-end features and site breakages etc. Hopefully early next year!

Would love to collaborate with folks doing similar work! 🙂

Felidae Conservation Fund

Wilde Pod | Felidae Conservation Fund

Felidae Conservation Fund combines top research, community engagement, and education to help humans live in harmony with wild cats and restore ecosystems.

Original URL: https://felidaefund.org/projects/community/wilde-pod

😎 Sarah, Michael Bunsen, Abhay

Sarah (thom1253@umn.edu)

2023-10-31 11:15:17

Thats really neat Prabath - It sounds like what so many of us are wishing for -- I will take a look!

👍:skin_tone_5: Prabath Gunawardane

Serge Wich (sergewich@gmail.com)

2023-11-10 17:03:38

I am not sure whether it can help for a complete workflow but we have a North American species detection model that we are trying to expand to more species and into result dashboards that are useful for users. https://www.conservationai.co.uk/ I am happy to discuss if this can be part of a workflow that is of use for you and others. We are also always looking for more images for species in North America to train the model to include more species of mammal.

👍 Sarah

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-10-30 13:28:31

My final “who’s going” conference round-up for this year (😂) - I will be at the Wildlife Society conference next week in Louisville, KY and would love to meet up with other conservation tech and AI folks! My stuff:

Workshop on acoustic monitoring (with some sprinkles of AI) on Sunday (Nov 5), 8am-12pm

Podium pres. on using ecoacoustics to track invasive species in Puerto Rico

❤️ Suzanne Stathatos, Toryn Schafer, Enis Berk Çoban, Jason Holmberg (Wild Me), Sara Beery, Prabath Gunawardane, Santiago Ruiz Guzman

Toryn Schafer (tschafer@tamu.edu)

2023-10-30 14:51:57

I am attending the conference Monday. I am in the symposium "Agent-based modeling for wildlife management". There is a get-together for the symposium planned afterwards at Down One Bourbon Bar Monday, Nov. 6th, at 6pm

🙌 Carly Batist

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-10-30 14:59:09

Awesome, thanks @Toryn Schafer!

Riley Knoedler (mknoedler@west-inc.com)

2023-11-08 12:14:53

I'm here as well, come by the WEST booth to chat!

Olof Mogren (olof.mogren@ri.se)

2023-11-02 10:19:59

Priya Donti in Learning Machines!
https://rise.zoom.us/j/208117140?pwd=SENUZTZ4SDdtc0tvcFdkNzlUQ2tNUT09

🙌 Justin Kay, Shir Bar, Carl Boettiger, Sara Beery

Olof Mogren (olof.mogren@ri.se)

2023-11-02 11:17:53

Amazing talk. Intrigued by the possibilities that optimization in the loop machine learning can give!

Ben Weinstein (benweinstein2010@gmail.com)

2023-11-02 10:29:39

Does anyone know of airborne detection of humans in natural landscapes datasets? I know of https://www.crcv.ucf.edu/data/UCF-ARG.php and https://arxiv.org/pdf/2209.00128.pdf. I am writing a proposal to expand the DeepForest backbone models, marine mammals, terrestrial ungulates, etc. Should we do a human detection model? I can imagine it being very helpful in human-wildlife conflict, anti-poaching work, etc. I know many members of our community have data in this area and we could package it together as we move towards a general ecological object (including humans) detector for open imagery. What does the community think of making such a model available? Are we treading on ethically questionable ground? Potential for nefarious misuse is pretty high for airborne human detection model? Discuss.

❤️ Jon Van Oast, Kakani Katija

Sara Beery (sara.beery@gmail.com)

2023-11-02 11:03:48

@Elizabeth Bondi-Kelly has a dataset for aerial detection in thermal data

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-11-02 11:31:43

Hi Ben, we have some experience in this context through our AirCap project. I will be glad to connect. you can check the project, pubs and dataset here : https://www.aamirahmad.de/projects/aircap/

flightrobotics

AirCap

Original URL: https://www.aamirahmad.de/projects/aircap/

😎 Jon Van Oast

👍 Ben Weinstein

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-11-02 11:34:41

... and all open source code here https://github.com/robot-perception-group/AirCap

robot-perception-group/AirCap

Public Code Repository of the Aerial Outdoor Motion Capture Project at the Max Planck Institute for Intelligent Systems, Tübingen.

Stars

Language

Elizabeth Bondi-Kelly (ecbk@umich.edu)

2023-11-02 14:52:44

Thanks for tagging me @Sara Beery! Yes, here is the dataset: https://sites.google.com/view/elizabethbondi/dataset?authuser=0. I also think that the ethical discussion is incredibly important, and I'd like to refer to some great work that I've seen in this space - this is of course not exhaustive, and much has come from Wildlabs!

https://conbio.onlinelibrary.wiley.com/doi/10.1111/csp2.374

https://besjournals.onlinelibrary.wiley.com/doi/10.1002/2688-8319.12033

https://www.youtube.com/watch?v=Ge6_EF0z83U

https://fnigc.ca/ocap-training/

https://news.usask.ca/articles/colleges/2020/engaging-northern-expertise-strengthens-ecological-science.php

https://link.springer.com/article/10.1007/s13280-015-0714-0

https://vimeo.com/771752163

sites.google.com

Elizabeth Bondi-Kelly - DATASET

BIRDSAI The Benchmarking IR Dataset for Surveillance with Aerial Intelligence (BIRDSAI, pronounced "bird's-eye") is a long-wave thermal infrared dataset containing nighttime images of animals and humans in Southern Africa. The dataset allows for benchmarking of algorithms for automatic detection

Original URL: https://sites.google.com/view/elizabethbondi/dataset?authuser=0

YouTube

Cabot Institute for the Environment (https://www.youtube.com/@TheCabotInstitute)

Drone Ecologies: Day 2, Session 1

Original URL: https://www.youtube.com/watch?v=Ge6_EF0z83U

❤️ Sara Beery, Alan Stenhouse

Patrick Beukema (patrickb@allenai.org)

2023-11-09 13:34:30

Does anyone here use ESA’s sentinel 1,2 datasets for a (near) real time streaming application? And if so what is your experience been like with the new (or old) API? We are running into occasional data availability issues, and not sure what we might do to improve (e.g. leverage a different data provider such as Sentinel-Hub).

Casey Youngflesh (caseyyoungflesh@gmail.com)

2023-11-13 12:50:08

I’m currently working on something that relies on the Sentinel API, the goal being to DL/process imagery daily. I’ve found the API extremely clunky but have a pipeline that I believe works. I’m very much trying to avoid going with a downloader-pays model, which seem to be getting more common… Interested in getting your perspective here and happy to chat more.

Patrick Beukema (patrickb@allenai.org)

2023-11-13 12:54:51

We are having quite a bit of friction with this and we are happy to pay but its not obvious if that would solve our problems. Its not trivial to support these NRT use cases, and requires significant resources to do so, and then there is the question of at what SLA. There are groups moving to the cloud/s3 (like NOAA for some of their products), which enables a more modern AWS API.

Our use case is rather demanding — we support organizations world wide and they need to know as soon as possible whether there are vessels somewhere/doing something they aren’t supposed to be doing). Our computer vision is actually quite fast in comparison to the downlink/transfer from ESA’s servers.

Patrick Beukema (patrickb@allenai.org)

2023-11-13 12:55:11

| Its not trivial to support these NRT use cases,
not us — I mean for NASA/ESA etc.

Patrick Beukema (patrickb@allenai.org)

2023-11-13 12:58:05

FWIW We have had great experience with NRT NASA data (we use SuomiNPP/NOAA20, and soon NOAA21 satellites) — there servers are basically always on -- and we only see performance degradation when there is an actual on board satellite issue causing data corruption/preventing downlink to earth. We

Patrick Beukema (patrickb@allenai.org)

2023-11-13 12:58:44

NASA’s NRT data is free — which is kind of surreal when you think about how serious/big of a lift supporting that volume of data and NRT on top of that

Patrick Beukema (patrickb@allenai.org)

2023-11-13 13:00:56

Our goal is to open source the best solutions we find so that groups don’t have to reinvent the wheel every time. Here is the service for NRT processing of Suomi-NPP and NOAA-20 — I linked to it earlier in this channel, but didn’t underscore the downloading component, in case you find the code useful: https://github.com/allenai/vessel-detection-viirs and happy to chat of course as well

allenai/vessel-detection-viirs

Model and service code for streaming vessel detections from VIIRS satellite imagery

Stars

Language

Python

Casey Youngflesh (caseyyoungflesh@gmail.com)

2023-11-16 12:27:39

Thanks for sharing the repo and for your take on this! Nice to know about the reliability of the Suomi/NOAA NRT data. From my memory Sentinel data is available within about 4 hours of acquisition (when DLing directly from the ESA Sentinel API) which I could see would not be ideal for your use case. I’m more concerned with satellite revisit time then time to data availability (once it’s in the hours range). Happy to share the code I have for Sentinel-2 once it’s cleaned up/fully operational.

Dan Stowell (dan.stowell@naturalis.nl)

2023-11-10 12:19:54

I've put some course material online for my "AI for Nature & Environment" including 2 video-lectures:
https://github.com/danstowell/ai_nature_environment

😍 Sara Beery, Justin Kay, Timm Haucke, Taiki Sakai - NOAA Affiliate, Oisin Mac Aodha, Prabath Gunawardane, Dylan Van Bramer (she/her), Katelyn Morrison, Rita Pucci, Jennifer, Carly Batist

🙏 Patrick Beukema, Alexander Merdian-Tarko, Enis Berk Çoban, Kalindi Fonda, Maddie Cusimano, charlotte

🙌 Alan Stenhouse, Talia Speaker, Carly Batist, Julien Boussard

❤️ Thomas Radinger

Violet Turri (vturri@andrew.cmu.edu)

2023-11-10 13:11:54

Hi everyone! @Katie Robinson and I are AI researchers at the Carnegie Mellon University Software Engineering Institute. We are interested in exploring applications of large language models (LLMs) within the conservation space. If anyone is working on a project in this space, please shoot us a message and we can schedule a time to chat more about your work. Thanks!

👍 Katie Robinson, Ankita Shukla, Negar Sadrzadeh, Patrick Beukema, Collin Abidi, Jason Holmberg (Wild Me), Henry Herzog, Douglas Mbura, Katelyn Morrison

👍:skin_tone_3: Alan Stenhouse

Patrick Beukema (patrickb@allenai.org)

2023-11-10 13:40:13

Hi we (Skylight) are exploring geospatial tuned LLMs (bespoke to Skylight) to remove barriers for entry/accelerate adoption/help growth etc — we built (and deployed) one for AI2's hackathon with AI2's Aristo team. I went to CMU and this caught my eye 🙂. Feel free to reach out to my email directly: patrickb@allenai.org Skylight is a marine conservation platform.

👍 Violet Turri

Patrick Beukema (patrickb@allenai.org)

2023-11-10 13:42:18

https://www.skylight.global/

skylight.global

Skylight | Home

Skylight helps improve maritime transparency and protect our oceans.

Original URL: https://www.skylight.global/

👍:skin_tone_3: Alan Stenhouse

Violet Turri (vturri@andrew.cmu.edu)

2023-11-13 11:03:06

Hi Patrick! Thanks for the response, this sounds like a really interesting project. We’ll follow up over email to chat more.

Eric Orenstein (eorenstein@mbari.org)

2023-11-13 12:16:28

Job opportunity at the Turing Institute on Autonomous Systems for Biodiversity Monitoring. Help develop algorithms for adaptive sampling on autonomous underwater vehicles!

cezanneondemand.intervieweb.it

Senior Research Associate, Autonomous Systems for Biodiversity Monitoring

ENVIRONMENT AND SUSTAINABILITY GRAND CHALLENGEThe Alan Turing Institute’s new strategy, Turing 2.0, was officially launched in March 2023 with a foc

Original URL: https://cezanneondemand.intervieweb.it/turing/jobs/senior-research-associate-autonomous-systems-for-biodiversity-monitoring-36910/en/

🙌 Justin Kay, Carly Batist, Oisin Mac Aodha, Violet Turri, Katie Robinson, Shir Bar, Sara Beery, gvanhorn, Katriona Goldmann, Suzanne Stathatos, Ted Schmitt

😀 Michael Bunsen

Blair Costelloe (blaircostelloe@gmail.com)

2023-11-20 02:56:49

Would anyone be interested in trying BlueSky? I have a couple invite codes available

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2023-11-21 13:47:19

Sure. jason@wildme.org please.

Graham Wallington (graham.wallington@natureeye.com)

2023-11-23 10:12:46

I too would love to have a look.

Olof Mogren (olof.mogren@ri.se)

2023-12-11 04:12:26

I created a Bluesky feed on AI for climate change! https://bsky.app/profile/did:plc:fyw3wpo7owjjc362rp42upzo/feed/aaabbqsgt6bd2

Olof Mogren (olof.mogren@ri.se)

2023-12-11 04:18:06

@Jason Holmberg (Wild Me), @Graham Wallington, did you register? I cannot find you at Bluesky.

Vardaan Pahuja (vardaanpahuja@gmail.com)

2023-11-21 17:51:23

Is there a survey paper about different camera trap datasets out there? I am looking for a resource for all available camera trap datasets. I am wondering if there are any datasets that aren't covered by LILA.

Dan Morris (agentmorris@gmail.com)

2023-11-21 19:50:51

I don't know if this counts as "not covered by LILA", but on the "other datasets" page on LILA, I list a few other camera trap datasets in the section called "Terrestrial wild animal images (ground-based sensors)":

https://lila.science/otherdatasets#images-terrestrial-animals-ground

Vardaan Pahuja (vardaanpahuja@gmail.com)

2023-11-21 21:51:50

Thanks a lot!

Robin Ranabhat (robinnarsingha123@gmail.com)

2023-11-26 02:12:24

Apologies in advance if this is not the right place for this inquiry.
I recently realized this new field computational sustainability through works from professor Bistra Dilkina, Carla Gomes. I have decided this is the kind of work I want to do.
My current skill-set is limited to software and applied machine learning. But this endeavor requires having good foundation of various sub-domains of CS like mathematical thinking, Operations-Research e.t.c and out of CS.
It's would be really really helpful to know if there are relatable Graduate Programs that specifically focus on this.

Katelyn Morrison (kcmorris@andrew.cmu.edu)

2023-11-26 09:31:16

I haven't seen specific programs, but I have seen people apply to either CS-related programs or another domain (i.e., civil engineering or engineering and public policy) and then work with/get advised by professors who work on CompSust related topics. There are also lots of summer schools like CV4Ecology summer school and the Climate Change AI summer school. 🙂

Robin Ranabhat (robinnarsingha123@gmail.com)

2023-11-27 10:22:10

Thank you !!

Olof Mogren (olof.mogren@ri.se)

2023-11-26 06:56:36

Looking for people to follow on Bluesky! Looks like a promising alternative to Twitter. Here is my profile: https://bsky.app/profile/olofmogren.bsky.social

Bluesky Social

Olof Mogren (@olofmogren.bsky.social)

AI for climatechange and biodiversity. PhD. Head of deeplearning research at RISE in Sweden. Love trail running and snowboarding. Husband, proud dad.

Original URL: https://bsky.app/profile/olofmogren.bsky.social

Blair Costelloe (blaircostelloe@gmail.com)

2023-11-27 03:24:10

I still have a spare code if anyone wants to sign up!

❤️ Olof Mogren, Aakash Gupta

Olof Mogren (olof.mogren@ri.se)

2023-11-27 07:50:16

I have now created a public feed for AI for climate change! https://bsky.app/profile/did:plc:fyw3wpo7owjjc362rp42upzo/feed/aaabbqsgt6bd2

Sako Arts (sako@fruitpunch.ai)

2023-11-27 04:42:43

CALL FOR BEAR 🐻 DATA
Hi all,
Together with BearID bearid , ARM, Hack the Planet and NXP, we at FruitPunch will be organizing an bear identification Challenge. This Challenge will cover a couple of things: low-powered bear classification, bear detection, bear face detection and bear identification. While we have some datasets, we are looking for additional datasets to end up with a more robust model. Do any of you have or know of a dataset that includes annotated images of bears?
The annotations can be anything like, simply having wild bears in them, boundingbox around the bear, boundingbox around the bear face and identified individual bears.

Thanks in advance and have a very fruitful day! 🍉

:bearid: Ed Miller

Casey Clifton (caseyclifton@proton.me)

2023-11-27 04:45:15

Apparently there's a 'bear guy' in alaska with a huge number of images, but he's pretty off the grid. A friend was trying to contact him recently - ill check and see if he got through!

Sako Arts (sako@fruitpunch.ai)

2023-11-27 05:31:29

Oeh sounds exciting, would be great to get in touch!

Casey Clifton (caseyclifton@proton.me)

2023-11-27 04:49:30

Does anyone have experience or insights in the use of eDNA vs camera traps (or other 'traditional' sensing approaches) for monitoring biodiversity? Both have needs for AI, and im looking to pick an area to focus working on.

I've just been on a 3 month camera trap project covering 35 different sites, and having recently learned of eDNA am wishing we took soil samples instead.

In the short term there are issues of sequence data availability, costly lab services, etc, and technical challenges like DNA degredation, but im assuming they are solvable with sufficient time and resources improving the tech and sampling methods.

Are there any fundamental unsolvable limitations of eDNA (or anyone you might recommend i reach out to on the matter)?

Any thoughts are much appreciated 🙏 🙂

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-11-27 08:52:39

Any of the folks at NatureMetrics or SimplexDNA would be great for eDNA advice!

🙌 Casey Clifton

Talia Speaker (talia.speaker@wildlabs.net)

2023-11-27 13:21:53

My team at WWF has done some comparison studies on species captured, costs, etc. between the two methods - happy to connect you if it's of interest! https://www.nature.com/articles/s41598-021-90598-5

Nature

eDNA sampled from stream networks correlates with camera trap detection rates of terrestrial mammals

Scientific Reports - eDNA sampled from stream networks correlates with camera trap detection rates of terrestrial mammals

Original URL: https://www.nature.com/articles/s41598-021-90598-5

🙌:skin_tone_3: Alan Stenhouse

🙌 Sam Lapp

Casey Clifton (caseyclifton@proton.me)

2023-11-27 23:24:26

Thanks @Talia Speaker! I'll read this and come back to you

Riley Knoedler (mknoedler@west-inc.com)

2023-11-28 14:27:53

Mark Davis over at Illinois Natural History Survey is doing some interesting work with eDNA, just saw a presentation from him comparing the species found with eDNA vs camera traps (surprisingly little overlap!)

🙌 Casey Clifton

Casey Clifton (caseyclifton@proton.me)

2023-11-28 20:59:37

That's interesting @Riley Knoedler - do you have a link to the pres i could see?

Riley Knoedler (mknoedler@west-inc.com)

2023-11-29 10:58:11

@Casey Clifton It was at the REWI Solar Symposium conference so I don't have a way to distribute the presentation slides, but you might be able to discuss it with Mark! I'm not sure if that work is published yet.

👍 Casey Clifton

Paul Allin (allinpaul@gmail.com)

2023-12-06 08:27:37

We have just finished the data collection for comparison of eDNA methods (air, dust, and water ) to camera trap data. Happy to chat

🎉 Carly Batist

👍 Casey Clifton

mimi (arandjel@eva.mpg.de)

2024-01-17 05:28:27

here's our paper on iDNA and camera traps https://onlinelibrary.wiley.com/doi/full/10.1002/edn3.46
Biggest limitation in our mind is how to go beyond species detection and into population estimates.

👍 Ștefan Istrate, Carly Batist, Casey Clifton

Edwin Reed-Sanchez (ereedsanchez@gmail.com)

2023-11-29 13:38:55

Hello Everyone, I am a researcher at CUNY and looking to deploy an automated camera trap system. Ideally using wifi enabled camera (video/stills). I would prefer it over any Cellular based camera. Are there any cameras that you recommend? Thank you.

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-11-29 18:48:21

Oh hey I’m at CUNY too!! I’m a PhD student at the GC (Biological Anthro dept). Based out of Hunter though.

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-11-29 18:49:24

For your question - there are lots of good resources on WILDLABS about different trail cams, and it might also be a good place to re-post your question as well.

wildlabs.net

Feed | WILDLABS

Original URL: https://wildlabs.net/feed

Edwin Reed-Sanchez (ereedsanchez@gmail.com)

2023-12-01 13:21:23

Thanks!

Edwin Reed-Sanchez (ereedsanchez@gmail.com)

2023-12-01 13:21:37

What are you currently using/

Filippo Varini (fppvrn@gmail.com)

2023-11-30 08:19:55

Hello everyone,
In the last few months, I have been interested in the use of AI and Machine Learning to develop better Ocean Biodiversity Monitoring technologies.

I would love your help in exploring the field and I aim to share a report that would benefit the whole community!

I would like to map out all advanced marine biodiversity monitoring technologies. So far I investigated eDNA (and Metabarcoding), BRUVs, UVC, Scientific Fishing, and Bioacoustics. What else should I look into?

I am particularly interested in successful applications of such technologies in monitoring Ecosystem Health and the success of Marine Protected Areas. Do you know any related study or project? What papers should I read? What organisation should I look into? Who should I speak with?

Thank you so much!

🌊 Talia Speaker, Katelyn Morrison, Sara Beery

Florence Cuttat (florence.cuttat@nature-counts.org)

2023-12-06 10:24:09

Hi Filippo,
We are working on using AI for fish monitoring. --> streamocean.io
Let me know if you have some questions

✅ Filippo Varini

Olof Mogren (olof.mogren@ri.se)

2023-11-30 10:14:02

@Ben Weinstein talking right now 🙂 https://rise.zoom.us/j/208117140?pwd=aWJsbnIyai92RUk1cjcrcFMxWDROUT09

👏 Malte Pedersen

Olof Mogren (olof.mogren@ri.se)

2023-12-01 03:43:27

Here is the recording of this excellent talk! https://www.youtube.com/watch?v=7yXsFFbWgcs&list=PLqLiVcF3GKy0-jZFGg-VqLzh51LqCfduN&index=1

YouTube

RISE Research Institutes of Sweden (https://www.youtube.com/@RiSeSweden)

Ben Weinstein: General models for airborne wildlife detection

Original URL: https://www.youtube.com/watch?v=7yXsFFbWgcs&list=PLqLiVcF3GKy0-jZFGg-VqLzh51LqCfduN&index=1

🙏 Michael Bunsen

👍 Michael Bunsen

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-11-30 15:15:39

For any of you on more of the software side - still time to apply for our Software Engineer position at Rainforest Connection (RFCx) & Arbimon!💻

Join the team to help develop software for biodiversity monitoring. Fully remote position, open until filled. Full job description, responsibilities and requirements here & attached below.

To apply: please submit your resume and a cover letter to contact@rfcx.org with the subject “Position: Software Engineer” and tell us a bit about yourself!

❤️ Lucia Gordon

Devis Tuia (devis.tuia@epfl.ch)

2023-12-01 02:33:19

just a reminder for this beautiful position for Deep SDMs in my team!
If you feel like your next life step could be in Switzerland, please apply! We will be starting looking at the application in 10 days!

Pleasewrite me a msg if you have questions, happy to give info!

Devis Tuia (https://aiforconservation.slack.com/team/U032W85K4SD)

Hello everyone, I am hiring a postdoc in Species Distribution Models and Deep Learning in my Lab at EPFL! <a href="https://recruiting.epfl.ch/Vacancies/3158/Description/2">https://recruiting.epfl.ch/Vacancies/3158/Description/2</a> Position for 2 years, in the worndeful SDM team with @Nina van Tiel and @Robin Zbinden, and also in collaboration with @Benjamin Kellenberger. Dream team!!

👍 Oisin Mac Aodha, Justin Kay, Olof Mogren, Omiros Pantazis, Robin Zbinden, Lloyd Hughes, Yonghao Xu, Sara Beery, Thomas Radinger

Otto Brookes (otto.brookes@bristol.ac.uk)

2023-12-02 09:13:35

📣 A reminder that the deadline for the IET Computer Vision Special Issue: Camera Traps, AI, and Ecology is coming up! More details here!

🎉 Sara Beery, Jon Van Oast, Dante Wasmuht, Vardaan Pahuja

Vardaan Pahuja (vardaanpahuja@gmail.com)

2023-12-06 00:37:19

Hi, thanks for sharing this! I have a few questions, will send them in private chat.

✅ Otto Brookes

Piotr Tynecki (piotr@tynecki.pl)

2023-12-05 09:37:42

Greetings from :flag_pl: 🦬 🐗 🌳!

👋 Elizabeth Campolongo, Sara Beery

🦌 Kalindi Fonda

Fadel Seydou (fadel.seydou@gmail.com)

2023-12-06 07:57:29

Hi all, hope you're doing great.

@Paul Allin and myself are working on the topic of "automated aerial census of large herbivores in southern africa using ML". Our specificity is that we want to experimentally assess the influence of flight/data acquisition parameters on the accuracy/precision/recall of an ML-based wildlife detector.

We are currently facing a few issues from which we would appreciate your inputs:

How to label efficiently a large amount of data with unknown species and limited workforce? Our current approach is to use a yolov8s model (trained to localize herbivores wildlife on similar data) to obtain bounding boxes and then attribute classes manually. Do you see flaws in our approach or have recommandations?

Do you have ideas on how we can improve the generalization of our wildlife localizer?

How can we account for False Negatives (introduced by our wildlife localizer) in our labeling workflow?

Thank you🙌

Devis Tuia (devis.tuia@epfl.ch)

2023-12-06 08:09:06

Hi Fadel, I think the AIDE platform could help you out (and sorry everyone if I bring that one up often, it’s really not for bragging). With AIDE you can start labeling your images and use a pre-trained model tuning on those. Then the trained model will propose you candidate detections to review within an active learning loop. You can customize species as you go.. You can contact @Benjamin Kellenberger, who is the mastermind behind it.

➕ Sara Beery, Ben Weinstein, Robin Zbinden, Dan Morris, Caleb Robinson

Fadel Seydou (fadel.seydou@gmail.com)

2023-12-07 10:00:32

Thank you @Devis Tuia, I will look into it 😁

David Russell (davidrussell327@gmail.com)

2023-12-07 15:44:54

Hi Fadel, you may be interested in a tool that my supervisor Derek Young at UC Davis developed for scripting runs of the Agisoft Metashape photogrammetry software. He also has some work assessing the impact of flight parameters on tree detection using geometric methods: https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/2041-210X.13860.

open-forest-observatory/automate-metashape

Easy, reproducible Metashape photogrammetry workflows

Stars

Language

Python

💯 Fadel Seydou

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2023-12-14 16:37:13

Hi @Fadel Seydou, we're building our second round aerial flight detector now for the KAZA Transfrontier survey. Happy to share ideas. @Lasha Otarashvili is the primary on the training effort.

🙌 Fadel Seydou, Paul Allin

Joseph Dimos (joedimos@gmail.com)

2023-12-07 08:06:08

Greetings everyone! A bit about me and my ongoing work. I am working on data assimilation models with alignment in ocean/atmosphere dynamics that draws upon the notion of ‘online/offline’ learning and associated sub-models. In addition to this, I am working on defining an objective function for maps of dynamical snapshots. I working on some biodiversity problems too, namely those that employ deep learning. For instance, I am optimising some YOLO algorithms for post-processing of camera trap data. I am, in particular focusing on bees with YOLO. I am adapting a dynamical model that is deployed in a YOLO framework (with ResNet50 R-CNN) to learn the dynamics of an environment. This is done with the COCO dataset, to which I’m working with fairly deeply.

👏 Jonah Fox, Dan Morris, Jon Van Oast

Dan Stowell (dan.stowell@naturalis.nl)

2023-12-07 09:06:04

Job in Austria? Junior Group Leader, Machine Learning in Acoustics https://www.oeaw.ac.at/fileadmin/subsites/Jobs/ISF/ISF156JGL223.pdf

🔈 Oisin Mac Aodha

🎉 Jon Van Oast

Casey Clifton (caseyclifton@proton.me)

2023-12-08 03:14:54

Hello all, does anyone have any scripts/resources/papers/tips on fine-tuning MegaDetector (e.g. on images taken in a specific region) that they'd be open to sharing? Thanks 🙂

Peter van Lunteren (peter@addaxdatascience.com)

2023-12-08 03:26:43

Hi Casey, here are some tutorials:

https://www.kaggle.com/code/agentmorris/fine-tuning-megadetector

https://pub.towardsai.net/train-and-deploy-custom-object-detection-models-without-a-single-line-of-code-a65e58b57b03

The second one is more for adding custom labels (i.e., species identification), but can also be used for fine-tuning the existing labels (animal, person, verhicle).

kaggle.com

Fine-tuning MegaDetector

Explore and run machine learning code with Kaggle Notebooks | Using data from megadetector-tutorial

Original URL: https://www.kaggle.com/code/agentmorris/fine-tuning-megadetector

Medium

Train and Deploy Custom Object Detection Models Without a Single Line Of Code.

Let AI process your imagery through an open-source graphical user interface.

Reading time

9 min read

Original URL: https://pub.towardsai.net/train-and-deploy-custom-object-detection-models-without-a-single-line-of-code-a65e58b57b03

👍 Casey Clifton, Dan Morris

Casey Clifton (caseyclifton@proton.me)

2023-12-08 03:31:04

Thanks!

Dan Morris (agentmorris@gmail.com)

2023-12-08 12:21:36

Casey, are you trying to fine-tune MegaDetector in the sense of making it work better on species or camera angles it doesn't work well on, or in the sense of adding a species classification stage? Just a reminder that if you're referring to the latter, you may find it easier not to fine-tune MegaDetector at all, and instead train a classifier on the crops that come from MegaDetector. It's not obvious one way or the other, but if I had to place my bets on one approach, I would place my bets on the two-stage approach, and I think Peter's conclusion has been the same (Peter, don't let me put words in your mouth, but I vaguely remember you saying that).

If it's the former - making MD work on things it kind of works on, but not as well as you'd like - fine-tuning is a good solution. I've just done this for the first time recently (for a couple of reptile-heavy use cases), and haven't written up a tutorial yet, but can provide some tips and tricks if this is your scenario.

Somewhere in between would be the case where MD works well on your animals, but you want to reduce false positives for a particular ecosystem or set of cameras; I don't know that anyone has looked into this, but I think I would recommend a post-hoc classifier in this case (junk/not-junk) rather than fine-tuning. But I would only bother doing this if you have an egregious number of false positives.

Peter van Lunteren (peter@addaxdatascience.com)

2023-12-08 13:59:34

Yes, absolutely @Dan Morris. I've tried fine-tuning MD in the sense of adding a species classification stage, but found that that just costs lots of training data, processing power and yielded mediocre results. If you want species classification, I would definitely work with a custom classifier on top of MD. But if it is for improving detection rate (of the original classes animal, vehicle, and person) on "images taken in a specific region", it might be a good approach - but I haven't tried it out myself 😉

👍 Dan Morris

Casey Clifton (caseyclifton@proton.me)

2023-12-10 18:02:49

Hey @Dan Morris thanks for the info! I'm looking at the former. Specifically, using MD in an Australian setting, where im finding it misclassifies native plants as animals, and also just doesn't recognize some animals - there seem to be more false negatives than other papers have reported. Would love to hear of any tips or tricks you have for this.

Dan Morris (agentmorris@gmail.com)

2023-12-10 18:48:40

Gotcha. We may quickly get to the point where it's faster to do this by email (especially if we need to look at sample images), but I'll take a shot here, and if we move to email, I solemnly swear that if I don't report back to this Slack with whatever we learn, I owe everyone here a granola bar.

But, next questions:

Confirm that you're using MDv5?

Are you running MD via rundetectorbatch.py, via EcoAssist, or via something else?

What confidence threshold are you using?

Are the misses** reptiles, mammals, birds, or some other scary thing you have in Australia that we don't even have an American word for? Like poison-dragon-spiders, or poison-spider-dragons.

Are the misses** on things that are really obvious to your eye, or are they tail-sticking-into-the-corner kind of misses?

Very rough ballpark: about what percentage of your images are actually empty?

Very rough ballpark: about what percentage of the animals are being missed**?

Very rough ballpark: about what percentage of MD's positives** are false positives?

For (6)/(7)/(8), I know the real answer is "it varies a lot across cameras", but just try to average overall all your cameras.

**All those asterisks are there as a reminder that "positive", "negative", "miss", etc. are just placeholder words for now, they're undefined in this thread until we know the answers to (1)/(2)/(3).

😂 Casey Clifton

Casey Clifton (caseyclifton@proton.me)

2023-12-11 07:58:56

Happy to email / share pics / supply granola

Yep v5

Via rundetectorbatch.py

I've tried a few levels and if i just score a classification of animal vs not animal i get the following precision/recall values

Conf 0.2 = 0.74 / 0.76

Conf 0.5 = 0.91 / 0.7

Conf 0.8 = 0.97 / 0.53

Most of them are night shots of Quenda - the species we're investigating that is native to Australia. As far as i know they're not poisonous, but it's about the only thing that hasn't bitten us yet so who knows...

They are usually obvious when you look at multiple images in a row (we take 5 in 5 seconds for each trigger) because the animal is the only thing that's moving, but they're basically impossible to detect from a single shot with no context.

Has anyone looked into detection with temporal context to solve this? I think it'd be essential because Quenda are nocturnal, small, and fast, so they can be pretty blurry, easily hide behind a small bush, etc.

About 85% of the ones we've labelled so far are actually empty, but i know there are sites we havent labelled yet that are 99% empty

As per recall numbers above, about 25-45%

False positives were high but since removing repeating bounding box locations a lot of the FPs due to plants are removed so this is negligible at the moment

Dan Morris (agentmorris@gmail.com)

2023-12-11 12:01:29

Super-helpful. And also I just learned what a quenda is.

At the end of the day, if the thing you're looking for is only visible in image sequences, MD can't see it. So I think there are two questions here: (1) what are your possible paths forward for finding things that are only visible in multiple images? and (2) how can you get the most out of MD? I'll take a shot at (1) here, though hopefully others will chime in too, since I have no particular expertise there, then I'll address (2) in a separate reply.

Re: moving beyond single images...

AFAIK, there is no off-the-shelf tool that will help you find animals that are mostly only visible in sequences. There's Sara's work on Context-RCNN:

https://arxiv.org/abs/1912.03538
https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/context_rcnn.md

...but I would say that's solving a lot more than just the problem of finding a quenda scurrying in the bushes, and is therefore justifiably pretty complex, and I don't know of anyone that's deployed it.

If you are willing to get into the business of training custom models but don't quite want to leave the universe of RGB images (it's hard to overstate how much more complicated the world gets when you leave the world of RGB images, you lose 10 years of computer vision infrastructure development built around RGB images and video), I was peripherally involved with a paper that proposed a bunch of relatively simple ways to fit sequence information into RGB images... things like "swap out the blue channel and replace it with a motion channel":

https://arxiv.org/abs/2005.00116

I'm still a fan of that approach, and if you have enough data, it's relatively easy to explore (since you're still training RGB models). But again, AFAIK no one has deployed this. I think it would be particularly neat to train a detector using one of these approaches, which was outside the scope of that paper (which was focused on classifiers), but you would need to get a whole bunch of boxes, and then do this complicated research-y thing.

And although this might not help you now, if you can capture and label a bunch of videos, you might try training a custom model via Zamba Cloud:

https://www.zambacloud.com/

👍 Casey Clifton

Dan Morris (agentmorris@gmail.com)

2023-12-11 12:34:08

Re: getting the most out of MD...

FWIW, even for easy datasets, I never use a confidence threshold higher than 0.2 with MDv5, and for difficult datasets, I'll go as low as 0.05. So I think you have some room to reduce your confidence threshold, though I don't think it will immediately get your recall up from 0.76 to something super-respectable, because invisible is still invisible. But if 85% of your images are blank, IMO you have a lot of room to compromise on precision and still have AI save you lots of time, if you can get to adequate recall.

But the main tricks I use for difficult datasets are, from most to least important:

Repeat detection elimination... good for maybe a 2-3% precision bump at reasonable thresholds. For easy datasets, I typically spend about 5 minutes per million images on this, maybe for difficult datasets, this goes up to 10 minutes per million images. Once you get the hang of it, this will always increase precision with no impact on recall. This is particularly important for cases where you want to user super-low confidence thresholds.

Test-time augmentation, by doing inference via this script instead of rundetectorbatch. This will raise the confidence of everything a little, but it will particularly raise the confidence of low-confidence animals. I.e., when difficult animals are present, there is real signal there, it's not just raising the confidence of every detection. In principle this can improve both precision and recall, but in practice, I've only ever seen it at least slightly hurt precision, while always either helping recall or leaving recall the same.

Combining results from MDv5a and MDv5b, using this script.

Your case sounds like a difficult one that may be out of reach of one-image-at-a-time ML, but I will DM you to see what we can squeeze out of MD.

Casey Clifton (caseyclifton@proton.me)

2023-12-11 19:36:41

This is great, thanks. Will reply in DMs!

Talia Speaker (talia.speaker@wildlabs.net)

2023-12-08 12:03:19

📣 Important news from WILDLABS: Survey & Funding Opportunity! 📣

🌎 Help shape the sector: If you're developing or using AI or any other digital technologies for conservation, please take a few minutes to complete and share the 2023 State of Conservation Technology Survey. This helps capture what's working and what the needs are in the community, and also builds an evidence base for all of our work: https://colostate.az1.qualtrics.com/jfe/form/SV_e5kiopCmrZXX1KS

💵 Apply for funding: There are $60,000, $30,000 and $10,000 grants available for up to 14 conservation technology projects through the brand new WILDLABS Awards! AI for conservation projects are totally fair game. Closes January 14th: https://wildlabs.net/funding-opportunity/wildlabs-awards-2024-supporting-accessible-affordable-and-effective-innovation

colostate.az1.qualtrics.com

WILDLABS Conservation Tech Survey 2023

Contribute your experiences with conservation monitoring technologies on this 4th annual survey by WILDLABS & Colorado State University

Original URL: https://colostate.az1.qualtrics.com/jfe/form/SV_e5kiopCmrZXX1KS

wildlabs.net

WILDLABS Awards 2024: Supporting accessible, affordable, and effective innovation for nature | WILDLABS

With $60,000, $30,000, and $10,000 grants available for 14 outstanding projects, the support of engineering and technology talent from Arm (the leading semiconductor design company), and access to the world’s biggest conservation technology community, the WILDLABS Awards are a powerful opportunity for users and makers of technologies for nature to take their projects to the next level. Closes January 14th, 2024.

Original URL: https://wildlabs.net/funding-opportunity/wildlabs-awards-2024-supporting-accessible-affordable-and-effective-innovation

🎉 Carly Batist, Jon Van Oast, Suzanne Stathatos, Adrien Pajot, Piotr Tynecki, Ando Shah, Stephanie O'Donnell, Ed Miller

Talia Speaker (talia.speaker@wildlabs.net)

2023-12-08 12:24:33

For questions on the awards, reach out to our new program manager @Adrien Pajot 🙂

Ben Weinstein (benweinstein2010@gmail.com)

2023-12-08 13:41:46

Here is one for the community to process that I think gives us a good sense of where we are and where we need to go. I reviewed a ecological machine learning paper for 2 rounds at an unnamed but prestigious journal. Good paper, needed a bit of tightening. I asked for them to make data available. They posted it online without complaint. I recommended to accept the paper, explicitly saying "Reviewer #1: Thank you for making the data available. This is an excellent paper." Then in the published article they did not include the data link and put "Data will be made available on request." in the data availability statement, which is in separate form from what the reviewers see. Even though the data is currently still posted online (I went back to the response to reviewer). While lots of harmless explanations exist, it doesn't look great. As reviewers, keep your eyes out for this.

😢 Ștefan Istrate, Sara Beery, Caleb Robinson, David Russell, Subhransu Maji, Akash Nagaraj, Suzanne Stathatos, Enis Berk Çoban, Mitch Fennell, Toryn Schafer, Emilio Luz-Ricca

😮 Jon Van Oast, Michael Bunsen, Enis Berk Çoban

Dan Stowell (dan.stowell@naturalis.nl)

2023-12-12 12:36:43

I think you should raise this with the editors. The article should be amended with an erratum, since it deviates from the article that was accepted (whether by "accident" or not). The editors should be willing to handle this.

Ben Weinstein (benweinstein2010@gmail.com)

2023-12-12 12:37:16

ya, update here, editors said they are looking into it.

👍 Dan Stowell

Luke Sheneman (sheneman@uidaho.edu)

2023-12-08 17:42:53

For anybody working with video of wildlife, etc. I wrote a little convenience tool around YOLOv5 to extract video clips from source **MP4 video files that contain animals. It is intended to be used with models like MegaDetector_v5 to efficiently process large collections of videos to maximize overall throughput. It performs inference at specified intervals (e.g. every x seconds). Using multiprocessing, it can process multiple videos simultaneously and uses locking to allow safe concurrent access to the GPU. In reality, I/O will be your bottleneck with video....not YOLOv5 inference. It spits out clips to an output folder and builds a small metadata report file which describes the clips relative to their source video file and includes summary statistics on the confidence scores for the clips. Hopefully you can find it useful!

https://github.com/sheneman/tigervid

sheneman/tigervid

Scans video looking for frames which include animals, produced clips and a final report

Language

Python

Last updated

2 months ago

❤️ Suzanne Stathatos, Yseult Hb, Sara Beery, Charlotte Mallo, Stephanie O'Donnell

👍 Dan Morris, Shir Bar, Sara Beery, Joseph Dimos, Otto Brookes, Jon Van Oast, Aakash Gupta, Evan Eskew, Valentin Gabeff, Kakani Katija, Alasdair Davies, Martin Marzidovsek

👍:skin_tone_3: Alan Stenhouse

Ben Weinstein (benweinstein2010@gmail.com)

2023-12-08 17:47:16

Thats great, I've been hoping someone would do this and update. I still get alot of downloads of https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/2041-210X.13011, which is getting pretty old at this point. Can I start pointing users towards you? What kinds of models are in there to start?

Luke Sheneman (sheneman@uidaho.edu)

2023-12-08 17:54:24

That looks really cool! Its currently just a wrapper around YOLOv5, so in theory it will work with any YOLOv5 weights. I am using it with a collaborator from University of Michigan with MegaDetector_v5 to detect simple presence/absence of animals in a giant video collection but it should work with other weights (species classifier, etc.) as well with minimal changes.

Dan Morris (agentmorris@gmail.com)

2023-12-08 21:37:32

I have one super-important question: did you make that ASCII art tiger?

🙂 Jon Van Oast

😂 Stephanie O'Donnell

Joseph Dimos (joedimos@gmail.com)

2023-12-10 13:13:20

Working on a neural pipeline for YOLOv5 myself. That way, feature extraction is optimised for data assimilation tasks

Anna Willoughby (arwill19@gmail.com)

2023-12-31 14:13:14

oo will take a look at using this.

Edward Bayes (bayesbayes@gmail.com)

2023-12-13 13:17:43

Hi everyone and happy holidays! 🎄 It’s been so awesome to see the explosion in semi-supervised, transformer-based models in recent months from general models like SAM (and now EfficientSAM) to specialised models like BioCLIP. I’ve heard smatterings of conversations about applications in conservation here, but was wondering if anyone has done any thorough evals on how they compare in terms of accuracy and speed to larger supervised models like MD? If anyone is down to hop on a call I’d love to brainstorm!

P.S. I’ve created a demo to test such models in the browser using ONNX runtime (no backend). It has CLIP and MobileNet at the moment, but I’m planning on adding other models later to try to do head to head evals of various models.

❤️ Elizabeth Campolongo, Sara Beery, Andy Viet Huynh, Emilio Luz-Ricca, Antón Álvarez

👀 Stephanie O'Donnell, Jeremy Forest

Carly Batist (cbatist@gradcenter.cuny.edu)

2023-12-13 13:31:33

@Jason Holmberg (Wild Me) @Sara Beery

Antonio Ferraz (antonio.a.ferraz@jpl.nasa.gov)

2023-12-13 19:00:50

Hi Everyone; what are the best international conferences to learn about AI Foundational Models and RNN/LSTM Models in the context of Satellite Earth Observations for monitoring ecosystems (e.g. monitoring ecosystem change including land cover/land use change)? Thank you in advance!

❤️ Jon Van Oast

Ruben Remelgado (ruben.remelgado@gmail.com)

2023-12-14 02:54:02

Geo BON

Devis Tuia (devis.tuia@epfl.ch)

2023-12-14 04:02:04

EarthVision at CVPR (though the focus in on remote sensing applications at large rather than ecosystems)

✅ Antonio Ferraz

Gabriel Tseng (gabrieltseng95@gmail.com)

2023-12-14 10:31:39

This may also be of interest: https://aiforconservation.slack.com/archives/CLWGQ4BJ6/p1702567887929979

Gabriel Tseng (https://aiforconservation.slack.com/team/U014B1TAPAB)

Hi everyone! I wanted to share a workshop that might be relevant here: Machine Learning for Remote Sensing at ICLR 2024. It will take place in person in Vienna on May 11th, 2024, and have a virtual participation option. > The Machine Learning for Remote Sensing workshop will provide a platform for machine learning research that is relevant for applications in remote sensing and/or environmentally important. A panel of domain scientists from international organizations, such as the IAEA or the Red Cross will discuss the role of data and benchmarks in applied machine learning and our keynote speakers are leading researchers in the intersection of machine learning and remote sensing. We also invite you to submit to the workshop; the submission deadline for a 4-page non-archival research paper is February 03, 2024. The detailed call for papers is available <a href="https://ml-for-rs.github.io/iclr2024/#call-for-papers">here</a>

Original URL: https://aiforconservation.slack.com/archives/CLWGQ4BJ6/p1702567887929979

✅ Antonio Ferraz, Gedeon

Gabriel Tseng (gabrieltseng95@gmail.com)

2023-12-14 10:31:27

Hi everyone!

I wanted to share a workshop that might be relevant here: Machine Learning for Remote Sensing at ICLR 2024. It will take place in person in Vienna on May 11th, 2024, and have a virtual participation option.

The Machine Learning for Remote Sensing workshop will provide a platform for machine learning research that is relevant for applications in remote sensing and/or environmentally important. A panel of domain scientists from international organizations, such as the IAEA or the Red Cross will discuss the role of data and benchmarks in applied machine learning and our keynote speakers are leading researchers in the intersection of machine learning and remote sensing.

We also invite you to submit to the workshop; the submission deadline for a 4-page non-archival research paper is February 03, 2024.
The detailed call for papers is available here

👍 Justin Kay, Devis Tuia, Olof Mogren, David Russell, Ronny Hänsch, takatomi-k, Sara Beery, Gedeon, Chris Yeh, Robin Zbinden, Jason Holmberg (Wild Me), Nico Lang, Joseph Dimos, Thijs van der Plas

👋 Sara Beery, Gedeon, Chris Yeh, Jason Holmberg (Wild Me)

Miaomiao (mzhang@hbs.edu)

2023-12-14 22:33:18

Good evening everyone! I am sharing an exciting opportunity in Sustainability x AI. See the poster, a video, or the blurb here for detail:

Get ready to navigate solutions for a sustainable future using the power of generative AI! The action kicks off with a 36-hour virtual round on January 6-7th, 2024, and continues in person on January 18th, 2024 at Microsoft Toronto, where 8 finalist teams will be invited to present to a panel of industry leaders, which include leading VCs, AI, and sustainability experts. Teams of up to four participants are eligible to participate. Register your interest here.

Stay up to date by following our website for the latest announcements and event details. Should you have any questions or concerns, don’t hesitate to reach out to us at info@genaicompetition.com.

genaicompetition.com

The GenAI Competition

Join our hackathon style GenAI competition to create venture ideas to solve sustainability challenges and win summer internship opportunities and other prizes from Manulife and other corporate partners!

Original URL: https://www.genaicompetition.com/

Ben Weinstein (benweinstein2010@gmail.com)

2023-12-15 16:12:11

I know many people in the community use label-studio for machine learning annotations. I've been working on getting model predictions overlaid on a label-studio server all week to create an active learning pipeline. I can make a blogpost if this is something other people need to do, it wasn't trivial. Eventually it will go into DeepForest + Label-studio integration. So on image load, the annotator sees existing boxes with labels and scores, can select a set of them, delete, change the labels, and save as annotations. Message me if you need help with this.

👍 Sara Beery, Dan Morris, Caleb Robinson, David Russell, Shir Bar, Yseult Hb, Thor Veen, Nico Lang, Aakash Gupta, Cameron Trotter, Georgia Atkinson, Emilio Luz-Ricca, Lucia Gordon, Mike Trizna, abdon

👍:skin_tone_5: Prabath Gunawardane

👍:skin_tone_3: Alan Stenhouse

Dan Stowell (dan.stowell@naturalis.nl)

2023-12-18 12:16:51

Postdoc job in computer vision for nature, in Denmark, deadline 3rd Jan: https://international.au.dk/about/profile/vacant-positions/job/postdoc-for-computer-vision-and-deep-learning-applied-to-biodiversity-monitoring

👍 Georgia Atkinson, Justin Kay, Lucia Gordon, Sara Beery

Brandon Hays (brandon.hays@duke.edu)

2023-12-18 13:54:06

Hey folks - long time listener, first time caller!

Does anyone here have leads on automated identification of individual elephants in camera trap photos? I'm an ecology PhD student with limited tech know-how working on elephant ecology/conservation in Thailand. I'd love to be able to get higher accuracy population estimates. I know that there are successful species-level elephant ID algorithms. And I know people have been manually identifying individuals in photos with good accuracy. But that seems less tenable for areas with elephant populations >500 individuals.

Any suggestions appreciated!!!

Sara Beery (sara.beery@gmail.com)

2023-12-18 13:58:05

@Peter Kulits and I have a participatory system that helps collect labeled data with IDs with humans in the loop, if you don't yet have sufficient training data. One of the trickiest things in this space is that there is very little public data (just the ELPephants dataset, which is only 40 individuals and highly curated) due to limitations in data sharing.

🎉 Jon Van Oast

❤️ Jon Van Oast

Sara Beery (sara.beery@gmail.com)

2023-12-18 14:00:05

A qualitative study in our side seems to suggest that re-ID from camera trap data for many species is only possible for a very small portion of the data even by experts, vs. human captured DSLR data

Brandon Hays (brandon.hays@duke.edu)

2023-12-19 09:22:52

Hey @Sara Beery and @Peter Kulits! Hmm, yea data accessibility and image quality are tricky. Did you see this paper that the Plotnik research group published in March?
https://peerj.com/articles/15130/#aff-2

They're doing everything by human verification of videos from camera traps and getting reasonable seeming results! I was also thinking that, if you could afford it, some of that difficulty in getting high quality imagery from camera traps could be resolved by doubling cameras per location (similar to leopard folks trying to catch both sides of the leopard).

👍 Sara Beery

Brandon Hays (brandon.hays@duke.edu)

2023-12-19 09:25:31

Though they are also subsetting the images they analyze in that paper, and only using the high quality images. Still I feel like you could leave camera traps out long enough to accrue higher quality images for most of the elephants in an area, given enough time. I think the next step, modeling populations, would be trickier though given that variability in data availability.

Also, I don't have camera traps or images yet, but I bet I could find people with that data willing to share for a machine learning project!

Paul Allin (allinpaul@gmail.com)

2024-01-02 04:37:13

Not sure about indian elephant, but on african elephant we use tears and wrinkles in the ears, or the wrinkle pattern from the forehead to the mouth which is unique for each. This allows us to us images of lesser quality

Brandon Hays (brandon.hays@duke.edu)

2024-01-04 09:01:05

Hi @Paul Allin and @Sara Beery. Thanks for the tips! Could I use that particaptory labelling system that Sara mentioned a little ways down the road? I'm writing a proposal to Wildlabs for money to buy camera traps. Once I have traps, I'm planning to take them to elephant sanctuaries in Thailand. There are hundreds to thousands of captive elephants that free range through forests at least part of the year, with humans that know the elephants very well and can ID them with high confidence. Seems to me like the best possible way to build a training dataset!
And I'm writing into the grant that the dataset would be publicly available too!

Paul Allin (allinpaul@gmail.com)

2024-01-04 12:03:56

sounds like a great way to collect data of a large group of elephants. What is the end goal?

Brandon Hays (brandon.hays@duke.edu)

2024-01-04 13:27:50

the end goal is to be able to estimate populations in and around a protected area in eastern Thailand. I'm hoping individual ID will give better results than gridded camera trap arrays using more traditional analyses

Paul Allin (allinpaul@gmail.com)

2024-01-05 08:02:28

Would be good to chat, I have some ideas for something similar for African elephant and perhaps there is some overlap in ML

Brandon Hays (brandon.hays@duke.edu)

2024-01-05 09:10:22

Yea, I'd love to chat! Do you have any free time next Monday or Tuesday?

Paul Allin (allinpaul@gmail.com)

2024-01-05 09:18:04

Yes that should work, what time zone are you on?

Brandon Hays (brandon.hays@duke.edu)

2024-01-05 09:19:17

I'm in EST

Paul Allin (allinpaul@gmail.com)

2024-01-05 09:22:55

Okay I’m GMT+2

Paul Allin (allinpaul@gmail.com)

2024-01-05 09:23:11

Which platform do you want to use?

Brandon Hays (brandon.hays@duke.edu)

2024-01-05 09:28:20

let me shoot you a zoom link

Brandon Hays (brandon.hays@duke.edu)

2024-01-05 09:29:54

Could we 9am my time, 4pm your time?

Brandon Hays (brandon.hays@duke.edu)

2024-01-05 09:30:21

on Monday**

Paul Allin (allinpaul@gmail.com)

2024-01-05 09:41:25

Allinpaul@gmail.com, 4pm is fine with me

Oscar W (omtinez@gmail.com)

2023-12-18 16:13:40

Hi everyone! It's my pleasure to announce that we finally were able to release a dataset containing images of two species of sea stars: https://lila.science/sea-star-re-id-2023/

The paper that goes with it has been approved but needs some minor tweaks before it can be published (hopefully in the upcoming weeks), I'm happy to share the manuscript with anyone interested in the meantime. And of course, happy to answer any questions about the dataset.

LILA BC

Sea Star Re-ID 2023 - LILA BC

Overview This data set contains 1,204 images of 39 individual Asterias rubens sea stars (ASRU) and 983 images of 56 individual Anthenea australiae (ANAU) sea stars. For the ASRU data set, images were taken on five distinct days. For the ANAU data set, images were taken in three locations (sunlight, shaded, naturalistic exhibit) on the...

Written by

lilawp

Est. reading time

2 minutes

Original URL: https://lila.science/sea-star-re-id-2023/

❤️ Timm Haucke, Gracie Ermi, Jason Holmberg (Wild Me), Sara Beery

⭐ Dan Morris, Benjamin Hoffman, Maddie Cusimano, Cameron Trotter, Aran Dasan, Andrew Schulz

🤩 Maddie Cusimano

👍 Aamir Ahmad

Cameron Trotter (cater@bas.ac.uk)

2023-12-19 04:58:48

Hi @Oscar W - I'd be interested in giving the paper a read if you're happy to share

👍 Oscar W

Ana Maria Quintero (quinteroossa37@gmail.com)

2023-12-19 11:07:07

Hi everyone! I hope you are all doing well. I'm Ana María from Colombia :flag_co:, working as a Data Scientist. Last week, I came across this fantastic initiative at NeurIPS, and I'm eager to redirect my academic and professional focus toward the field of conservation. I would love to participate as an intern or volunteer in any available projects to gain experience and knowledge in conservation, particularly if they are based in Colombia or Latin America in general. This aligns with my goal as I prepare to apply to academic programs next year.
Thank you, and Happy Holidays! :)

❤️ Suzanne Stathatos, Catherine, Piotr Tynecki, Carly Batist, Arjun Subramonian (they/them), Jon Van Oast, Sara Beery, Andy Viet Huynh, Gustavo Perez, Talia Speaker, Shir Bar, Andrew Schulz, Iván Higuera-Mendieta, Aaron Ferber, Robin Zbinden, Gabriel Manso, Alan Stenhouse

Ben Weinstein (benweinstein2010@gmail.com)

2023-12-19 12:20:20

Hi Ana, I work with a number of Colombian and Ecuadorian teams. If you are near Medellin, highly recommend check out Juan Parra's lab at Antioquia. Happy to chat further if helpful.

Ana Maria Quintero (quinteroossa37@gmail.com)

2023-12-19 12:33:48

Amazing! Thank you so much, I will contact him

Jose Ruiz-Munoz (jfruizmu@unal.edu.co)

2023-12-20 11:25:40

Hi Ana, I am a faculty member at Universidad Nacional de Colombia. My areas of interest include machine learning and data analysis, with a focus on their applications in environmental monitoring. Please feel free to direct message me if you are interested in discussing potential projects

🙌 Ana Maria Quintero

Aaron Ferber (aferber@usc.edu)

2023-12-20 11:53:58

Also just wanted to say that I worked with Ana co-organizing the NeurIPS 2023 Latinx in AI workshop and she was extremely on top of things, eager to push projects forward on her own initiative, had great insights about the different research directions people were working on. Overall the social event we organized had around ~450 people registered which I think says something about her ability to manage large projects :)

❤️ Ana Maria Quintero, Tiziana Gelmi Candusso

🙌 Ana Maria Quintero

Atriya Sen (atriya@atriyasen.com)

2023-12-19 17:56:26

Hello all, I'm a computer science assistant professor at the U of New Orleans (from 2024, Oklahoma State U). I'm soliciting interest in joining us in a collaborative proposal with Queen's University in Belfast, from biologists in the US who are interested in the high-conservation-value demersal elasmobranchs; species that lay eggs (mermaids purses) and that utilize egg laying sites: for example large Skate species in the genera Dipturus, Bathyraja, etc, such as the Barndoor Skate (Dipturus laevis). Thank you.

Magali Frauendorf (magali.frauendorf@slu.se)

2023-12-21 06:15:22

For those interested: the Nordic Society Oikos conference in Lund (Sweden), March 12-15 2024, https://nordicsocietyoikos.glueup.com/event/nordic-oikos-2024-80737/, will have a thematic session on computer vision in ecology and evolution https://nordicsocietyoikos.glueup.com/event/nordic-oikos-2024-80737/thematic-sessions.html. The extended deadline for abstract submission is 12th January 2024!

Glue Up

Nordic Oikos 2024 | Nordic Society Oikos

WELCOME Velkominn // Velkommen // Tervetuloa // Velkommen // Välkommen We are delighted to invite you to the upcoming biennial meeting of the Nordic Society Oikos 12-14th of March 2024 with the theme New perspectives in ecological and evolutionary research. The meeting will be jointly organised by the Nordic Society Oikos, EEF and Lund University. The conference will be held at AF Borgen in the picturesque area Lundagård in Lund, southern Sweden.

Original URL: https://nordicsocietyoikos.glueup.com/event/nordic-oikos-2024-80737/

Glue Up

Thematic sessions ⇽ Nordic Oikos 2024 | Nordic Society Oikos

Original URL: https://nordicsocietyoikos.glueup.com/event/nordic-oikos-2024-80737/thematic-sessions.html

👍 Piotr Tynecki, takatomi-k, Sara Si-Moussi, Martin Marzidovsek, Tiziana Gelmi Candusso

❤️ Suzanne Stathatos, Shir Bar

Suzanne Stathatos (suzanne.stathatos@gmail.com)

2023-12-21 11:05:35

@Oskar Åström 👀

Burooj Ghani (buroojghani@gmail.com)

2023-12-21 11:28:26

Pleased to share this work with y'all. Enjoy the read! 🙂

https://www.nature.com/articles/s41598-023-49989-z

Nature

Global birdsong embeddings enable superior transfer learning for bioacoustic classification

Scientific Reports - Global birdsong embeddings enable superior transfer learning for bioacoustic classification

Original URL: https://www.nature.com/articles/s41598-023-49989-z

👍 gvanhorn, Suzanne Stathatos, Georgia Atkinson, Benjamin Hoffman, Dan Morris, Timm Haucke, Ana Maria Quintero, Jason Holmberg (Wild Me), Justin Kitzes, Ghazi Randhawa, Chris Lange

🐦 Holger Klinck, Dan Morris, Taiki Sakai - NOAA Affiliate, Yseult Hb, Jason Holmberg (Wild Me), Takumi Sato, Namitha Suresh, Timm Haucke, Alan Stenhouse

😊 Maddie Cusimano, Jason Holmberg (Wild Me)

🙌 Ben Williams

Dan Morris (agentmorris@gmail.com)

2023-12-22 19:12:21

A few weeks ago @Ian Ingram and I had a chat on this Slack that went something like this (paraphrasing):

__

Ian: "Is there a model zoo for camera trap models where we can publish our models and try other people's models?"

Me: Model zoos are hard, blah blah, compatibility, blah blah, general tone of discouragement and doom.

Ian: "Well what if it was more like just a list of models and instead of a 'model zoo', we called it a 'model wilderness'?"

Me: Oh yes, brilliant then, I like lists and I'm going to start saying "model wilderness".

--

So... I finally got around to collecting a list of all the models I'm aware of for camera trap data, in the sense of publicly-available models you can download as model weight files. I.e., this particular list does not include models that exist only in specific platforms, unless the weights are downloadable.

https://agentmorris.github.io/camera-trap-ml-survey/#publicly-available-ml-models-for-camera-traps

I'm not sure if that's what a "model wilderness" is, but it's a start. I'm not trying to make something particularly structured, but it's not that far from a structured collection of models with automatic download from the sources, sample code, some model-card-like information, etc. If someone wants to formalize that and put together code to actually run all those classifiers, that would not be a ton of work, and it would start to be somewhere between a "model wilderness" and a "model zoo". A "model park" maybe?

Let me know what's missing from that list!

🙌 Casey Clifton, Suzanne Stathatos, David Russell, Taiki Sakai - NOAA Affiliate, Aakash Gupta, Sara Beery, Mitch Fennell, Shir Bar, Gaspard Dussert, Felipe Parodi, Mohamed Belja, Malte Pedersen, Carly Batist, Giacomo May, Matthias Zuerl, Aran Dasan, Elizabeth Campolongo, Jason Holmberg (Wild Me), Ștefan Istrate, Mike Trizna, Timm Haucke, Joseph Dimos, Edward Bayes, Catarina Silva, Talia Speaker, Rebecca Wilks, Tiziana Gelmi Candusso, Olivier Dietrich, Lauren Harrell

🎄 Aakash Gupta, Jason Holmberg (Wild Me), Sam Lapp

🙌:skin_tone_5: Prabath Gunawardane

😎 Jon Van Oast, Elizabeth Campolongo, Jason Holmberg (Wild Me), Barry Brook, Joseph Dimos

👍 Thor Veen, Piotr Tynecki, Vincent Christlein, Ed Miller, Joseph Dimos, mimi

💯 Gaspard Dussert, Valentin Gabeff, Chris Yeh, Tiziana Gelmi Candusso

👍:skin_tone_3: Pen-Yuan Hsing

👍:skin_tone_2: Cara Appel

🙌:skin_tone_3: Alan Stenhouse

❤️ Alan Stenhouse

🦏 Ian Ingram

Aakash Gupta (aakash@thinkevolveconsulting.com)

2023-12-22 20:30:46

Hi Dan you could also probably add the EU Rewilding project. Model weights are uploaded on HF

HF Spaces

https://huggingface.co/spaces/skylord/European-ReWilding-Demo-Yolov8

Model weights

https://huggingface.co/skylord/ReWilding-Europe-Yolov8

huggingface.co

European ReWilding Demo Yolov8 - a Hugging Face Space by skylord

Discover amazing ML apps made by the community

Original URL: https://huggingface.co/spaces/skylord/European-ReWilding-Demo-Yolov8

huggingface.co

skylord/ReWilding-Europe-Yolov8 · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Original URL: https://huggingface.co/skylord/ReWilding-Europe-Yolov8

Dan Morris (agentmorris@gmail.com)

2023-12-22 20:51:41

Added, thanks!

I tested it on an image with deer from my backyard, it did the right thing.

❤️ Jon Van Oast

🦌 Mitch Fennell, Shir Bar, Aakash Gupta

🎄 Aakash Gupta

Piotr Tynecki (piotr@tynecki.pl)

2024-01-08 02:06:40

@Dan Morris feel free to update the list of the models for mammals with the paper:

Recognizing European mammals and birds in camera trap images using convolutional neural networks

Two models based on EfficientNetV2 and ConvNextBase + test set are here:
https://data.uni-marburg.de/handle/dataumr/246
https://github.com/umr-ds/Marburg-Camera-Traps

Paper:
https://inf-cv.uni-jena.de/wordpress/wp-content/uploads/2023/09/Talk-8-Daniel-Schneider.pdf

Dan Morris (agentmorris@gmail.com)

2024-01-08 16:30:45

@Piotr Tynecki Done, thanks! Sample code looks very straightforward. Also added your dataset to this list.

Piotr Tynecki (piotr@tynecki.pl)

2024-01-09 03:38:48

(it’s not my paper/model/code but thanks!)

Sam Lapp (sam.lapp@pitt.edu)

2024-01-12 18:41:51

On the bioacoustics side of model zoos/parks/wildernesses, we’ve been developing the “Bioacaoustics Model Zoo” here. It’s admittedly OpenSoundscape-focused, but it does include APIs for generating predictions or embeddings with Perch and BirdNET. In addition to adding more models to the zoo, the section of the ReadMe “Other automated detection tools for bioacoustics” could be expanded to fit the spirit of “model wilderness” - I’d be happy to work or collaborate on that effort (or port it to another location) if @Dan Morris or others think such a list is needed on the acoustics side of thiings

🙌 Suzanne Stathatos, Matt Weldy, Carly Batist, Dan Morris, Shir Bar, Mitch Fennell, Enis Berk Çoban, Sara Beery, Andrew Schulz, Thijs van der Plas, Antón Álvarez, Ben Williams, Lauren Harrell

😎 Jon Van Oast, Maddie Cusimano, Tiziana Gelmi Candusso

👀 Alba Márquez-Rodríguez, Meredith Palmer

Dan Morris (agentmorris@gmail.com)

2024-01-12 20:51:30

@Sam Lapp Yes I really like your model zoo! You did not just the easy version, but the proper model zoo that I claimed earlier on this thread is very hard. 🙂 Coincidentally I played with your model zoo for the first time last week, and made my first PR to your repo. Worlds collide!

👍 Sam Lapp

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2023-12-29 16:27:42

Hi everyone! Continuing our trend from last year, this time also we have created a 2023 Year in Review Playlist of our group's work (Flight Robotics and Perception Group at Uni Stuttgart). A large part of it is likely of interest for this slack workspace. Enjoy it here : https://www.youtube.com/playlist?list=PLZM-Zi7aahIvGecEUjNafPWhb2pYuwBVO

YouTube

FRPG's 2023 Year in Review

Flight Robotics and Perception Group's 2023 Year in Review

Original URL: https://www.youtube.com/playlist?list=PLZM-Zi7aahIvGecEUjNafPWhb2pYuwBVO

😎 Timm Haucke, Alen Lin

👍:skin_tone_3: Alan Stenhouse

Alen Lin (alenlin2752@gmail.com)

2023-12-30 06:04:46

Hi everyone, I'm Alen from Taiwan! Although I'm a high school student, I have a deep passion for birds (and any and all things related to them 🦉), really excited to join this slack feed. I'm currently working on a project which utilizes computer vision to track migratory patterns of birds and analyze the impact of climate change on annual migration (alongside some other small ideas!) and am deeply interested in participating in larger research or volunteer programs, particularly over the summer. I would love to pick your collective brains on what's available to a motivated high school student like myself, or if there are any organizations you recommend I reach out to. Thank you in advance, and I look forward to reading about all of the papers, workshops and competitions you've shared in here dad_parrot

🎉 Kai Hung, Dan Morris, Takumi Sato, Isabella, Alan Stenhouse

👋 Viktor Domazetoski, Mitchell Rogers, Shir Bar, Sara Beery, Edward Bayes, Talia Speaker, Priscilla Ye

Dan Morris (agentmorris@gmail.com)

2023-12-31 19:05:14

Wow, in high school and already so motivated... when I was in high school, I think my biggest achievement was beating Final Fantasy (the NES version, in case anyone else here is old enough to make that distinction) in one continuous, pizza-fueled session. And here you are training AI models to analyze the impact of climate change on bird migration. Kudos.

And welcome to this community! If you're looking to "meet" (in the 2023 sense of the word) more folks doing this kind of work, in addition to hanging out on this Slack, I recommend hanging out at as many WILDLABS events as you can; many of us are big fans of the Variety Hour series as a way to get a feel for what's going on in this field that's as close to in-person as one could get without teleporting everyone on this Slack into one room:

https://wildlabs.net/article/variety-hour-2023-lineup

❤️ Talia Speaker

Dan Morris (agentmorris@gmail.com)

2023-12-31 13:33:08

A couple posts ago I shared a link to a list of publicly-available models for camera trap images, I am also trying to put together an analogous list of models for detecting fish in underwater images/video (same criteria... listing downloadable model weights):

https://github.com/agentmorris/agentmorrispublic/blob/main/fish-datasets.md#publicly-available-models-for-fish-detection

Let me know if I'm missing stuff there too?

👍 Justin Kay, Jason Holmberg (Wild Me), Edward Bayes, Thor Veen, Sara Beery, Luke Sheneman, Catarina Silva, Levi Cai

🙌 Anna Willoughby, Jason Holmberg (Wild Me), Shir Bar, Sara Beery, Carly Batist, Kalindi Fonda

🎉 Jon Van Oast, Sara Beery, Tiziana Gelmi Candusso

🐟 Malte Pedersen, Aran Dasan, Shir Bar, Sara Beery, Meredith Palmer

👍:skin_tone_3: Alan Stenhouse

Edward Bayes (bayesbayes@gmail.com)

2024-01-05 11:14:41

Love this recent work compiling these lists Dan, super useful! Do you know of any similar lists for birds (from terrestrial, not aerial, images), insects, plants, or reptiles/amphibians? Or know of any evals of how MD performs for these (I imagine it’s only really performant on birds given the training dataset)?

If these don’t exist, I'm doing some research in this area and happy to compile a first draft to add to the ‘wilderness’.

Dan Morris (agentmorris@gmail.com)

2024-01-05 14:17:42

Sorry, I don't know of similar lists for other terrestrial modalities/taxa.

I don't know of formal evaluations of MD on birds and reptiles, though subjectively I would say for birds, MD works almost all well as it does for mammals, but not quite, and for reptiles and amphibians, it can be anywhere from "excellent" to "a total catastrophe".

❤️ Edward Bayes

Edward Bayes (bayesbayes@gmail.com)

2024-01-05 15:05:00

"anywhere from "excellent" to "a total catastrophe"." - love it! 😂

Edward Bayes (bayesbayes@gmail.com)

2024-01-05 15:05:06

And thanks so much! This is really useful

Sam Lapp (sam.lapp@pitt.edu)

2024-01-02 08:25:30

Job opportunity: KAUAI AVIAN RESEARCH/MANAGEMENT COORDINATOR
Deadline to apply: extended to Jan 5 (soon!)
Reposting from an email from project coordinator Cali Crampton
(I’ll post the description in a reply thread because its long)

❤️ Sara Beery, Suzanne Stathatos

Sam Lapp (sam.lapp@pitt.edu)

2024-01-02 08:25:50

KAUAI AVIAN RESEARCH/MANAGEMENT COORDINATOR – ID# 223806.
CLOSING DATE: January 5, 2024. INQUIRIES: Lisa Crampton 808-335-5078 (Kauai).
Regular, Full-Time, RCUH Non-Civil Service position with Pacific Cooperative Studies Unit (PCSU), Kauai Forest Bird Recovery Project (KFBRP) located in Hanapepe, Kauai. Continuation of employment is dependent upon program/operational needs, satisfactory work performance, availability of funds, and compliance with applicable Federal/State laws.
MONTHLY SALARY: $5,140/Mon.
DUTIES: Leads planning, organization, and implementation of research, monitoring, and management projects on Kauai for the benefit of native forest birds, with particular focus on threatened and endangered forest bird species. Performs and coordinates all project activities including research, conservation management, field work and logistics, budget tracking, regulatory compliance. Liaises with partners to facilitate research activities, ex situ conservation, control of introduced species and/or habitat restoration in study areas. Recruits and provides oversight, instruction, and guidance for a team of up to ten (10) members (including staff, interns, and volunteers). Produces annual work plans and reports for submission. Ensures that proper environmental compliance documentation is prepared for all projects and that all permits and regulatory approvals and permits are obtained. Coordinates and performs analysis of field data with guidance from Program Manager. With Program Manager, develops proposal budgets and writes grant proposals. Carries out routine awareness-raising events to ensure ongoing project funding. Initiates and conducts public outreach efforts, working with media, community organizations, civic leaders, and individuals through an effective program using personal contact, media briefings, brochures, press releases, presentations, and public service announcements. May travel to other islands for site visits and fieldwork. Drives to project activities and field work locations.
PRIMARY QUALIFICATIONS:
EDUCATION
Master’s Degree from an accredited college or university in Biological Sciences, Natural Resource Management, Biological Conservation, or related field. (Bachelor’s Degree from an accredited four (4) year college or university and at least two (2) years of independent research in conservation of endangered avian species may substitute for a Master’s Degree.).
EXPERIENCE
Three to five (3-5) years of experience conducting biological research, ornithology, invasive species management, or conservation management. Experience includes one-three (1-3) years of experience mist-netting, banding and taking blood samples from passerines; conducting surveys of plants and animals; and locating and monitoring bird nests. Experience includes one to two (1-2) years of supervisory or team oversight/leading experience.
KNOWLEDGE
Detailed knowledge of the principles and techniques of conservation management, remote field operations, and avian species biology. Working knowledge of natural history relevant to native Hawaiian wildlife, or similar environments. Working knowledge of native Hawaiian ecosystems, native Hawaiian plants and wildlife, or similar environments. Detailed knowledge of techniques used to inventory and monitor wildlife, and other natural resource assets, including experimental design and data analysis. Working knowledge of rules and regulations relating to field operations, and pertinent laws, regulations, licensing and permitting requirements related to program. Knowledge of management principles including, but not limited to, supervising/developing employees, EEO, workplace safety, corrective/disciplinary actions, and administration of policies and procedures.
ABILITIES & SKILLS
Strong organizational ability to plan, lead, and execute logistically complex field operations. Strong ability to solve logistical problems and innovate solutions related to biological threats. Ability to lead a field crew and work as a team member for safe and efficient field operations. Ability to maintain a positive professional attitude in support of a productive work environment. Must have excellent communication and program management skills. Ability to perform and coordinate data analysis and present findings and recommendations in written report format. Ability to estimate costs associated with complex projects and to project budget requirements for future program needs. Proficient use of Microsoft Office word processing and spreadsheet programs, databases (e.g. Access), GIS (e.g., AGOL), and statistical software (e.g. R or SAS). Must possess a valid driver’s license (and if use of personal vehicle on the job is required, must also have valid personal driver’s insurance equivalent to Hawai’i’s No-Fault Driver’s Insurance) and maintain throughout the duration of employment. Post Offer/Employment Conditions: Must possess the American Red Cross Certification in First Aid/CPR. Must be able to complete basic helicopter safety course (A100) and external sling load (A219) course within twelve (12) months from date of hire and become certified as a PCSU Helicopter Manager when training is offered. Must be able to complete chainsaw training within twelve (12) months from date of hire and maintain throughout duration of employment. Must be able to qualify for DOFAW’s Bird Banding Laboratory (BBL) subpermit to band and take blood from threatened and endangered passerines. Must be able to complete the UH Information Security Awareness Training (ISAT) within two (2) weeks from date of hire, and re-certify every twelve (12) months.
PHYSICAL/MEDICAL DEMANDS
Must be able to conduct fieldwork in dense vegetation and remote areas under difficult conditions (e.g., heat, rain, cold temperatures, poor footing). Must be able to hike over difficult terrain and long distances of eight to ten (8-10) miles with a backpack weighing up to forty (40) pounds unassisted. Must be able to work around and fly in approved helicopters.
POLICY/REGULATORY REQUIREMENT
As a condition of employment, employee will be subject to all applicable RCUH policies, procedures, and trainings and, as applicable, subject to University of Hawai’i’s and/or business entity’s policies, procedures, and trainings. Violation of RCUH’s, UH’s, or business entity’s policies and/or procedures or applicable State or Federal laws and/or regulations may lead to disciplinary action (including, but not limited to possible termination of employment, personal fines, civil and/or criminal penalties, etc.).
SECONDARY QUALIFICATIONS:
Management, public relations, and administration skills. Knowledge of funding sources in Hawai’i and nationally. Proven grant writing and fund-raising abilities. Demonstrated ability in publishing peer-reviewed scientific papers. Experience coordinating with land managers in Hawai’i. Familiarity with the Alakai Wilderness Area, Kauai and its native birds. Expertise in database design and management. Experience in reintroduction or translocation, or founding of captive populations of passerine birds. Aviculture skills, including husbandry and fluid administration. Experience monitoring animal movements using radio-tracking.
APPLICATION REQUIREMENTS: Please go to www.rcuh.com and click on “Job Postings.” You must submit the following documents online to be considered for the position: 1) Cover Letter, 2) Resume, 3) Supervisory References, 4) Copy of Degree(s)/Transcript(s)/Certificate(s). All online applications must be submitted/received by the closing date (11:59 P.M. Hawai’i Standard Time/RCUH receipt time) as stated on the job posting. If you do not have access to our system and the closing date is imminent, you may send additional documents to rcuh_recruitment@rcuh.com. If you have questions on the application process and/or need assistance, please call (808)956-7262. Please visit https://www.rcuh.com/document-library/3-000/benefits/rcuh-benefits-at-a-glance/ for more information on RCUH’s Benefits for eligible employees.

RCUH’s mission is to support and enhance research, development and training in Hawai’i, with a focus on the University of Hawai’i.

We are an equal opportunity employer and all qualified applicants will receive consideration for employment without regard to race, color, religion, sex, sexual orientation, gender identity or expression, pregnancy, age, national origin, disability status, genetic information, protected veteran status, or any other characteristic protected by law.
--
Dr. Lisa “Cali” Crampton
Program Manager
Kauai Forest Bird Recovery Project
PO Box 27 (USPS mail) or 3751 Hanapepe Rd (courier packages)
Hanapepe HI 96716
808.335.5078

Ghazi Randhawa (muhammadghazirandhawa@gmail.com)

2024-01-02 18:39:43

@Sidhika

Patrick Beukema (patrickb@allenai.org)

2024-01-02 20:15:10

I know that last year was hard on the planet -- but there was also a lot of innovation and progress in environmental AI and I was inspired by the excellent talks at NeurIPS (especially Sara, David, and Zaira's, just to name a few). I wrote a short recap of some key takeaways in AI4Earth throughout the year. This is not exhaustive by any means, and it is certainly opinionated, and of course any feedback is welcome.
https://blog.allenai.org/ai-for-earth-2023-in-review-49a27cb731a8

Medium

AI for Earth, 2023 in Review

It was not an easy year for our planet. But we have also seen breakthroughs in Environmental AI that are reasons for hope and optimism.

Reading time

7 min read

Original URL: https://blog.allenai.org/ai-for-earth-2023-in-review-49a27cb731a8

❤️ Suzanne Stathatos, Sara Beery, Ted Schmitt, Shir Bar, Mike Gartner, Konstantin Klemmer, Edward Bayes, Shravan Ambudkar, Omiros Pantazis, Devis Tuia, Robin Zbinden, Chris Llorca, Joseph Dimos, Vinicius Amaral, Alexander Merdian-Tarko, Alan Stenhouse, Chris Lange

💕 Jon Van Oast

Patrick Beukema (patrickb@allenai.org)

2024-01-02 20:17:25

I wrote this somewhat spontaneously and last moment and wanted it to get out there before the new year -- but I think it would have been much better had it been a more collaborative/jointly written especially to offer more diverse perspectives.

Devis Tuia (devis.tuia@epfl.ch)

2024-01-03 05:27:28

THanks @Patrick Beukema! It is important to keep an optimistic drive and your recap is a great read to start the year with a lot of positive energy!

❤️ Patrick Beukema

Ben Weinstein (benweinstein2010@gmail.com)

2024-01-03 13:05:31

I get 1-3 requests for review a week at this point. I'm considering a blanket approach that if the journal doesn't require data availability, then I won't review. I encourage others to join me. I'm trying to decide how stringent and reasonable the requirement should be. I think my minimum is that if within 2 min I can find a paper I'd want to read in the latest issue, but the data "Is available on request", I won't review. Use that and the guide for authors as an indicator. As a survey in my inbox I have a request from Ecological Indicators, Biogeosciences, International Journal of Applied Earth Observation and Geoinformation. Ecological Indicators has the typical weak data availability statement (https://www.elsevier.com/researcher/author/tools-and-resources/research-data/data-statement), and I found an interesting paper within 15 seconds. The other journals appear equally vague. Are there other criteria I should use? Is this fair? Does it bias towards more expensive journals and have any other unintended consequences? Of course there are legal and logistical reasons not to share data, but those are the vast minority of cases. A reasonable explanation in the data availability section is usually sufficient in those cases. FYI, credit to Methods in Ecology and Evolution, which has probably the most comprehensive statement (https://besjournals.onlinelibrary.wiley.com/hub/editorial-policies)

Data are important products of the scientific enterprise, and they should be preserved and usable for decades in the future. The British Ecological Society thus requires, as a condition for publication, that all data supporting the results in papers published in its journals are archived in an appropriate public archive offering open access and guaranteed preservation. For theoretical papers the underlying model code must be archived.

sciencedirect.com

Estimation of chlorophyll content for urban trees from UAV hyperspectral images

Urban trees provide important ecosystem services to improve cities’ liveability and sustainability. Leaf chlorophyll content (Cab) estimation by remot…

Original URL: https://pdf.sciencedirectassets.com/272637/1-s2.0-S1569843223X00107/1-s2.0-S1569843223004417/main.pdf?X-Amz-Security-Token=IQoJb3JpZ2luX2VjEMr%2F%2F%2F%2F%2F%2F%2F%2F%2F%2FwEaCXVzLWVhc3QtMSJHMEUCIQCeq8rVfsuD6kwwap4Ean3c0FY3meO2LFH5h3VNDGTmrwIgeJRmVgOoYa%2FqQNAuNXTRsadLxrrvngtSpvhg01YUaQIqsgUIYhAFGgwwNTkwMDM1NDY4NjUiDEAf6oTFFVTCaYfxCSqPBRC4hAI2d76ydyauTgLwx4Pp%2FYoxeYRP6loc2AGgUNYMGv3kIH%2FPdOIiJtmj1XhOuX446yOAU19QvMj1RRHKr3fCLemamb%2FbbOQ7%2BHmwY%2BR4MctxGMYRr1QV0H%2BMp7EKjgR6e4%2FhWmZ8KlEYFvvK9mnAIPLpIhF%2FTJyKsdX92tST%2BZk7kkKPLY45q9%2FpR3NwvQGv0T73qRvgYKmZOBr3gKC%2BEpddks8WEKJsJER3ROxAcLk9e%2BUljBapaBZS%2FZ3p4YrM7yQCCgZXfaD0ARpoChlDH0KptPOZ2O2WSMWzZK72I%2Bm%2BElYwBluP%2Fai9Lg7ZV2vnxoLX1VEoMHN3%2Fes%2FaSsUTHVVckUsjnNfAGPf7NAZ2bxUa5SgyEpsnbGss6tSpdUhIPini4NZzlewKtg6FizaZR1oi8WBpnrhIhF9OHQGFLcRwe0SPYbyvdiEcwcPcaRsrb4yD5aFs%2BXPY7Q90H1SMAQ3dUM7iJOj%2F4BOVJq7DaYb0m30zbdeiMLezfpnvav87bsItHLU1mD%2BTnY6ZlcFrVA%2F0sd0591Fm%2FHqyLjAVsyESm7gDxzERvTn9b%2F7ZNlnvsMHJR0ggkPJ1e%2FjlM%2FRpVX1v6P70vgYgB9p8uFLKCEMLw28Hbx1dphGyLP0Bbtayjcu6kqpQRWgAfFL%2F8pmpqcQEnMx0771mKRMa%2BYkqnuYN%2BBu%2BbUaDEBJHFR94uFiZhcPGRvxFzsYxdObl8%2Ba2vr8GEiT4mjwoNFAEoTiDcyUbrx1d0wdwbnese%2F95M%2FPD7OM%2BNyXSkHEXmHEf8D%2BuW55L1Qn3xLHp6LASweosb7zK0umAjJrVDVgUb1GPg%2FODODmn0IeBGcphF6cSeaaisIZim6GjJ4EXBPCpK4wsKzWrAY6sQH0jHx7chb%2F7WUXKyvpXrgXT0Wab%2BGhz%2FDUqLC8RczuILfIFeES%2FC8cnPB0Z4EUwAFUirysNIiDgAC2mLcEqy0Ri8YFvOrRni6HlAWAN0EKARkmvwaRldRJY7kBqHrSnBX9lLkNiccvoaIoDfkEyFWghQ43Nq4MlTVy%2FdHNSr9Dnxeh1NHrP7kajCupqYZHVVaqeA3tSIIVlx6hUnYwBpshTwBAJI%2F3blSkY3KECI%2Fs8A8%3D&X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Date=20240103T175754Z&X-Amz-SignedHeaders=host&X-Amz-Expires=300&X-Amz-Credential=ASIAQ3PHCVTYVDTMZKDA%2F20240103%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Signature=bf2c855ca6f7e0e03b1d16ee7616c8d58ca58234ac089eda3a59e84691c3e713&hash=e25bf956b2d66971b4143a387f007733879bec598becce1e45a7aec94ec08ed4&host=68042c943591013ac2b2430a89b270f6af2c76d8dfd086a07176afe7c76c2c61&pii=S1569843223004417&tid=spdf-101f8213-f738-4c37-8ada-35d5d0f696d0&sid=1a0126d47157b040387ab0d29247e2cf0c19gxrqa&type=client&tsoh=d3d3LnNjaWVuY2VkaXJlY3QuY29t&ua=17105d520005545e54&rr=83fd20774857ef98&cc=us

🎉 Jon Van Oast, Ariel Chamberlain, Sara Beery, David Russell, Jason Holmberg (Wild Me)

👏 Leonardo Viotti, Enis Berk Çoban, Dan Morris, Rowan Converse, Jason Holmberg (Wild Me), Martin Marzidovsek

Caleb Robinson (calebrob6@gmail.com)

2024-01-03 16:05:21

I think this is reasonable -- (lack of) dataset availability came up several times for me while reviewing for NeurIPS Datasets and Benchmarks and AAAI AI for Social Impact tracks. It is impossible to properly review some papers without data availability.

Dan Morris (agentmorris@gmail.com)

2024-01-03 21:47:08

I generally agree with what Ben says (everyone is almost certainly tired of hearing me shout about public data on this Slack), but I have a slightly softer approach as a reviewer...

If there's a good reason that data can't be released, and it's either mentioned specifically or it's self-evident, I don't totally ignore the issue as a reviewer, but I look at it as slightly narrowing the audience that this paper might impact and slightly raising the bar for reproducibility. There are a zillion other things that make an audience narrow or broad, or make it more/less difficult to reproduce a result, and this goes in the pile somewhere, not even really as a strike against the paper, more as a bonus point they didn't get from me. The most common reason I see is licensing of commercial satellite imagery... but that's also an example of where the models/methods are still useful to other people, because if a paper says "we used WorldView-3 imagery but we can't release it", readers can still purchase WV-3 imagery and make use of the methods/models from the paper. I.e., this is really different than saying "we built a totally custom hyperspectral drone you'll never see and here are our results that you can never verify". I think we don't want to create a culture where no one can ever write papers about AI methods for commercial satellite imagery. Ditto for papers that use data that came from sources that don't allow public release; in the camera trap space, for example, requiring data release would in practice bias significantly against writing papers about data from indigenous land, or from whole countries (e.g. India) where data release is hard or impossible, or at least much harder than it is in the US.

If a paper doesn't release the data and doesn't really say anything about why, that's a more significant strike from me, because I agree with your inclination toward encouraging data release, but it's still not a dealbreaker, just an imperfection. If a paper is otherwise awesome, with awesome open-source code and/or public models and/or methods that are obviously a good idea, I might still give an above-the-rejection-line rating.

If the paper says that part of their contribution is that the data is publicly available, and I click the link and it's obviously not really available, now I get really cranky as a reviewer. As often as not it's just carelessness, and maybe they intend to really release the data, but IMO you can't have it both ways. Same for the case where data is "publicly available" but I have to log in to access it during the anonymous review process. Grrrr.

Ben Weinstein (benweinstein2010@gmail.com)

2024-01-03 23:30:57

I totally agree with these guidelines for once you accept the review. I'm just talking about trying to choose among the many review requests we are all getting. These are all good points. Usually you just get an abstract, so you can't even tell any of these things yet.

Drea Burbank (drea@savimbo.com)

2024-01-06 15:08:50

Hey guys, really bad at checking this channel but FYI our Indigenous-led biodiversity credit methodology is in open review now and will be the first certified biodiversity credit in the world ~ Feb. If you’re a biodiversity nerd like me please consider giving us a supportive review.

isbm.savimbo.com

Executive summary

Conservation-first, indicator-species-based biodiversity crediting

Original URL: http://isbm.savimbo.com

😍 Eric Greenlee, Ted Schmitt, Tiziana Gelmi Candusso

Aran Dasan (aran@sntech.co.uk)

2024-01-07 13:10:10

Sadly we’re not eligible for this one (marine monitoring is out of scope), but here’s a new funding call from InnovateUK:

https://apply-for-innovation-funding.service.gov.uk/competition/1838/overview/8e7ae74c-af9d-4c8b-8bee-084865a57276#summary

Defra and Innovate UK will invest up to £5 million in collaborative innovation projects that develop environmental monitoring solutions.
Projects must develop new or repurpose existing sensor systems and capabilities, for example observation systems, sensor or sampler carrying platforms or modelling systems and focus on one or more of the following challenge areas:
- biodiversity and natural capital
- soil health (including measuring soil carbon)
- water quality
- greenhouse gas (GHG) and ammonia emissions from Defra sectors

apply-for-innovation-funding.service.gov.uk

Competition overview - Environmental Monitoring Innovation - Innovation Funding Service

Original URL: https://apply-for-innovation-funding.service.gov.uk/competition/1838/overview/8e7ae74c-af9d-4c8b-8bee-084865a57276#summary

👀 Dan Watson

❤️ Dan Watson, Pen-Yuan Hsing

Dan Morris (agentmorris@gmail.com)

2024-01-07 20:17:12

New dataset on LILA, courtesy of New Zealand DOC:

https://lila.science/datasets/nz-trailcams

"This data set contains approximately 2.5 million camera trap images from various projects across New Zealand. These projects were run by various organizations and took place in a diverse range of habitats using a variety of trail camera brands/models. Most images have been labeled by project staff and then verified by volunteers. Labels are provided for 97 categories, primarily at the species level. For example, the most common labels are mouse (49% of images), possum (6.7%), and rat (5.5%). No empty images are provided, but some can be made available upon request. "

But also I'm going to remind everyone in a reply to this post why all the camera trap datasets on LILA are just a figment of your imagination.

:flag_nz: Mitchell Rogers, Alan Stenhouse, Isabella

🐧 Edward Bayes, Carly Batist, Chris Yeh

👍 Piotr Tynecki, Meredith Palmer, Joris Tinnemans, David Will, Luke Sheneman, Sam Lapp

😎 Jon Van Oast

Dan Morris (agentmorris@gmail.com)

2024-01-07 20:21:53

As promised, let me convince you that camera trap datasets don't exist on LILA.

That is, camera trap "datasets" still exist on LILA, but I encourage folks to think of them as arbitrary divisions within the one big dataset that is "LILA camera trap stuff". E.g., if you want images of coyotes, you probably don't care whether they came from the "Idaho Camera Traps" dataset or the "Caltech Camera Traps" dataset.

So for most ML training applications, rather than dealing with individual data sets, consider working with the Really Big CSV File that contains all the camera trap images on LILA, mapped into a common taxonomy:

https://lila.science/taxonomy-mapping-for-camera-trap-data-sets/

The Really Big CSV File has also been imported as a Hugging Face dataset:

https://huggingface.co/datasets/society-ethics/lila_camera_traps

🔥 Elizabeth Campolongo, Carly Batist, Meredith Palmer, Dan Stowell

Dan Morris (agentmorris@gmail.com)

2024-01-07 20:23:08

Also, if you're wondering "how many images did Dan look at to find the penguin doing the penguin-i-est thing a penguin could be doing for that thumbnail?", the answer is... maybe 5,000? Worth it.

😂 Mitchell Rogers, Jason Holmberg (Wild Me), Shir Bar, Alan Stenhouse, Yseult Hb, Alan Papalia, Elizabeth Campolongo, Carly Batist, Valentin Gabeff, Neha Hulkund

😆 Aakash Gupta, Rachael Laidlaw

🌊 Kalindi Fonda

Dan Morris (agentmorris@gmail.com)

2024-01-10 13:20:10

I also see that @Joris Tinnemans, who did all the hard work to collect and organize this dataset, has joined this Slack... welcome Joris, and thank you for your contribution to LILA!

Emily Lines (erl27@cam.ac.uk)

2024-01-08 10:16:15

A call to academic researchers: some of my colleagues in Computer Science, University of Cambridge have developed a Declaration on Academic Response to the Planetary Crisis: Declaration Text

If you would like to sign it, please add your name here:
https://docs.google.com/document/d/1TCrQIIDsJAI6JvOuDOvlzldkrEwhmtCoHOxcy1HlGsc/edit

charlotte (deshchang@gmail.com)

2024-01-08 10:26:49

Hi folks! If you’re using any form of natural language processing to analyze problems related to conservation, I wanted to share an opportunity to present your work at the North American Congress for Conservation Biology (NACCB, Vancouver, June 23-28 2024). @Amrita Gupta and I are co-organizing a symposium entitled “Analyzing text data for conservation problems and action”.

We have 1-2 more 15 minute presentation slots available.

If you’re interested and would like to learn more, please get in touch with me ASAP via DM (deadline: January 20).

I’d also be happy to share more information with you if you know folks that I should contact.

We’re especially interested in research on the following topics:

1) the barriers to using technology for conservationists, and

2) additional real-world case studies using NLP such as tracking discourse or attitudes toward species or other environmental entities, analyzing reports of wildlife harvesting, or corporate disclosures for environmental impact (though this latter list is certainly non-exhaustive).

👍 Carly Batist, Joseph Dimos, Suzanne Stathatos, Meredith Palmer, Jon Van Oast, Jason Holmberg (Wild Me), Dan Morris

👍:skin_tone_3: Alan Stenhouse

Abhay (abhaykash12@gmail.com)

2024-01-08 14:44:14

cc: @Peter Bull
(spoke to him recently about this space and he wanted to learn more given his recent relevant work)

💯 charlotte

❤️ charlotte

Dan Morris (agentmorris@gmail.com)

2024-01-16 19:23:10

Not relevant to your question, but I didn't realize until you mentioned it on this thread that NACCB is in Vancouver this year. I never go anywhere, and I consider Vancouver not going anywhere (from Seattle), so I'm in! Excited to meet folks at NACCB. Thanks @charlotte for pointing that out!

💯 charlotte

👋 Amrita Gupta

charlotte (deshchang@gmail.com)

2024-01-17 14:02:03

Thanks @Dan Morris and glad to hear! NACCB is a great venue for meeting folks in the policy space as well — the previous NACCB meeting featured a large “America the Beautiful” (aka 30x30) US eNGO meeting where scientists and policy leaders in these orgs were triangulating their strategies for landscape prioritization, communicating to policymakers, etc.

:flag_ca: Dan Morris

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-01-20 05:29:19

I'll be there as a speaker in another symposium, looking forward to see your symposium!!

🙏 charlotte

💯 charlotte

👋 Amrita Gupta

Ronny Hänsch (rww.haensch@gmail.com)

2024-01-08 17:00:46

Happy New Year everybody! Let's start with some good news: EarthVision will be again at CVPR this year!! Get your papers ready and see you in Seattle. https://www.grss-ieee.org/events/earthvision-2024/
(attached a happy memory from last year when @Sara Beery gave EarthVision a very nice shout-out at the last CVPR panel discussion after giving an awesome keynote at the workshop)

❤️ Caleb Robinson, Shir Bar, Devis Tuia, Nico Lang, Chris Lange, Robin Zbinden, Thor Veen, Akash Nagaraj, Alan Stenhouse, Andy Viet Huynh, Luke Sheneman

🎉 Jon Van Oast, Dan Morris, charlotte, David Russell, Oisin Mac Aodha, Georgia Atkinson, Mitchell Rogers, Valentin Gabeff, Yonghao Xu, Andy Viet Huynh

John Payne (drjohnpayne@gmail.com)

2024-01-14 01:28:17

I would be grateful for papers or advice from anyone who has experience in identifying tree species from aerial photography (airplane or drone), using ML. The goal is to find and identify extra-large trees in a closed-canopy tropical forest, for an ecology/conservation research project. The sample size of identified trees to use for training is limited and there is a very long distributional tail of rare species, so our expectation is that only a few common species will be identifiable to species.

Ben Weinstein (benweinstein2010@gmail.com)

2024-01-14 10:09:20

Happy to help. We have been working in this area for a few years. We have an RGB tree algorithm (https://deepforest.readthedocs.io/en/latest/), that has been fine-tuned for tropical forests https://ieeexplore.ieee.org/abstract/document/9387530 (but needs more annotations). 4 or 5 papers on tree species classification in hyperspectral data (https://scholar.google.com/citations?user=7POnELAAAAAJ&hl=en) and we are working on a global tree detection benchmark https://milliontrees.idtrees.org/. We would love to help and have your data included, every dataset gets the community closer to a usable baseline. RGB only classification is hard, but I can point to several interesting papers (a couple attached).

ieeexplore.ieee.org

Multisensor Data Fusion for Improved Segmentation of Individual Tree Crowns in Dense Tropical Forests

Original URL: https://ieeexplore.ieee.org/abstract/document/9387530

MillionTrees

A Benchmark Dataset for Airborne Machine Learning

The MillionTreesBenchmark

Original URL: https://milliontrees.idtrees.org/

🙌:skin_tone_3: Alan Stenhouse

👍 Martin Marzidovsek

John Payne (drjohnpayne@gmail.com)

2024-01-14 14:13:55

Wow, thank you Ben for that rich vein of information. I look forward to reading those references and I’ll respond with questions when I have done so.

Dan Stowell (dan.stowell@naturalis.nl)

2024-01-15 04:44:43

Hi all - If you were looking for audio datasets of animal sound, where would you look? (I know lots of new datasets recently, but I don't think they're all listed in one place. I'd like to fix that... but there's no point listing them where people don't look...)

🙌 Sara Beery

👀 Ben Williams

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-01-15 09:18:46

https://lila.science/otherdatasets#bioacoustics

LILA BC

https://bioacousticsdatasets.weebly.com/index.html

LILA BC - List of other conservation-related data sets

Written by

lilawp

Est. reading time

32 minutes

Original URL: https://lila.science/otherdatasets#bioacoustics

👍 Dan Stowell, John Martinsson, Taiki Sakai - NOAA Affiliate, Sam Lapp, Sara Beery

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-01-15 09:19:10

bioacousticsdatasets.weebly.com

BIOACOUSTICS DATASETS

A list of bioacoustics datasets. All datasets listed below are easily accessible and can be downloaded in one go. See below for a list of Bioacoustics Online Repositories that offer a more...

Original URL: https://bioacousticsdatasets.weebly.com/index.html

👍 Dan Stowell, John Martinsson, Taiki Sakai - NOAA Affiliate, Sara Beery

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-01-15 09:21:50

do you mean like labeled datasets? raw soundscape recordings? completely open-source you-can-download-all-files? all of the above?

Dan Stowell (dan.stowell@naturalis.nl)

2024-01-15 09:24:20

Good questions! I guess I mean anything that has already been somewhat curated/selected, and open data. My secret motivation is... I have new PhD students starting this year, and I'd like it to be clear for people like them to know where to find "all the good new bioacoustics datasets we could use in our ML"

John Martinsson (john.martinsson@ri.se)

2024-01-15 10:33:22

I'd be very interested in a curated list of audio datasets of animal sounds as well!

Where would I look? I would personally do a google site:github.com search and look for a GitHub repo which maintains a curated list of animal sounds (did not find any). However, there seems to be a curated list of audio technologies (https://github.com/DolbyIO/awesome-audio, including some audio datasets at the end). When this fails me I would probably just do a google search and see what I find. I may also look at: https://dcase-repo.github.io/dcase_datalist/ which would lead me to the list by Justin that Carly linked (https://bioacousticsdatasets.weebly.com/). I'd completely miss the lila.science list with this approach. Thank you Carly for that link!

Personal thought: There are a lot of lists on GitHub with the "awesome-" where the purpose is to curate a list for a certain topic. Not sure how popular these lists are, but I like the community aspect. GitHub makes it easy for others to contribute with a pull request, and then the community can review and accept the addition and keep the list alive. Seems to be one for bioacoustics by Yann Bayle (https://github.com/ybayle/awesome-bioacoustic), but mainly focusing on references.

Maybe this is an opportunity to create a "awesome-animal-sound-datasets" repository?

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-01-15 11:16:11

There is also another list that would be of interest, which is that of open-source models. For example, the Kitzes lab (@Justin Kitzes @Tessa Rhinehart @Sam Lapp) is working on a bioacoustics model zoo - https://github.com/kitzeslab/bioacoustics-model-zoo/tree/main

Elly Knight (ecknight@ualberta.ca)

2024-01-15 11:32:27

https://wildtrax.ca/ has almost ~750K acoustic recordings where the processed data (the first detection of all species identified by human expert human listeners) is public. You can download flac clips of those detections for model training

Wildtrax

Copernicus Sentinel-2 pieces together mosaic of Antarctica - Sentinel Online

WildTrax Manage, store, process, and share environmental sensor data.

Original URL: https://wildtrax.ca/

Benjamin Hoffman (benjaminsshoffman@gmail.com)

2024-01-15 14:59:46

there are 10 collected here: https://github.com/earthspecies/beans

earthspecies/beans

BEANS: The Benchmark of Animal Sounds

Stars

Language

Python

Lauren Harrell (laurenaharrell@gmail.com)

2024-01-28 21:44:31

Since it hasn’t been mentioned: https://xeno-canto.org is an awesome call repository for birds, and has some insects and bats as well

👆 Carly Batist

Sam Lapp (sam.lapp@pitt.edu)

2024-01-29 10:29:10

here’s one list more by Tessa, with contents from Justin’s and Dan’s https://docs.google.com/spreadsheets/d/1KrmCB0vvSK7V3znJfycO-eOMZJKP2F-Ih6neRYPz1Xc/edit#gid=0

Jan Huus (jhuus1@gmail.com)

2024-02-08 13:32:37

Another good one is https://www.inaturalist.org/. I used it to get recordings of squirrels and amphibians for my bird sound recognizer: https://github.com/jhuus/HawkEars. It has a nice API too: https://github.com/pyinat/pyinaturalist

inaturalist.org

iNaturalist

iNaturalist is a social network for naturalists! Record your observations of plants and animals, share them with friends and researchers, and learn about the natural world.

Original URL: https://www.inaturalist.org/

jhuus/HawkEars

Language

Python

Last updated

7 months ago

pyinat/pyinaturalist

Website

<https://pyinaturalist.readthedocs.io>

Stars

115

Devis Tuia (devis.tuia@epfl.ch)

2024-01-16 04:09:21

@Ben Koger is great! apply with him!

❤️ Ben Koger, Sara Beery

George Darrah (george.darrah@systemiq.earth)

2024-01-16 06:02:49

Hi everyone - our friends at Ground Effect are hiring a Natural Capital Portfolio and Investment Director, please do shoot me a dm if interesting/you know anyone who'd be up for this! I think it's probably the most exciting nature investing remit out there... across companies, academia and NGOs. https://www.groundeffect.io/

Burooj Ghani (buroojghani@gmail.com)

2024-01-16 08:51:03

Can you design a system that learns from five example vocalizations in a lengthy animal sound recording to identify similar sound events throughout the audio?

We will continue to host the Few-shot Bioacoustic Event Detection challenge at DCASE 2024. Stay tuned and spread the word!

DCASE challenge 2024 short task descriptions are out and the challenge will be live on April 1.
https://dcase.community/challenge2024/

👍 Georgia Atkinson, Edward Bayes, Sara Beery, Carly Batist, Douglas Mbura, Benjamin Hoffman, Enis Berk Çoban, Shubhr singh, Maddie Cusimano

Ciera Martinez (ccmartinez@berkeley.edu)

2024-01-17 13:08:25

📣 Come work with me! We have a number of Postdoc positions available and 1-2 of them will focus on leveraging AI for biodiversity monitoring.

DSE, along with an opportunity from our partners in the James M. and Cathleen D. Stone Center for Large Landscape Conservation, is accepting applications from recent PhDs in environmental science and/or data science fields interested in providing domain specific research that informs the development of data enabled solutions to our most pressing environmental challenges.

Competitive pay in a unique academic environment that values open science, inclusiveness, and impact driven research. Please help send to your networks and reach out to me or @Carl Boettiger if you have any questions.

Learn more here: https://dse.berkeley.edu/postdocs

💚 Carl Boettiger, Sara Beery, Meredith Palmer

Patrick Beukema (patrickb@allenai.org)

2024-01-17 16:48:17

Question for the group especially anyone who might know folks at ESA. We are working on a new dataset that depends on WorldCover ESA — WorldCover ESA did not include Antartica, but i came across this press release from Copernicus about a cloud free mosaic (3000 Sentinel-2 images) covering Antarctica: https://sentinels.copernicus.eu/web/sentinel/-/copernicus-sentinel-2-pieces-together-mosaic-of-antarctica That page says “the data are free and open to all scientists and researchers” but I can’t find any link or any other reference to that dataset. Does anyone know about that data or works at ESA who might be able to help?

Sentinel Online

Original URL: https://sentinels.copernicus.eu/web/sentinel/-/copernicus-sentinel-2-pieces-together-mosaic-of-antarctica

Patrick Beukema (patrickb@allenai.org)

2024-01-17 17:07:12

It might be that one is simply supposed to use the Copernicus browser — but the press release made it sound like there would be some specific data product https://dataspace.copernicus.eu/browser/?zoom=7&lat=-77.75217&lng=-175.61199&themeId=DEF[…]urce3D=%22MAPZEN%22&cloudCoverage=9&dateMode=TIME%20RANGE

dataspace.copernicus.eu

Copernicus Browser

Search, visualize and download Sentinel satellite imagery

Original URL: https://dataspace.copernicus.eu/browser/?zoom=7&lat=-77.75217&lng=-175.61199&themeId=DEFAULT-THEME&visualizationUrl=https%3A%2F%2Fsh.dataspace.copernicus.eu%2Fogc%2Fwms%2Fa91f72b5-f393-4320-bc0f-990129bd9e63&datasetId=S2_L2A_CDAS&fromTime=2023-01-01T00%3A00%3A00.000Z&toTime=2023-02-28T23%3A59%3A59.999Z&layerId=1_TRUE_COLOR&mosaickingOrder=mostRecent&demSource3D=%22MAPZEN%22&cloudCoverage=9&dateMode=TIME%20RANGE

Patrick Beukema (patrickb@allenai.org)

2024-02-05 19:33:16

i know this wasn’t wildly popular but in case anyone else ends up looking for it, support at Copernicus kindly led me to this: https://s2gm.land.copernicus.eu/mosaic-hub

Devis Tuia (devis.tuia@epfl.ch)

2024-01-18 10:27:02

Hello everyone!
I wanted to share with everyone the video of our expedition in Djibouti :flag_dj:, where with a team of data scientists and marine ecologists from Switzerland, Sudan and Djibouti we went studying the status of coral reefs 🪸 at the entrance of the Red Sea. It was an inspirational experience and we are now analysing the data, more news will follow! For now enjoy the beauty of Djibouti! PS: if you are on a schedule, we talk about AI starting minute 6:40 with @Jonathan Sauder 😉

https://vimeo.com/903195229?share=copy

Vimeo

TRSC (https://vimeo.com/trsc)

The Seven Brothers Mission - Djibouti, November 2023

Original URL: https://vimeo.com/903195229?share=copy

🎉 Ronny Hänsch, Sonny Burniston, Dan Morris, Justin Kay, Robin Zbinden, Elizabeth Campolongo, Shir Bar, Chris Llorca, Oisin Mac Aodha, Ted Schmitt, Steve Haddock, Elie Alhajjar, Sara Beery, Aakash Gupta

😍 Carly Batist, Nicolas Arrieta Larraza, Sonny Burniston, Justin Kay, Robin Zbinden, Shir Bar, Chris Llorca, Leonardo Viotti, Gustavo Perez, Mitchell Rogers, Elie Alhajjar, Eric Colson, Subhransu Maji, Aakash Gupta

🙌 Aran Dasan, Elie Alhajjar

❤️ Patrick Beukema, Elie Alhajjar, Nora Gourmelon, Valentin Gabeff, Hannah Kerner, Rebecca Wilks

💪 Giacomo May

Hugo Magaldi (magaldi.hugo@gmail.com)

2024-01-19 10:22:55

👋 Hello everyone

Thrilled to join the community! I'm a French engineer with a PhD in mathematics and 3 years of industry experience in ML/DL, eager to enter the field of conservation. I'm looking to join a challenging (public or private) project where I can be of use, and ideally get some field exposure.

Happy to talk, don't hesitate to reach out!

👋 Sara Beery, Dan Morris, Jacob Adkins, Nicolas Arrieta Larraza, Jon Van Oast, Robin Zbinden, Andrew Schulz, Shir Bar, Elizabeth Campolongo, Mitchell Rogers, Aakash Gupta, Phuc Le, Jason Holmberg (Wild Me), Catarina Silva, Andy Viet Huynh, John Martinsson, Piotr Tynecki, Antón Álvarez, Kishore Panaganti, Martin Marzidovsek

slackbot

2024-01-23 09:33:46

This message was deleted.

Ben Williams (ben.williams.20@ucl.ac.uk)

2024-02-26 05:38:21

Thanks Aakash this is awesome! I won't be trying it but my colleague will so I'll pass it on., this could definitely be useful. Perhaps we're onto a new grain counting benchmark!

Kari Kuester (kari.kuester@sunbird.tv)

2024-01-24 04:08:08

Maybe something more old-school with erosion+dilation+edge detection+thresholding+contouring could work here? https://www.youtube.com/watch?v=DsePM4F3tEw

https://shrishailsgajbhar.github.io/post/OpenCV-Apple-detection-counting

YouTube

Anushka Agarwal (https://www.youtube.com/@anushkaagarwal8250)

Part3: Counting the number of objects in an image using Edge detection, Thresholding, Contouring etc

Original URL: https://www.youtube.com/watch?v=DsePM4F3tEw

shrishailsgajbhar.github.io

[OpenCV] Detecting and Counting Apples in Real World Images using OpenCV and Python - Tech(B)log of Dr. Shrishail Gajbhar

Tech(B)log of Dr. Shrishail Gajbhar

Original URL: https://shrishailsgajbhar.github.io/post/OpenCV-Apple-detection-counting

Cameron Trotter (cater@bas.ac.uk)

2024-01-24 05:06:27

Hi @Ben Williams, perhaps SegmentEveryGrain could work here? It's an offshoot of SAM that claims to focus grain-like objects. If I remember right a colleague of mine tried using it for ice floe segmentation, though I can't remember how well it worked

zsylvester/segmenteverygrain

A SAM-based model for instance segmentation of images of grains

Stars

290

Language

Jupyter Notebook

❤️ Ben Williams

Ben Williams (ben.williams.20@ucl.ac.uk)

2024-01-24 05:11:07

These are super useful, thank you! SegmentEveryGrain sounds ideal, will pass it all on

👍 Cameron Trotter

Ben Weinstein (benweinstein2010@gmail.com)

2024-01-24 12:23:44

def report back on SegmentEveryGrain, we've been thinking about retraining SAM for tree crowns and I don't know how worthwhile fine-tuning is, versus just using vanilla.

Jonah Fox (jonahfox@gmail.com)

2024-01-26 08:13:18

Hi everyone - I'm part of a startup trying to improve biodiversity with AI. I'm really interested (but new to) using drones with multispectral sensors mixed with AI. Is there anyone here who does this ?

Casey Clifton (caseyclifton@proton.me)

2024-01-26 08:14:42

Also new to this but very interested in and trying to get involved! Got any key papers or particular tech you're excited about?

Jonah Fox (jonahfox@gmail.com)

2024-01-26 08:15:37

just reading this meta paper to start with !

Jonah Fox (jonahfox@gmail.com)

2024-01-26 08:15:38

https://www.mdpi.com/2504-446X/3/1/9

MDPI

Identifying Species and Monitoring Understorey from UAS-Derived Data: A Literature Review and Future Directions

Understorey vegetation plays an important role in many ecosystems, yet identifying and monitoring understorey vegetation through remote sensing has proved a challenge for researchers and land managers because understorey plants tend to be small, spatially and spectrally similar, and are often blocked by the overstorey. The emergence of Unmanned Aerial Systems (UAS) is revolutionising how vegetation is measured, and may allow us to measure understorey species where traditional remote sensing previously could not. The goal of this paper was to review current literature and assess the current capability of UAS to identify and monitor understorey vegetation. From the literature, we focused on the technical attributes that limit the ability to monitor understorey vegetation—specifically (1) spatial resolution, (2) spectral sensitivity, (3) spatial extent, and (4) temporal frequency at which a sensor acquires data. We found that UAS have provided improved levels of spatial resolution, with authors reporting successful classifications of understorey vegetation at resolutions of between 3 mm and 200 mm. Species discrimination can be achieved by targeting flights to correspond with phenological events to allow the detection of species-specific differences. We provide recommendations as to how UAS attributes can be tailored to help identify and monitor understorey species.

Original URL: https://www.mdpi.com/2504-446X/3/1/9

Ben Weinstein (benweinstein2010@gmail.com)

2024-01-26 12:17:13

I work a bunch in this area. Happy to connect you with labs, are you a developer, have a target use case? Kind of organism you are interested in?

✔️ Jon Van Oast

Jonah Fox (jonahfox@gmail.com)

2024-01-26 13:49:20

fab ! - yes im a developer - currently interested in habitat assessment using drones / GIS .

Casey Clifton (caseyclifton@proton.me)

2024-01-29 19:19:55

hey @Jonah Fox & @Ben Weinstein - im also a developer and have actually recently launched a business to work on AI for biodiversity with a small team of AI devs
would love to stay in touch & chat about any specific projects / collabs! atm we're working on wildlife camera trap stuff and GIS for analysing forests to predict carbon abatement

Aakash Gupta (aakash@thinkevolveconsulting.com)

2024-01-29 22:49:33

I am interested.

Especially if there is a way to identify metallic traps that are placed by poachers in dense forests. Sometimes these are electrified, leading to significant stress to the trapped animals.

David Russell (davidrussell327@gmail.com)

2024-01-30 09:07:41

Hey all, we're just getting up and running but you may be interested in our work at Open Forest Observatory. We're developing easy-to-use workflows for processing drone imagery for forest ecology. My main project is the multiview mapping toolkit which allows you to do deep learning model training and prediction directly on individual drone images instead of stitched orthomosaics.

openforestobservatory.org

Open Forest Observatory

Drone-based forest maps and mapping tools for ecologists and forest managers

Original URL: https://openforestobservatory.org/

open-forest-observatory/multiview-mapping-toolkit

Stars

Language

Python

👍 Martin Marzidovsek

Jonah Fox (jonahfox@gmail.com)

2024-02-05 05:10:57

@David Russell @Casey Clifton - would love to have a chat some time ! Are you around this week ?

🙌 Jonah Fox

David Russell (davidrussell327@gmail.com)

2024-02-05 09:08:48

@Jonah Fox that would be great. Would Thursday or Friday morning (say 9-12) EST work for you?

Jonah Fox (jonahfox@gmail.com)

2024-02-05 09:13:48

let's go for Thursday at 9 EST if thats ok

✅ David Russell

Casey Clifton (caseyclifton@proton.me)

2024-02-05 19:17:19

I'll have to pass this time but will check out the above links!

Panayiotis Danassis (pdanassis@g.harvard.edu)

2024-01-26 09:54:24

Hello everyone,

We invite you to submit any work related to social impact to the Autonomous Agents for Social Good (AASG) workshop at AAMAS.

DEADLINE: Feb 29, 2024

For more details, please see: https://panosd.eu/aasg2024/

panosd.eu

Autonomous Agents for Social Good 2024

Original URL: https://panosd.eu/aasg2024/

:flag_nz: Mitchell Rogers, Jason Holmberg (Wild Me), Sara Beery, Sepand Dyanatkar

Aakash Gupta (aakash@thinkevolveconsulting.com)

2024-01-29 23:25:01

I just realized that older messages are lost in the channel. Whenever I find something useful, I save it for later. But I think because this channel is on the free version of slack the older messages get deleted!

Changed my workflow to copy the links and entire messages to google docs. !

Piotr Tynecki (piotr@tynecki.pl)

2024-01-30 00:08:44

In 2023 and before it was possible to get Slack for free workspace but with Paid plans features (like unlimited history). There are two options for get that: Slack for nonprofits and Slack for education but I see they offer 85% of discount this year.

👍 Aakash Gupta, Jon Van Oast

Elizabeth Campolongo (e.campolongo479@gmail.com)

2024-01-30 09:45:37

Same! Saving for later used to be a work-around.

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-01-30 10:23:05

until last year, slack would keep a certain number of messages (megabytes) in the free version and delete the oldest when the space limit was reached. it would not delete pinned or saved messages.
They changed that. Everything older than 90 days is now hidden, including saved and pinned messages. it sometimes shows it blurred out to tease people to buy the paid plan, internally the history is kept, but inaccessible unless you pay.

😞 Elizabeth Campolongo

Jon Van Oast (jon@wildme.org)

2024-01-30 10:42:11

i wonder if the 90-day-limit applies to the self-DM? i sometimes put things in there.

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-01-30 10:43:37

self DMs are apparently not affected when they are text, but the limit does apply to files and attachements - those get a placeholder that they are hidden because they are older than 90 days

✔️ Jon Van Oast, Elizabeth Campolongo

Sara Beery (sara.beery@gmail.com)

2024-01-30 12:45:08

I've looked into this several times, but don't have any easy/good solutions. If someone has the capacity to come up with a plan for how to switch this to something that maintains history and how to pay for/maintain funding support for it I would be very welcome to someone leading that charge! Issues I've run into in the past are that with large communities like this the cost is still high even for nonprofits/education. And there are some limitations to both that end up being restrictive (ie we're not all at one institution)

💕 Jon Van Oast

👆 Steve Haddock

Piotr Tynecki (piotr@tynecki.pl)

2024-01-30 12:48:37

@Sara Beery should we consider reaching out to organizations like WILDLABS for support? They could potentially advocate for the community and help us secure Slack for nonprofits, with us covering the remaining 15% of the monthly budget.

Sara Beery (sara.beery@gmail.com)

2024-01-30 12:51:10

They already have their own internal slack that is using their nonprofit ID, and I think you only get one per nonprofit. We could also start a nonprofit, but that seems complicated as well 🙂

😅 Jon Van Oast

Piotr Tynecki (piotr@tynecki.pl)

2024-01-30 12:51:39

Hmmm got it.

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-01-30 13:17:06

another option is to go back to non-proprietary chat options (XMPP based, Matrix based, ...) The budget needed to host a free open source chat/messaging server - beyond initial setup - should be a fraction of what Slack demands these days, but the learning curve can be higher and people need yet another app, which makes it harder to connect

❤️ Jon Van Oast

Jon Van Oast (jon@wildme.org)

2024-01-30 14:10:03

i am a fan (and user) of matrix/element - basically use it as a slack/discord substitute for a couple different groups i am part of. but it seems to be a big ask for folks to adopt and learn yet another app. 😕 i wonder if we could find an ai/conservation ngo which does not already use slack (and never wants to!) who might want to sponsor us. 🤔

Chris Yeh (chrisyeh96@gmail.com)

2024-01-30 22:02:31

Discord and MS teams are free and maintain history indefinitely, but those require shifting platforms

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-01-31 01:59:48

@Chris Yeh these don't solve the problem though - of a single 3rd party provider having control over both server/protocol, client/app and the terms of service. We'd be back to the situation with slack before they changed the rules, hoping they wouldn't change the rules (charge more money/limit features) or like google chat - stopping the service altogether.

Anastasia Pagán (anastasia@wildme.org)

2024-01-31 13:23:23

Let's go back to IRC. Slack, Discord, Teams, etc. are all just modern attempts at reinventing it anyway. 😄

🥲 Jon Van Oast, Eric Price, Aakash Gupta, Kakani Katija

Piotr Tynecki (piotr@tynecki.pl)

2024-01-31 13:24:30

yeah, IRC on freenode 🔥, I spent more than 10 yrs with the black screen

💚 Jon Van Oast

Steve Haddock (haddock@mbari.org)

2024-02-02 12:39:02

Every non-profit / academic Slack I know has been struggling with this. Some are limping along with the truncated history, while others have switched to Discord.
Really sad Slack changed their free model, and it is completely unrealistic to get the paid plan for a large but informal academic group like this.

👍 Jon Van Oast

Anastasia Pagán (anastasia@wildme.org)

2024-02-02 12:40:52

Discord has its advantages. Because its primary audience is the gaming community, they're likely to lose a large part of their audience by making it paid to use.

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-02-02 12:42:31

yeah but the gaming community would let them get away with other things the professional field wouldn't - I could totally see them do stuff like adding advertising revenue based on what topic is being discussed in chat

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-02-02 12:43:15

unless it causes lag in games, that'd be their death sentence

😂 Anastasia Pagán, Jon Van Oast

Shiva Muruganandham (shivamurug@gmail.com)

2024-02-03 23:30:08

I've been part of a few (pseudo-) academic Slacks that shifted away to Discord for this reason - if there's enough buy-in from the community on here, shifting platforms might work for the better.

Aakash Gupta (aakash@thinkevolveconsulting.com)

2024-01-30 06:10:40

@Ben Williams I am curious if you were able to count the number of grains. I got a grand total count of 39,690 grains.

The workflow is provided in this github repo:
https://github.com/Think-Evolve-Consulting/Counting-grains
and a short blog:
https://www.thinkevolveconsulting.com/2050-2-counting-grains/

slackbot

2024-02-05 02:37:52

This message was deleted.

Serge Wich (sergewich@gmail.com)

2024-03-01 15:50:18

Hi, happy to have a chat. I am involved in https://www.conservationai.co.uk/
Best, Serge

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-02-05 08:44:59

Hi! We at Rainforest Connection and Arbimon have lots of projects in SE Asia using sound and AI for biodiversity monitoring! Feel free to email me if you’d like to talk more 🙂 carly@rfcx.org

Mohit Dubey (mohit.dubey96@gmail.com)

2024-02-05 13:54:28

Does anyone know of a website for metadata of global plant species? Something like https://www.calflora.org/ but worldwide

Brandon Hays (brandon.hays@duke.edu)

2024-02-05 20:31:10

Not sure if this is what you mean by metadata, but the Global Biodiversity Information Facility has both occurrence data and species fact sheets for a lot lot of species
https://www.gbif.org/

gbif.org

GBIF

Global Biodiversity Information Facility. Free and Open Access to Biodiversity Data.

Original URL: https://www.gbif.org/

Anna Willoughby (arwill19@gmail.com)

2024-02-07 20:22:02

I help maintain a biodiversity dataset repo. You can check it out for plant databases, and feel free to ping me for additions https://earthskysea.org/biodiversity-databases/

Alan Stenhouse (alan.stenhouse@csiro.au)

2024-02-08 00:04:42

Perhaps https://www.worldfloraonline.org/ or https://powo.science.kew.org/

Lauren Harrell (laurenaharrell@gmail.com)

2024-02-05 14:49:08

To those of you in Colorado: we're organizing a local meetup of people/organizations involved in conservation technology in the Boulder area on Tuesday, Feb 20 at 2 PM- 5 PM at the Boulder Library. Details are here and RSVP here. If you can't make it but are interested in future events, please indicate in the RSVP form.

wildlabs.net

WILDLABS Colorado Conservation Tech Meetup | WILDLABS

Join us on February 20th in Boulder, Colorado for a WILDLABS meetup convening local groups working in conservation technology to connect, exchange ideas, and brainstorm future in-person events in the area.

Original URL: https://wildlabs.net/event/wildlabs-colorado-conservation-tech-meetup

🙌 Talia Speaker, Jason Holmberg (Wild Me)

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-02-05 15:45:39

so jealous!

Lauren Harrell (laurenaharrell@gmail.com)

2024-02-05 15:46:21

You're always welcome to come visit Carly!

💯 Talia Speaker

Brandon Hays (brandon.hays@duke.edu)

2024-02-05 20:49:39

Sawadee krap Lyn! I'm doing my PhD work in Thailand on elephant impacts on forests. I'm going to be doing some drone work and camera trapping, hopefully starting this summer. I'd love to connect and collaborate on some outreach material down the road! Shoot me an email at brandon.hays@duke.edu

Nanticha Ocharoenchai (Lyn) (lynnanticha.o@gmail.com)

2024-02-05 21:27:47

Thank you for your replies @Carly Batist and @Brandon Hays! I will drop you an email now. Excited to hear more 🙂

stefano puliti (stefano.puliti@nibio.no)

2024-02-13 08:59:07

Hi ,

does anyone has good tips on image sources for forests beyond the GBIF data (inclusive of iNaturalist)? I am generally interested in the tree component rather than the understory 🙂

thanks in advance!

Ben Weinstein (benweinstein2010@gmail.com)

2024-02-13 17:33:55

Do you mean photos from cell phones of individual trees? Or photos of forests from above. What’s the use case, we have collected a lot.

Autumn Nguyen (ngoc54n@mtholyoke.edu)

2024-02-13 09:26:48

Hi! Does anyone know of any climate or environmental projects that I can contribute to by processing their data and training machine learning models on them? I am going to do a data science and ML project for my course this semester, and though it’d be much easier to work with some popular Kaggle dataset, I want to spend my time working on some real-world ongoing research/development project. I’d appreciate any pointers 🌱!
- Autumn

Piotr Tynecki (piotr@tynecki.pl)

2024-02-13 09:33:19

Hey @Autumn Nguyen,
If you're interested in working with camera trap data (images/videos) for an NGO based in Europe and implementing state-of-the-art AI/CV algorithms to address specific issues, please feel free to reach out to me. I'd be happy to collaborate.

slackbot

2024-02-15 09:00:53

This message was deleted.

Caleb Robinson (calebrob6@gmail.com)

2024-04-01 11:05:03

NYT's "Opinion Today" title is "Birding can change your life" -- https://www.nytimes.com/2024/03/30/opinion/birding-spring-merlin-ebird.html. Made me think about this thread :)

😁 Sara Beery

slackbot

2024-02-15 11:15:16

This message was deleted.

Shir Bar (shirbar@tauex.tau.ac.il)

2024-03-29 02:17:06

Hey all,
The deadline for applications to the CV4E 2025 winter workshop has been extended till Monday April 1st!
Please consider applying, everyone knows the best thing to do in winter is to sit somewhere sunny and write awesome AI for Conservation code 🙂
See more info in the thread above.

🙌 Carly Batist, Sara Beery, Suzanne Stathatos, Justin Kay, Antón Álvarez, Ishan Nangia

🐋 Taiki Sakai - NOAA Affiliate

😎 Jon Van Oast

🙌:skin_tone_3: Alan Stenhouse

Kalindi Fonda (kalindi.fonda@gmail.com)

2024-02-18 09:11:20

Looking for courses/masters in Ai for conservation.

If longer, preferably remote, but open to all suggestions.

I am doing a computer science masters online, remote and part time (I always like to have some kinds of course to keep learning). I do one course every few months.
As I am progressing I figured it would probably be more rewarding if the topics and courses were more specific and more relevant to my interests. So in line with this slack group.

Thank you 💪 🪸 ❤️‍🔥

👍 Aakash Gupta

Aakash Gupta (aakash@thinkevolveconsulting.com)

2024-02-18 11:06:22

Recently released Sora (by OpenAI) could be a gamechanger for those long-tailed events. Consider showing the model a single image of a rare species and asking it to model the behavior. The data generated could then be piped for training more models.

👀 Antón Álvarez, mimi, Jonathan Roberts

🤯 Rita Pucci

❤️ Ben Williams

Cara Appel (appelc@oregonstate.edu)

2024-02-18 17:05:40

Union Bay Natural Area by UW campus is excellent and is one of the most heavily birded areas in the region. I'd love to join!

😍 Sara Beery

💯 Jes Lefcourt, Angela Zhu

slackbot

2024-02-20 07:25:16

This message was deleted.

Alan Stenhouse (alan.stenhouse@csiro.au)

2024-04-30 20:22:06

BioCLIP might be of interest: https://imageomics.github.io/bioclip/

slackbot

2024-02-20 17:34:40

This message was deleted.

slackbot

2024-02-21 16:08:23

This message was deleted.

Michael Yair (m1cha3l.ya1r@gmail.com)

2024-02-23 07:36:57

Yeah... It seams like a cyber security action our IT took, they are informed and probably restore abroad communication by the end of the weekend.
Here you may find a youtube playlist with demos.
A Colab notebook of the AI training.
A demo data set from our database.
And the Gitlab group.

Those links can be found on our website too, once it will be available to others outside the university.

It is still under development, and for the safety of the data we and our colleagues gather,
to use the tool you need to approach the admin to get username and password,

Have a great weekend!

YouTube

STARdbi

STARdbi demonstrations and development evolution

Original URL: https://www.youtube.com/playlist?list=PLh47sVEa670tK6Os45oAGXSlTcAouRZSc

GitLab

STARdbi · GitLab

<a href="http://GitLab.com">GitLab.com</a>

Original URL: https://gitlab.com/stardbi

Michael Yair (m1cha3l.ya1r@gmail.com)

2024-02-23 07:43:31

The admin: Prof Chen Keasar - keasar@bgu.ac.il

Michael Yair (m1cha3l.ya1r@gmail.com)

2024-03-02 14:31:09

Just to inform all, the cyber attack we were under for a week or so, is behind us.
The site is back and active,

Anna Willoughby (arwill19@gmail.com)

2024-03-04 16:29:23

thats great news! ill check it out

Michael Bunsen (notbot@gmail.com)

2024-03-07 12:00:44

Awesome, thanks for the update. I look forward to trying it out!

Anna Willoughby (arwill19@gmail.com)

2024-02-22 11:54:16

Hi, none of the webpages loaded... I was using chrome? I'm interested in this, as have hundreds of sticky traps with insects from the southwest with no feasible plan to id them.

Jenna Kline (jennamkline@gmail.com)

2024-02-22 15:12:40

hi @Rowan Converse! thanks for sharing this dataset! are you able to share the drone telemetry information associated with the data?

Nicolas (nicolas.lecomte@umoncton.ca)

2024-02-22 15:59:50

The links are also not working on safari and chrome on mac…

Cynthia Wu (cynthiaswu@gmail.com)

2024-02-26 19:43:07

Hello! Does anyone know what types of tree disease can be detected by NDVI? Are there certain types of diseases that are less detectable?

I am looking at a case where root fungus was found inside many eucalyptus trees that needed to be removed. However, looking at the trees, the NDVI from the previous year seemed quite good. Is it because the fungus develops quickly, or because the fungus does not affect the leaves, and therefore NDVI cannot catch it?

It was caught using sonic tomography, I’m wondering if near infrared images of the trunk would have caught it too?

Akshit Gupta (akshitgupta1695@gmail.com)

2024-04-02 17:06:37

Hi Cynthia, I saw this message now. Were you able to resolve this already? I am also curious about this and would love to know your findings

Evan Eskew (eveskew@gmail.com)

2024-02-26 20:24:46

Hi Cynthia, interesting question! It's not my area of expertise, but maybe this will help point you in the right direction.

I'm guessing the problem is partly that these fungal diseases can spread rapidly (meaning even relatively new data is too old to be useful) and/or the impacts within a given plant are spotty:
https://www.fs.usda.gov/nrs/pubs/jrnl/2022/nrs_2022_sapes_001.pdf

We treated each pixel as an observation because oak wilt disease does not manifest uniformly across the canopy of a tree, especially during early stages of infection. At early stages, the fungus may have infected only a fraction of the vessels within the tree trunk. Thus, curtailing the water supply to a few branches that become symptomatic while others remain asymptomatic. Treating pixels -rather than the whole tree- as observations is critical to prevent false negatives that result from early infected trees displaying a small number of symptomatic pixels.

It seems others do use different measures to try to get at what NDVI can miss: https://ageagle.com/use-cases/using-the-micasense-chlorophyll-map-to-identify-fungus-missed-by-ndvi/

🌳 Cynthia Wu

❤️ Cynthia Wu

Cynthia Wu (cynthiaswu@gmail.com)

2024-02-27 03:56:17

Thank you so much Evan for such a thoughtful and helpful response! We’re going to try some of these and see how it goes. Really appreciate it!!

👍 Evan Eskew

George Darrah (george.darrah@systemiq.earth)

2024-02-27 05:15:36

Our friends at Silverstrand are launching applications for their Biodiversity Accelerator... if you're building a biodiversity focused venture at pre-seed/seed stage, this is for you! Can vouch for how awesome these folk are... https://www.linkedin.com/posts/silverstrand-capital_impact-innovation-capacitybuildin[…]765654171648-ZArZ?utm_source=share&utm_medium=member_desktop

linkedin.com

Silverstrand Capital on LinkedIn: #impact #innovation #capacitybuilding

📢 Calling all nature tech founders working to address the biodiversity crisis… Applications for the Biodiversity Accelerator+ are now open! Don’t miss your…

Original URL: https://www.linkedin.com/posts/silverstrand-capital_impact-innovation-capacitybuilding-activity-7165515765654171648-ZArZ?utm_source=share&utm_medium=member_desktop

❤️ Cynthia Wu

👍 Martin Marzidovsek

Xiaojuan Liu (xjliu@climatechange.ai)

2024-02-28 14:48:24

Hi AI for Conservation community! My name is Xiaojuan, and I’m a research scientist at Climate Change AI (CCAI). I’m reaching out to see if any of you would be open to participating in either a 60-minute group virtual workshop session or a one-on-one interview focused on critical data gaps in using machine learning for work in biodiversity/ecosystems. CCAI, partnering with Google DeepMind, is conducting a global stocktake of critical data gaps that inhibit effective machine learning (ML) solutions to climate-related challenges. The goal of this initiative is to report on the most important data gaps, and lay out pathways for funders, data providers, and researchers to address them.

If you’re interested in joining the group workshop, may I ask you to indicate your availability in the following poll (by March 15): http://whenisgood.net/3xpsa9c?

If you’re interested in participating in a one-on-one interview, please book a time here: https://calendar.app.google/RbDx9GkeDiVxBYJA8

Thank you, and we hope you can join us!

whenisgood.net

WhenIsGood: CCAI Data Gaps Virtual Workshop

Plan the time to have your meeting or event by coordinating availability with all the participants using this fuss-free online tool.

Original URL: http://whenisgood.net/3xpsa9c

👍 Justin Kay, Carly Batist, Sara Beery, Ankita Shukla

Sara Beery (sara.beery@gmail.com)

2024-03-02 17:01:45

Two week environmental hackathon coming up in May with ETH AI Center: https://hack.biodivx.org/

hack.biodivx.org

AI + Environment Ecohackathon 2024

AI + Environment Ecohackathon is an unforgettable 2 week journey to deploy impactful solutions to the Amazon Rainforest

Original URL: https://hack.biodivx.org/

🌿 Suzanne Stathatos, Robin Zbinden, Edward Bayes

:female_technologist: Suzanne Stathatos, Negar Sadrzadeh, Robin Zbinden

❤️ Patrick Beukema, Yseult Hb, Millie Chapman, Chase Van Amburg, Robin Zbinden, Amee Assad, Alison Ketz, Jon Van Oast, Ronan Wallace, Jennifer, David, Olivier Dietrich, Alan Stenhouse

👍 Piotr Tynecki, Martin Marzidovsek

👍:skin_tone_5: Prabath Gunawardane

😎 Jon Van Oast

Ronan Wallace (rwallace@macalester.edu)

2024-03-05 21:12:28

Thanks for sharing this Dr. Beery! Seems like a great opportunity, and I would love to pass it on to others. Do we know if this is exclusively an in-person event? I can't seem to find a distinction between in-person and virtual participation.

👍 Jennifer

Sara Beery (sara.beery@gmail.com)

2024-03-06 04:55:49

I'm not sure! You could reach out to the organizers and ask maybe?

Sara Beery (sara.beery@gmail.com)

2024-03-06 04:56:23

@David?

David (dwddao@gmail.com)

2024-03-16 07:21:46

Thanks for sharing, we are aiming to make it hybrid!!! @Ronan Wallace 🌿

Sara Beery (sara.beery@gmail.com)

2024-03-04 14:45:59

East Coast opportunity to present student work:

Are you a graduate student, post-doctoral fellow, or early-career professional (from anywhere in the world!) pursuing or considering the field of conservation? Join us at the Student Conference on Conservation Science (SCCS-NY) at the American Museum of Natural History, October 9-11, 2024! Applications to present a talk, speed talk, or poster are due Monday, April 1 at 5:00 PM EDT.
Learn more about this year's conference & how to apply: amnh.org/sccsny

😎 Jason Holmberg (Wild Me), Jon Van Oast, Arjun Subramonian (they/them), Andrew Schulz, David Russell, Eric Greenlee, Elizabeth Campolongo

🎉 Jon Van Oast, Antón Álvarez, annie finneran, Alan Stenhouse

👍 Luke Sheneman

Davide Coppola (davidc9320@gmail.com)

2024-03-05 22:39:09

🙋‍♂️ Hi everyone! My name is Davide, I'm a Data Scientist and AI Engineer. For the last 5 years I have been studying and working in Singapore, mainly focusing on Computer Vision, AI robustness and applications to the domain of healthcare.
🌳 Nature and wildlife have always been a cornerstone interest of mine and over the last year I started dedicating my skillset to AI4Good applications, specifically for conservation efforts. I did this thanks to the challenges on the FruitPunch AI platform, which has been a great experience so far.
🐦 Outside of work and AI, I have plenty of hobbies and among them are wildlife photography and birding!
🤝 I am here to expand my network with other people from around the world who share the same objectives and to find new opportunities to put my skills to good use.
:flag_nz: Lastly, I am currently looking to move to New Zealand, so if you're from there I would be very happy to connect!

:flag_nz: Mitchell Rogers, Shir Bar, Cynthia Wu, Alan Stenhouse

👋 Ed Miller, Ștefan Istrate, Shir Bar, Robin Zbinden, Antón Álvarez, Cynthia Wu, Alexander Merdian-Tarko

:bearid: Ed Miller, Antón Álvarez, Cynthia Wu

Ed Miller (ed@hypraptive.com)

2024-03-06 01:38:38

Welcome, Davide, and thank you for participating in the AI for Bears challenge!

🐻 Davide Coppola

Davide Coppola (davidc9320@gmail.com)

2024-03-06 02:16:40

Thanks Ed! I'm very happy to be able to contribute to this project 🙂

🙌 Ed Miller

Nate Harada (nharada1@gmail.com)

2024-03-06 06:21:13

Hey y’all! I’m Nate, a longtime machine learning researcher and practitioner who now has to brand himself as “AI guy” thanks to the current hype. I’ve spent nearly all my career (and grad school) in machine perception across a very wide modality of sensors (camera, lidar, audio, medical sensors like ECG and PPG, accelerometer, radar, some I’ve probably forgot).

A few quick hits:

I’m the creator and maintainer of Moonshine which is open source pre-trained remote sensing models for satellite. I’m looking into branching out to drone or aerial photography as well. I think modern computer vision foundation models (which I’ve spent the last 2 years on) have a lot to offer here.

I also recently built Zeroshot which helps create efficient classifiers using just text (i.e. type “shark” and get a shark classifier).

The majority of my conservation focus is climate focused, and I’ve done work as well for UNICEF helping with refugee tracking for aid resourcing, but I’m looking at spending more focus on conservation specifically. I’m new so please be gentle!

I’m based in SF, although I’m currently traveling Asia for vacation which is why it’s like 3am Pacific.

Outside work I love to cook and bake (both!), play and listen to music, and fly paragliders.

If you do any aerial conservation work such as drones, fixed wing, etc I’d love to hear what kinds of problems you’re working on and what pain-points you face 🙏

🙌 Shir Bar, Amee Assad, Justin Kay, Dante Wasmuht, Suzanne Stathatos, Omiros Pantazis, Dan Morris, Henrik Cox (Sentinel), Chase Van Amburg, Rebecca Wilks, Abhay, Elizabeth Campolongo, Cynthia Wu

😎 Jon Van Oast, Elizabeth Campolongo, Cynthia Wu

🙌:skin_tone_3: Alan Stenhouse

Jonah Fox (jonahfox@gmail.com)

2024-03-06 08:45:06

wow just checked them out - really nice projects!

Amee Assad (aa3628@columbia.edu)

2024-03-06 09:23:50

Cool projects!

Dante Wasmuht (dante@conservationxlabs.org)

2024-03-06 09:58:30

@Lasha Otarashvili @Jason Holmberg (Wild Me)

❤️ Lasha Otarashvili

Brandon Hays (brandon.hays@duke.edu)

2024-03-06 11:25:06

@David Johnston moonshine could be cool for coral reef segmentation

Dan Morris (agentmorris@gmail.com)

2024-03-06 11:57:08

Re: Moonshine...

While everyone else is complimenting the Moonshine library, I'd like to take a moment to compliment your logo. I see you, "O" shaped like a moon.

Re: drones and conservation...

You may find this page useful; I try to keep track of open datasets and open models in this area:

https://github.com/agentmorris/agentmorrispublic/blob/main/drone-datasets.md

I'm not saying you have to go train a big model on all that data and put it in the Moonshine package, but if you were already thinking of adding object detection... just saying...

Re: music...

If you're ever in the PNW, we can jam and talk about conservation technology; my real contribution to society is:

http://awesomesongbook.com/

Ben Weinstein (benweinstein2010@gmail.com)

2024-03-06 12:01:49

Hi Nate, welcome. We have worked in this area for a couple years on tree detection, bird species classification from airborne imagery. A quick list of challenges for the field might include

Generalization across image resolution for multiple data acquisition systems

Cross-sensor calibration for remote-sensing with different numbers of bands

interactive tools for batch labeling across hundreds of images using unsupervised clustering of embedded features

cross-geometry prediction of different annotation types, so going between point/polygon/boxes to make sure we have access to every data source we can get (see SAM-geo).

We work on a number of benchmark datasets for tree detection and developing more (https://milliontrees.idtrees.org/), and release our models through a python package https://deepforest.readthedocs.io/en/latest/prebuilt.html. There are large datasets that speak all these challenges, if interested, or anyone else here. @Devis Tuia’s team works a bit more on some of the robotics/engineering challenges through their new collaborative that is really interesting https://wilddrone.eu/

MillionTrees

A Benchmark Dataset for Airborne Machine Learning

The MillionTreesBenchmark

Original URL: https://milliontrees.idtrees.org/

WildDrone - Drones for Nature Conservation

Home - WildDrone

WildDrone Autonomous Drones for Nature Conservation Missions Recruitment - Join the WildDrone Team Who Are We? WildDrone is an international training network funded by the EU Marie Skłodowska Curie Actions. The network will operate from 1 January 2023 – 31 December 2026. We aim to revolutionize wildlife conservation practices by using autonomous drone technology as a

Original URL: https://wilddrone.eu/

Nate Harada (nharada1@gmail.com)

2024-03-06 19:56:48

Wow thanks for the warm welcome everyone, it’s super cool seeing what everyone is up to! I might reach out to some of you to try and learn more about what you’re doing, but also feel free to message me if you wanna chat sometime!

Lasha Otarashvili (otarashvililasha@gmail.com)

2024-03-08 10:56:25

Hi Nate! Cool projects.

I work with the Wild Me team for aerial surveying with the Scout (UI) & Scoutbot (ML plugin) products. Currently Scout is being used for aerial surveys in Africa in Kenya and Kavango-Zambezi transfrontier, with expansion work underway. Two challenges we are currently working to solve:

Imbalanced data - given the long-tailed distribution of species, trying to predict the objectness along with the species label often reduces performance for objectness, even with a decoupled head.

Dense images - tight herds. Two problems - obtaining ground-truth labels is hard when there are 1000+ objects present in a single image. Models trained on less dense images have a hard time stepping up. I feel solutions will be outside of localization-as-bbox approach.

Happy to chat if interested.

WildMeOrg/scout

Stars

Language

JavaScript

WildMeOrg/scoutbot

Stars

Language

Jupyter Notebook

:zebra_face: Jason Holmberg (Wild Me)

😎 Jason Holmberg (Wild Me)

Elizabeth Campolongo (e.campolongo479@gmail.com)

2024-03-21 19:07:40

@Dan Morris, awesome list! I'm bookmarking some of these to look at later. Also, KABR is actually available on Hugging Face (along with the telemetry data from the flights).

Dan Morris (agentmorris@gmail.com)

2024-03-21 19:19:42

Glad the list is useful. And thanks for the pointer, I added the HF link to the listing for KABR.

👍 Elizabeth Campolongo, Jason Holmberg (Wild Me)

Cynthia Wu (cynthiaswu@gmail.com)

2024-04-02 17:46:14

Welcome @Nate Harada! I have been working on tree monitoring at Taro AI 🌳 (taroai.com). So cool what you’re building at Moonshine, it looks so useful! I’m also located in SF if you ever want to go for a walk!

Sara Beery (sara.beery@gmail.com)

2024-03-06 13:12:14

https://twitter.com/WildAudioJack/status/1764595424608821691

X (formerly Twitter)

Jack Greenhalgh (@WildAudioJack) on X

Introducing The Freshwater Sounds Archive, an initiative led by myself in collaboration with @fishsoundsweb! Contribute to a global archive of soniferous freshwater life and receive recognition as a co-author in a resulting data paper. See link 👇4 more <https://t.co/QfAVAcoWZa>

Original URL: https://twitter.com/WildAudioJack/status/1764595424608821691

🎧 Jon Van Oast, Lasha Otarashvili

🐟 Carly Batist, Maddie Cusimano, Benjamin Hoffman, Burooj Ghani, Alba Márquez-Rodríguez, Holly Houliston, Shir Bar

🔊 Carly Batist, Michael Bunsen

💚 Benjamin Hoffman, Michael Bunsen, Martin Marzidovsek

👀 Ben Williams

charlotte (deshchang@gmail.com)

2024-03-08 17:21:35

Hi all! For potential NACCB (North American Congress for Conservation Biology, Vancouver, June 2024) attendees & presenters: note that there are limited travel funds available for students/early career scholars and the Google Form applications are due on March 15, 2024. Otherwise, you can register for the conference and get a discounted rate by the early bird deadline of April 12, 2024.

❤️ Sara Beery, Remi Gosselin

Mai Lazarus (mai.lazarus@gmail.com)

2024-03-12 07:27:23

Hi all, I'm new here, and it's great to get to know such an important space. I'm an ecologist (with no experience in AI) studying fish communities across life stages, and one of my projects revolves around juvenile coral reef fish. I have 7 very similar juvenile Parrotfish species that I would like to identify to the species level. I don't need to have an ID as an output, just to cluster ~100 images of these species to groups of the most similar ones. Is there some available application that does this? Any input would be appreciated. Thanks!

🐟 Shir Bar, Justin Kay

🐠 Shir Bar

Ben Weinstein (benweinstein2010@gmail.com)

2024-03-12 12:55:17

can you show us a few images? Draw the thing in the image you want as a output.

Mai Lazarus (mai.lazarus@gmail.com)

2024-03-17 07:55:07

Hi, attaching some images. What do you mean by what I would want as output?

Ben Weinstein (benweinstein2010@gmail.com)

2024-03-17 10:11:16

machine learning models take in data and return predictions do you want to 1) label each entire image with a species -> classification, 2) Label each individual in an image to species -> detection. If you want to classify them, how many images do you have of each species? Can you go to other data sources and get more from citizen science (iNat?). @Justin Kay what's the latest on the fish foundation models. @Mai Lazarus do you know any R or python, or are you look for a desktop application?

Mai Lazarus (mai.lazarus@gmail.com)

2024-03-18 07:22:49

Hi, I would like to get a classification, the identity within each picture is not as important (I would like to crop images with several individuals to have as many images as possible, if the resolution is good enough). I have about 100 images and for sure I can get much more from other data sources. I am quite experienced with R. Thanks for the help!

Ben Weinstein (benweinstein2010@gmail.com)

2024-03-18 10:17:52

probably start here. https://cran.r-project.org/web/packages/keras/vignettes/ You'll probably need atleast 50 images for most classes to get a decent model. I don't know the fish space as well, hopefully someone can jump in and say if we have a backbone model to start training from.

👍 Justin Kay

🙏 Mai Lazarus

Ben Weinstein (benweinstein2010@gmail.com)

2024-03-12 13:27:54

@Nate Harada and I had nice chat and we discussed some of the needs for model-assisted computer vision annotation environments. I am obviously biased towards airborne tools. I jotted down some thoughts here: https://github.com/weecology/AirborneFieldGuide/blob/main/README.md, copied below, what else would be really useful? Let's brainstorm! Calling out a few members that have experience here -> @Benjamin Kellenberger @Jon Van Oast @Ben Koger @Dan Morris @Zhongqi Miao

# Airborne Field Guide

# Ideas, guiding principles, and wish list

Human review is here to stay. We need rapid model integration to create faster labeling environments specific to airborne biological use-cases.

** Create an annotation platform that using existing tools (e.g. Label-studio) to detect and classify biological objects. We don't want to get bogged down in re-inventing annotation classes and UI.
** Pre-annotate imagery with existing model classes and varying levels of taxonomic detail ("duck" versus "White-winged Scoter")
** Batch labeling tools that operate on the flight or campaign level to interactively cluster and label detections grouped by distance in embedded space. In the clustering you can re-label groups of points with existing or new labels. Clicking on individual points takes you to images.
** Prompt-level engineering to find images with certain characteristics. "Find me all the images over water"
** Both detection level and image-level query to find detections similar to target.
** Pre-computed zoom levels based on detection density -> https://openaccess.thecvf.com/content_CVPRW_2020/papers/w11/Li_Density_Map_Guided_Object_Detection_in_Aerial_Images_CVPRW_2020_paper.pdf
** Nightly model training and re-labeling, re-clustering.
** Label propogation at the image level. If I click on one animal in a flock/herd, it should auto-update nearby objects.
** Label propogation at the annotation level, using SAM to go between points, boxes, and polygons.
** On new mission, draw a bounding box of geographic area, query the ebird API/map of life/Inaturalist to get abundance curve and filter species list.

# Remaining questions
** Local Desktop installer? Especially for field researchers around the world? A stripped down version.
** How to learn from AIDE? From Scout? Fathomnet Portal, SAM-geo. Should we just merge with there? How do we promote community collaboration and avoid re-invention.
** https://www.tator.io/? Another option.
** BioCLIP foundation model -> https://arxiv.org/abs/2311.18803 versus more bespoke models? Engaging teams INat/Ebirds teams.

👀 Suzanne Stathatos, Timm Haucke, Rebecca Wilks, Shir Bar, Edward Bayes, Subhransu Maji, Sara Beery, Elizabeth Campolongo, Sam Lapp

🎉 Jon Van Oast, Elizabeth Campolongo, Alan Stenhouse

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2024-03-12 13:28:43

@Lasha Otarashvili

Suzanne Stathatos (suzanne.stathatos@gmail.com)

2024-03-12 13:32:08

@Kakani Katija

Dan Morris (agentmorris@gmail.com)

2024-03-12 14:00:04

My go-to recommendation has been "ask Ben Weinstein for his Label Studio template and do exactly what he did"; I'm not sure what I should make of the fact that my recommendation for Ben's setup is stronger than Ben's recommendation for Ben's setup. But FWIW, Ben, what you're doing felt like exactly the right amount of not reinventing the wheel, but also exactly the right amount of not taking dependencies on complicated ML features in Label Studio (rather just having good tools for fine-tuning and running models outside of LS, e.g. overnight, and loading the results into LS). So IMO a great starting point would be for you to document what you've done with Label Studio to the point where others could try it. Eh?

👍 Ben Weinstein, Sara Beery

😂 Jon Van Oast

👍:skin_tone_3: Alan Stenhouse

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2024-03-12 14:07:02

Nice list above! We have been actively developing SmarterLabelme https://github.com/robot-perception-group/smarter-labelme, for many of the listed tasks, focusing on aerial videos with many individuals in the view. We are using it for detection, tracking, label propogation, etc. but also a lot for quickly auto-annotating behaviors... feel free to check out or ask me about it. The behavior version is here: https://github.com/robot-perception-group/animal-behaviour-inference

robot-perception-group/smarter-labelme

Stars

Language

Python

robot-perception-group/animal-behaviour-inference

Stars

Language

Python

👍 Ben Weinstein

😎 Jon Van Oast

Ben Weinstein (benweinstein2010@gmail.com)

2024-03-12 14:23:57

That's great, can you talk alittle about what the largest time investments were? What have you learned so far?

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2024-03-12 14:42:10

Sure. It depends on the exact task, but typically one would spend most time in the initial expert labeling to get a good detection model. If such a model exists (which is the case for several species now), then you gain speed even there. For behaviors classification, for example, we spent a couple of days to annotate some initial sequences. Then used it to learn a classifier, which was then used to auto annotate (using the framework) to 'quickly' annotate more and more and then use those many new annotations further improve the accuracy of the model. What we have learned -- one thing, at least for the behavior classification task, that stands out to us is that some behaviors are very 'temporal' in nature and hard to learn from 'still annotations'. Another thing is that to use the auto-generated annotations for retraining, one has to still go over the annotations to correct the few wrong ones. Even though this is super fast than what it would take to annotate all those videos in the first place, it is still unavoidable.

Ben Weinstein (benweinstein2010@gmail.com)

2024-03-12 14:56:58

Great! but I meant what the time investments were for designing the system. What took the most time to create, using QT, image backend, model deployment?

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2024-03-12 14:59:56

@Eric Price can answer that best

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-03-12 15:11:21

Hi. We, too extended an existing open source tool for annotation and integrated models for detection and tracking into it, so we spent almost no time on the user interface, except adding a few buttons/key shortcuts for the new functionality. Maybe 2 days to understand the code and API, and one more day extending GUI. - but depending on the complexity and functionality of the tool that can be more elaborate, especially for web based tools. The main work was definitely model deployment - on the order of weeks, but thats mostly because we combined multiple models and extended them beyond what the base models could do on their own so there was new functionality. Deploying an existing model should be much quicker 1-2 days maybe. it depends also if the frameworks are compatible. in our case the GUI was python and the model was written in pytorch, so we could just call it directly, but there's combinations where that could be more time consuming, especially for cloud based tools.

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-03-12 15:12:30

does that answer your question?

👍 Ben Weinstein, Sara Beery

Kakani Katija (kakani@mbari.org)

2024-03-12 16:13:51

@Ben Weinstein been meaning to email you back and set up a time to chat. In two weeks?

Ben Weinstein (benweinstein2010@gmail.com)

2024-03-12 16:32:02

for sure, I know you are super busy, take your time.

Ben Weinstein (benweinstein2010@gmail.com)

2024-03-12 17:43:10

I feel like I should also mention VIAME here @Daniel Davila https://www.viametoolkit.org/

👍 Aamir Ahmad, Sara Beery

Jonah Fox (jonahfox@gmail.com)

2024-03-13 16:55:50

wow some amazing stuff here guys!

Jonah Fox (jonahfox@gmail.com)

2024-03-13 16:57:13

I'm interested in tagging of aerial data for habitats. E.g. "road", "tree", "hedgerow" and then more specific if there's the detail - e.g. specific species - or "species rich hedgerow"
Does anyone know of any work in this area in particular ?

Lasha Otarashvili (otarashvililasha@gmail.com)

2024-03-14 14:21:43

Something that could be useful - dealing with image overlap. Many fly-by styles that do continuous collection produce images that are partially overlapping - making it possible to overcount the bounding-box predictions unless dealt with. Scout has a feature to draw a straight line to discard an area in the image that appears elsewhere. This works well for plane-mounted cameras as they fly in long straight lines - so there is only one axis of overlap. Drone-mounted cameras might produce overlap in both dimensions - think of a reference image having 4 adjacent overlapping images from N,E,S,W.

Sachin Wani (sachin27071998@gmail.com)

2024-03-13 10:31:33

Hi all! I was invited to this Slack group by Dan Morris. I am a new Data Scientist currently at Lenovo in NC, USA. I did my Masters at Rutgers (graduated last Summer). I was lucky to work with Island Conservation to build a custom-trained classifier that can detect native species of Robinson Crusoe Island (off the coast of Chile). Thanks to Dan's help and Kyra Swanson's animl-py repository, I was able to build a classifier that works on top of MegaDetector to detect and recognize Coatis, cats, rabbits, rodents, birds, and other mammals. I look forward to interacting with you all in this group and it's great to be part of this community of amazing folks!

🙌 Justin Kay, Suzanne Stathatos, Dan Morris, Jon Van Oast, Ted Schmitt, Piotr Tynecki, Jason Holmberg (Wild Me), Bernie Boscoe, Chris Lange, Elizabeth Campolongo, Omiros Pantazis, Mohit Dubey, Sara Beery, Shir Bar, Ben Williams, David Will

David Will (david.will@islandconservation.org)

2024-03-20 22:36:41

Great to see you here Sachin!

❤️ Sachin Wani

Anna Willoughby (arwill19@gmail.com)

2024-03-21 11:50:28

Hi! it seems like your classifier has a lot overlap in species with my camera trap project canyoncritters.org I'd be interested in applying it if its available

Sachin Wani (sachin27071998@gmail.com)

2024-05-02 08:41:04

@Anna Willoughby Hi, I would be happy to help. I am not much active on slack, but you can send me an email (sachin27071998@gmail.com) and I can share the model weights that can be used with animl-py repository. https://github.com/conservationtechlab/animl-py

conservationtechlab/animl-py

Animl comprises a variety of machine learning tools for analyzing ecological data. This Python package includes a set of functions to classify subjects within camera trap field data and can handle both images and videos.

Stars

Language

Jupyter Notebook

Jennifer (jzhuge@alumni.cmu.edu)

2024-03-13 14:15:20

Hey all, those who were in normal tech and managed to successfully break into climate tech, did you do volunteer work on top of a job or quit your job and do unpaid AI for Conservation work before you were able to get an offer in climate tech? I've been applying to all these climate tech jobs (not limited to AI for conservation) to radio silence (I have an MSc in AI).

➕ Nate Harada, Alexander Merdian-Tarko

❤️ Kalindi Fonda

Cynthia Wu (cynthiaswu@gmail.com)

2024-04-02 17:25:21

When I was at Google, I got started by doing a 20% project in climate!

Tim Gardner (circuitformation@gmail.com)

2024-03-13 18:54:39

Hi All, I just heard about this group and joined. Our research team at the University of Oregon is developing self-supervised models (BERT related) to automatically parse animal vocalizations. I described some of our preliminary work in this recent talk for the Interspecies Internet - link below. I hope that we can apply these models to conservation challenges in addition to research on animal vocalizations. If we consider models built to detect the presence of a species in environmental recordings, I think it will be interesting to research what kinds of self-supervised pre-training processes lead to the most sample-efficient fine-tuned models for conservation... Happy to chat! https://www.interspecies.io/lectures/timgardner

Interspecies Internet

Tim Gardner, TweetyBERT: Uncovering Canary Song Syntax — Interspecies Internet

Tim presents a groundbreaking self-supervised transformer model, TweetyBERT, for analyzing bird songs without human-labeled data. By operating directly on spectrograms, TweetyBERT offers insights into canary vocalizations' statistical structure, uncovering long-range rules shaping song sequences. Th

Original URL: https://www.interspecies.io/lectures/timgardner

😎 Jon Van Oast, Elizabeth Campolongo

🐦 Dan Morris, Shir Bar, Sankalpa Ghose, Chase Van Amburg, Sara Beery, Alan Stenhouse

🙌 Ben Williams

Cara Appel (appelc@oregonstate.edu)

2024-03-14 14:37:13

Hi Tim, so interesting to hear about your research! There are quite a few of us at Oregon State University who are using bioacoustics to detect birds. It would be great to connect sometime.

Tim Gardner (circuitformation@gmail.com)

2024-03-15 12:03:39

I'd love to connect! Feel free to reach out to timg@uoregon.edu and we can schedule a time to talk.

Petar Gyurov (pgyurov93@gmail.com)

2024-03-14 10:34:18

Hi guys. I’m curious to hear what tools others here use for labelling? At my company, we’ve been self-hosting the open source edition of LabelStudio but it doesn’t check all the boxes for us.

We’re a small team so, I’m interested in something that is:

handles labelling for classification, object detection and segmentation (bonus points for video)

has some sort of data and user management

self-hostable

ideally free, but open to affordable pricing

I am doing my own research too but wanted to see if there’s something people strongly recommend! Thanks.

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-03-14 11:34:58

Not sure if that's what you are looking for:
We developed an open source tool for video annotation (smarter-labelme). It can do classification, object detection, but only limited segmentation. For bounding box annotation, it uses inbuilt detector and trackers to streamline the annotation process (the tool makes annotations and the user corrects them as needed). Its using python+QT and is not web based, ideally using a computer with GPU for running the detection/tracking models. There's no inbuilt data or user management though, beyond keeping track of the annotated frames per video in the form of json files, We handled that by handing out individual video files to annotators and then collecting the annotation data on a shared folder.
https://github.com/robot-perception-group/smarter-labelme

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2024-03-14 11:45:28

Thanks @Eric Price... @Petar Gyurov there is a recent discussion on this in a thread 2 messages above in the general channel!

Ben Weinstein (benweinstein2010@gmail.com)

2024-03-14 11:50:28

linking to above thread https://aiforconservation.slack.com/archives/CLWGQ4BJ6/p1710264474399949

Ben Weinstein (https://aiforconservation.slack.com/team/UMDLZLL1K)

@Nate Harada and I had nice chat and we discussed some of the needs for model-assisted computer vision annotation environments. I am obviously biased towards airborne tools. I jotted down some thoughts here: <a href="https://github.com/weecology/AirborneFieldGuide/blob/main/README.md">https://github.com/weecology/AirborneFieldGuide/blob/main/README.md</a>, copied below, what else would be really useful? Let's brainstorm! Calling out a few members that have experience here -> @Benjamin Kellenberger @Jon Van Oast @Ben Koger @Dan Morris @Zhongqi Miao ```# Airborne Field Guide <h1>Ideas, guiding principles, and wish list</h1> Human review is here to stay. We need rapid model integration to create faster labeling environments specific to airborne biological use-cases. ** Create an annotation platform that using existing tools (e.g. Label-studio) to detect and classify biological objects. We don't want to get bogged down in re-inventing annotation classes and UI. ** Pre-annotate imagery with existing model classes and varying levels of taxonomic detail ("duck" versus "White-winged Scoter") ** Batch labeling tools that operate on the flight or campaign level to interactively cluster and label detections grouped by distance in embedded space. In the clustering you can re-label groups of points with existing or new labels. Clicking on individual points takes you to images. ** Prompt-level engineering to find images with certain characteristics. "Find me all the images over water" ** Both detection level and image-level query to find detections similar to target. ** Pre-computed zoom levels based on detection density -> <a href="https://openaccess.thecvf.com/content_CVPRW_2020/papers/w11/Li_Density_Map_Guided_Object_Detection_in_Aerial_Images_CVPRW_2020_paper.pdf">https://openaccess.thecvf.com/content_CVPRW_2020/papers/w11/Li_Density_Map_Guided_Object_Detection_in_Aerial_Images_CVPRW_2020_paper.pdf</a> ** Nightly model training and re-labeling, re-clustering. ** Label propogation at the image level. If I click on one animal in a flock/herd, it should auto-update nearby objects. ** Label propogation at the annotation level, using SAM to go between points, boxes, and polygons. ** On new mission, draw a bounding box of geographic area, query the ebird API/map of life/Inaturalist to get abundance curve and filter species list. <h1>Remaining questions</h1> ** Local Desktop installer? Especially for field researchers around the world? A stripped down version. ** How to learn from <a href="https://github.com/microsoft/aerial_wildlife_detection">AIDE</a>? From <a href="https://www.wildme.org/scout.html">Scout</a>? <a href="https://fathomnet.org/fathomnet/#/">Fathomnet Portal</a>, <a href="https://github.com/opengeos/segment-geospatial">SAM-geo</a>. Should we just merge with there? How do we promote community collaboration and avoid re-invention. ** <a href="https://www.tator.io/">https://www.tator.io/</a>? Another option. ** BioCLIP foundation model -> <a href="https://arxiv.org/abs/2311.18803">https://arxiv.org/abs/2311.18803</a> versus more bespoke models? Engaging teams INat/Ebirds teams.```

Original URL: https://aiforconservation.slack.com/archives/CLWGQ4BJ6/p1710264474399949

Petar Gyurov (pgyurov93@gmail.com)

2024-03-14 12:20:41

Thanks, I wasn’t able to see the previous messages in the channel for some reason but I can now. @Eric Price , Smarter-Labelme looks great, but I am not sure it fills the gaps that my team faces in LabelStudio. Will definitely keep an eye on its development though.

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-03-14 13:14:25

Thanks. If you know any features that you'd like to see implemented or think are missing, feel free to tell us 🙂

Dan Morris (agentmorris@gmail.com)

2024-03-14 14:59:50

@Petar Gyurov Which of those boxes does the self-hosted Label Studio not check? I'm not claiming it's perfect, it just seems to check those boxes, and I've been impressed with it to the degree that I've tried it. My only gripes with Label Studio were that for very small (n=1 person) efforts where some amount of customization is required, it was easier to quickly modify a client-side tool like Label Me, but Label Studio still feels like the right solution for what you're describing.

👀 Sara Beery

Petar Gyurov (pgyurov93@gmail.com)

2024-03-18 04:48:59

@Dan Morris You’re right, I left out some detail. Here’s some things we struggle with on the free version:

user management — I believe the paid version has this but currently you can’t restrict users to a specific project on the free version. Also, users can’t even reset their password (unless I’m missing something)

data exploration — there’s no easy way to filter on “images with label x ” ; You can filter on the “Annotations” field but that just gives you the raw JSON for some reason and it’s a pain to work with. Can be crafted as a SQL query but not on the UI. Same goes for more complicated filters like “has labels x and y but not z”

I’d also love a summary of my data/label distribution (e.g: only 5% of your dataset represents class z)

We deal a lot with large image files:

For some reason LS takes a lot longer to load/render a large image than it should.

I’d love a way to automatically tile large images and label the tiles then load those tiles into my model (this is something we do manually at the moment)

We deal a lot with metadata:

Each image has a lot of information behind it that currently sits in a separate database

Related to the dataset distribution I mentioned above, I’d love to be able to plug that data in LS and explore it

The complexity of LS’s architecture is making us hesitant to fork it and develop against it. For example, I am not sure why the database needs to be so involved but I’m sure there’s a good reason for it.

I have some complaints about the UI in places but in general it’s OK.

We’re not on the latest version so perhaps some of my gripes have been addressed. The paid version is probably worth it just for the user management but as a small team that only has spikes of labelling throughout the year, the pricing is a bit prohibitive. Having said that, LS Enterprise is probably the cheapest out there from my research.

We’ll never have a perfect tool that has all the features we want, and LS gets close to it, but just wanted to see what else is out there.

Dan Morris (agentmorris@gmail.com)

2024-03-18 15:46:24

This is a really good summary, thanks. This is consistent with the reason that for my own small projects - where I'm the one doing the labeling - I chose Label Me, even after being impressed by Label Studio. The issue I had with LS was primarily the architectural complexity, but it's complex because it's powerful and has all that user management functionality and the ability to handle every imaginable data source and modality, I don't think we can have simplicity and all those things. I also found the metadata management in LS - including just importing existing annotations - to be a bit of a hassle. Well-documented, but a hassle.

Sidebar: this isn't really related to LS, but in terms of running a model with automatic tiling of large images, check out SAHI (https://github.com/obss/sahi).

In any case, thanks for this summary! I don't think any of these problems are unique to conservation, and I'll still come down on the side of re-using off-the-shelf tools rather than re-building. Everyone is likely to have a slightly different set of priorities, but it's good that there's a growing number of stable labeling tools that hit different points on the spectrum from "agile but feature-limited" to "cumbersome but feature-rich" (Label Me and Label Studio are maybe total opposites on that spectrum).

Petar Gyurov (pgyurov93@gmail.com)

2024-03-19 05:56:04

SAHI looks really interesting, thank you for sharing 👍

Urs (urs.waldmann@uni-konstanz.de)

2024-03-15 13:21:29

Send in your favorite animal papers to our CV4Animals workshop @CVPR by March 27!

Look forward to the exciting program and hope to see many animal enthusiasts in Seattle!

https://www.cv4animals.com

🙌 Sara Beery, Suzanne Stathatos, Jon Van Oast, Justin Kay, Yseult Hb, Mark Goldwater, Ishan Nangia, Subhransu Maji, Piotr Tynecki, Oisin Mac Aodha, Mitchell Rogers, Gengshan Yang, Levi Cai, Rohan Sawahn

Urs (urs.waldmann@uni-konstanz.de)

2024-03-24 08:22:21

Deadlines extended!

Submission Deadline: 11:59 pm, April 9, 2024 (Pacific Time)
Notification of Decision: April 29, 2024

Urs (urs.waldmann@uni-konstanz.de)

2024-03-24 08:22:30

We will also provide a number of free conference and workshop registrations for the outstanding papers, thanks to our generous sponsors.

Ishan Nangia (ishannangia.123@gmail.com)

2024-03-15 15:09:44

Hey everyone. New here.
I'm a conservation scuba diver + data scientist working on computer vision tasks for a couple of NGOs based out of India and diving for a marine conservation NGO here.
Love the discussions happening here. Glad to have found this space! 🌊

🌊 Suzanne Stathatos, Ariel Chamberlain, Shir Bar, Yseult Hb, Arthur Caillau, Malte Pedersen, Rebecca Wilks

😎 Jon Van Oast, Gracie Ermi

🐟 Shir Bar

🙌 Ben Williams

Ishan Nangia (ishannangia.123@gmail.com)

2024-03-16 14:05:31

Yup. Already joined it ^_^

I am working with Coastal Impact India where we have set up artificial coral reefs that we are now monitoring using AI. Using instance segmentation algorithms to detect and size up corals.

🌊 Kalindi Fonda

👍 Kishore Panaganti

:i_love_you_hand_sign: Tarun

Owen Xing (owenxing1994@gmail.com)

2024-03-17 21:29:27

Hi guys😆! My name is Owen Xing. I am a first-year PhD student at Griffith University, Australia. As you may be aware, koalas are Australia's symbol and iconic animal. I am currently engaged in a project entitled "Predicting Koala Road Crossing Behaviors using AI-Powered Observation Networks." to use camera traps to spot wild koalas in the suburban area. Recently, I came across the website climatechange.ai, which caught my attention due to its relevance and exciting content related to my research area. I have just joined this organization and am keen to connect with individuals who share a passion for integrating AI with conservation efforts, particularly in monitoring parks and wildlife.
Looking forward to talking to everyone. ❤️

❤️ Ankita Shukla, Jason Holmberg (Wild Me), Wenyuan Zhang, Ishan Nangia, Cynthia Wu

🐨 Kalindi Fonda, Shir Bar, Chase Van Amburg, Elizabeth Campolongo, Wenyuan Zhang, Dan Morris, Holly Houliston, Cynthia Wu, Rebecca Wilks

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2024-03-17 21:57:37

Welcome!

Jonah Fox (jonahfox@gmail.com)

2024-03-18 17:15:43

Hi - is anyone aware of any classifiers for habitats from aerial images? Open source if possible !

Nate Harada (nharada1@gmail.com)

2024-03-18 18:19:23

Hey! Can you share more about your problem? Or even better, that plus a few example images and what you want the classifier to output for each one?

Ben Weinstein (benweinstein2010@gmail.com)

2024-03-18 18:22:12

With some prodding from @Dan Morris I'm writing a blog post for label studio on how we 1) cut large geospatial tiles into reasonable chunks that can be read by label-studio, 2) use foundation models to pre-annotate images using the label-studio python sdk, and 3) a couple useful utility functions for converting between geospatial (projected) and image-based (cartesian) coordinate representations of annotations. I would like some feedback on the how to make the explanations clearer. https://github.com/weecology/LabelStudio_BlogPost/blob/main/blogpost.ipynb

<https://github.com/weecology/LabelStudio_BlogPost/blob/main/blogpost.ipynb | blogpost.ipynb>

🎉 Jon Van Oast, Justin Kay, Dan Morris, Nate Harada, Timm Haucke, Sara Beery, Shir Bar, Jason Holmberg (Wild Me), Edward Bayes, Ishan Nangia, David Russell, Elizabeth Campolongo, Mitch Fennell, Thor Veen, Rebecca Wilks

Dan Morris (agentmorris@gmail.com)

2024-03-18 19:18:09

Let the record show that I didn't just prod and run, I got Ben's tutorial running (proof in the image below!). Will send some minor suggestions via email, but this looks great!

If you need more prodding, I still think there's a Chapter 2 about how to do stuff in Label Studio, then a Chapter 3 where you show all the stuff you do beyond notebook-scale, but I wouldn't recommend actually writing either of those chapters, instead IMO they should be a presentation you give at the next appropriate conference/workshop/AI4RS seminar.

But the ability to do stuff like this is another reason we shouldn't re-invent the wheel; this is some serious power-user Label Studio stuff you're doing here.

🎉 Elizabeth Campolongo, Mitch Fennell

Finn J (finnjanson@gmail.com)

2024-04-10 18:51:48

Really love this work!

Peter van Lunteren (peter@addaxdatascience.com)

2024-03-19 09:25:36

Hi! I'm looking into the feasibility of developing a sound recognition model that can detect human and mammalian sounds in Northern Namibia. The idea is to use this model to real-time notify park management about any detections. The goal is to identify (1) a general class for mammalian sounds (i.e., lion, elephant, antelope, hyena, canid, etc.) and (2) a general class for any human sounds (voice, vehicle, music, gunshots, etc.). Any other sounds will be labelled as background (birds, insects, rain, thunder, etc.). Does anyone know of certain datasets, sound-libraries, or repositories where I can find sound files for this project? A semi-extensive search online yielded in the following sources. At this point I'm just looking at global sound datasets, but the preference would be Southern Africa.

If you have any other recommendations or dos and don'ts, please let me know! Thanks in advance :)

Mammal:
- 4K mammals: https://www.gbif.org/dataset/b7ec1bf8-819b-11e2-bad2-00145eb45e9a/metrics

Human:
- 550 hunter gunshots: https://zenodo.org/records/5068271
- 850 gunshots: https://www.kaggle.com/datasets/emrahaydemr/gunshot-audio-dataset/data
- 2k gunshots: https://zenodo.org/records/7004819
- 750 gunshots: https://commons.datacite.org/doi.org/10.17632/x48cwz364j.1
- 10k gunshots: http://cadreforensics.com/audio/
- 150 gunshots: https://zenodo.org/records/3519845
- 5TB of urban sounds (cars, motorcycles, music, talking, etc.): https://osf.io/h285u/, https://github.com/brain-bzh/SilentCities
- 9k urban sounds: https://urbansounddataset.weebly.com/urbansound8k.html
- chainsaws: https://zenodo.org/records/5824433

Background (bird):
- 200k birds in several datasets: https://lila.science/otherdatasets#bioacoustics
- 700k birds: https://www.gbif.org/dataset/b1047888-ae52-4179-9dd5-5448ea342a24/metrics
- 3k birds: https://www.gbif.org/dataset/cb6e66f1-3056-404d-a341-bb856762c57c/metrics
- Unknown (website under maintenance): https://www.macaulaylibrary.org/

Background (insect):
- 20K insects: https://www.gbif.org/dataset/ecce3417-d4fd-4fab-9009-09ad3f3f62b3/metrics
- 1k insects: https://www.gbif.org/dataset/bc7c19ec-eefd-403b-931a-16a2ea1e71fc/metrics
- ?k crickets: https://osf.io/5ub4c/

Background (amphibian):
- 1536 frogs: https://www.kaggle.com/datasets/mehmetbayin/anuran-sound-frogs-or-toads-dataset
- 93k frogs: https://github.com/soundclim/anuraset

Background (ambient):
- rain: YouTube e.g.: https://www.youtube.com/watch?v=Mr9T-943BnE
- thunder: YouTube e.g.: https://www.youtube.com/watch?v=T9IJKwEspI8

gbif.org

Animal Sound Archive

The Animal Sound Archive at the Museum fuer Naturkunde Berlin (German: Tierstimmenarchiv) is one of the oldest and largest worldwide. Founded in 1951 by Professor Guenter Tembrock the collection consists now of around 130 000 records of animal voices.

Original URL: https://www.gbif.org/dataset/b7ec1bf8-819b-11e2-bad2-00145eb45e9a/metrics

Zenodo

Human animal hunting audio dataset

YouTube collected audio samples from human hunting animals, such as Coyotes, Bobcats, Wild Boars. The samples are in MP3 format, having a maximum duration of 3 seconds. The audio was classified by a semi-supervised process using a sound event classification system.

Original URL: https://zenodo.org/records/5068271

kaggle.com

Gunshot audio dataset

Gather gunshot audios from YouTube

Original URL: https://www.kaggle.com/datasets/emrahaydemr/gunshot-audio-dataset/data

Zenodo

Gunshot/Gunfire Audio Dataset

This dataset presents a multi-firearm, multi-orientation gunshot/gunfire audio dataset collected using several edge devices positioned in and around an outdoor firearm range. This dataset will aid researchers who are interested in the task of gunshot detection, firearm classification, and direction of arrival, thereby helping advance the capabilities of Internet-of-Battlefield-Things (IoBT) systems.

Original URL: https://zenodo.org/records/7004819

Ben Williams (ben.williams.20@ucl.ac.uk)

2024-03-20 10:06:47

FSD50K contains strongly labelled human sound events (+ a limited number of animal sound events) https://zenodo.org/records/4060432

Zenodo

FSD50K

FSD50K is an open dataset of human-labeled sound events containing 51,197 Freesound clips unequally distributed in 200 classes drawn from the AudioSet Ontology. FSD50K has been created at the Music Technology Group of Universitat Pompeu Fabra. Citation If you use the FSD50K dataset, or part of it, please cite our TASLP paper (available from [arXiv] [TASLP]): @article{fonseca2022FSD50K, title={{FSD50K}: an open dataset of human-labeled sound events}, author={Fonseca, Eduardo and Favory, Xavier and Pons, Jordi and Font, Frederic and Serra, Xavier}, journal={IEEE/ACM Transactions on Audio, Speech, and Language Processing}, volume={30}, pages={829--852}, year={2022}, publisher={IEEE} } Paper update: This paper has been published in TASLP at the beginning of 2022. The accepted camera-ready version includes a number of improvements with respect to the initial submission. The main updates include: estimation of the amount of label noise in FSD50K, SNR comparison between FSD50K and AudioSet, improved description of evaluation metrics including equations, clarification of experimental methodology and some results, some content moved to Appendix for readability. The TASLP-accepted camera-ready version is available from arXiv (in particular, it is v2 in arXiv, displayed by default). Data curators Eduardo Fonseca, Xavier Favory, Jordi Pons, Mercedes Collado, Ceren Can, Rachit Gupta, Javier Arredondo, Gary Avendano and Sara Fernandez Contact You are welcome to contact Eduardo Fonseca should you have any questions, at <a href="mailto:efonseca@google.com">efonseca@google.com</a>. ABOUT FSD50K Freesound Dataset 50k (or FSD50K for short) is an open dataset of human-labeled sound events containing 51,197 Freesound clips unequally distributed in 200 classes drawn from the AudioSet Ontology [1]. FSD50K has been created at the Music Technology Group of Universitat Pompeu Fabra. What follows is a brief summary of FSD50K's most important characteristics. Please have a look at our paper (especially Section 4) to extend the basic information provided here with relevant details for its usage, as well as discussion, limitations, applications and more. Basic characteristics: FSD50K contains 51,197 audio clips from Freesound, totalling 108.3 hours of multi-labeled audio The dataset encompasses 200 sound classes (144 leaf nodes and 56 intermediate nodes) hierarchically organized with a subset of the AudioSet Ontology. The audio content is composed mainly of sound events produced by physical sound sources and production mechanisms, including human sounds, sounds of things, animals, natural sounds, musical instruments and more. The vocabulary can be inspected in vocabulary.csv (see Files section below). The acoustic material has been manually labeled by humans following a data labeling process using the Freesound Annotator platform [2]. Clips are of variable length from 0.3 to 30s, due to the diversity of the sound classes and the preferences of Freesound users when recording sounds. All clips are provided as uncompressed PCM 16 bit 44.1 kHz mono audio files. Ground truth labels are provided at the clip-level (i.e., weak labels). The dataset poses mainly a large-vocabulary multi-label sound event classification problem, but also allows development and evaluation of a variety of machine listening approaches (see Sec. 4D in our paper). In addition to audio clips and ground truth, additional metadata is made available (including raw annotations, sound predominance ratings, Freesound metadata, and more), allowing a variety of analyses and sound event research tasks (see Files section below). The audio clips are grouped into a development (dev) set and an evaluation (eval) set such that they do not have clips from the same Freesound uploader. Dev set: 40,966 audio clips totalling 80.4 hours of audio Avg duration/clip: 7.1s 114,271 smeared labels (i.e., labels propagated in the upwards direction to the root of the ontology) Labels are correct but could be occasionally incomplete A train/validation split is provided (Sec. 3H). If a different split is used, it should be specified for reproducibility and fair comparability of results (see Sec. 5C of our paper) Eval set: 10,231 audio clips totalling 27.9 hours of audio Avg duration/clip: 9.8s 38,596 smeared labels Eval set is labeled exhaustively (labels are correct and complete for the considered vocabulary) Note: All classes in FSD50K are represented in AudioSet, except Crash cymbal, Human group actions, Human voice, Respiratory sounds, and Domestic sounds, home sounds. LICENSE All audio clips in FSD50K are released under Creative Commons (CC) licenses. Each clip has its own license as defined by the clip uploader in Freesound, some of them requiring attribution to their original authors and some forbidding further commercial reuse. Specifically: The development set consists of 40,966 clips with the following licenses: CC0: 14,959 CC-BY: 20,017 CC-BY-NC: 4616 CC Sampling+: 1374 The evaluation set consists of 10,231 clips with the following licenses: CC0: 4914 CC-BY: 3489 CC-BY-NC: 1425 CC Sampling+: 403 For attribution purposes and to facilitate attribution of these files to third parties, we include a mapping from the audio clips to their corresponding licenses. The licenses are specified in the files dev_clips_info_FSD50K.json and eval_clips_info_FSD50K.json. In addition, FSD50K as a whole is the result of a curation process and it has an additional license: FSD50K is released under CC-BY. This license is specified in the LICENSE-DATASET file downloaded with the FSD50K.doc zip file. We note that the choice of one license for the dataset as a whole is not straightforward as it comprises items with different licenses (such as audio clips, annotations, or data split). The choice of a global license in these cases may warrant further investigation (e.g., by someone with a background in copyright law). Usage of FSD50K for commercial purposes: If you'd like to use FSD50K for commercial purposes, please contact Eduardo Fonseca and Frederic Font at <a href="mailto:efonseca@google.com">efonseca@google.com</a> and <a href="mailto:frederic.font@upf.edu">frederic.font@upf.edu</a>. Also, if you are interested in using FSD50K for machine learning competitions, please contact Eduardo Fonseca and Frederic Font at <a href="mailto:efonseca@google.com">efonseca@google.com</a> and <a href="mailto:frederic.font@upf.edu">frederic.font@upf.edu</a>. FILES FSD50K can be downloaded as a series of zip files with the following directory structure: root │ └───FSD50K.dev_audio/ Audio clips in the dev set │ └───FSD50K.eval_audio/ Audio clips in the eval set │ └───FSD50K.ground_truth/ Files for FSD50K's ground truth │ │ │ └─── dev.csv Ground truth for the dev set │ │ │ └─── eval.csv Ground truth for the eval set │ │ │ └─── vocabulary.csv List of 200 sound classes in FSD50K │ └───FSD50K.metadata/ Files for additional metadata │ │ │ └─── class_info_FSD50K.json Metadata about the sound classes │ │ │ └─── dev_clips_info_FSD50K.json Metadata about the dev clips │ │ │ └─── eval_clips_info_FSD50K.json Metadata about the eval clips │ │ │ └─── pp_pnp_ratings_FSD50K.json PP/PNP ratings │ │ │ └─── collection/ Files for the *sound collection* format │ └───FSD50K.doc/ │ └───README.md The dataset description file that you are reading │ └───LICENSE-DATASET License of the FSD50K dataset as an entity Each row (i.e. audio clip) of dev.csv contains the following information: fname: the file name without the .wav extension, e.g., the fname 64760 corresponds to the file 64760.wav in disk. This number is the Freesound id. We always use Freesound ids as filenames. labels: the class labels (i.e., the ground truth). Note these class labels are smeared, i.e., the labels have been propagated in the upwards direction to the root of the ontology. More details about the label smearing process can be found in Appendix D of our paper. mids: the Freebase identifiers corresponding to the class labels, as defined in the AudioSet Ontology specification …

Original URL: https://zenodo.org/records/4060432

👍 Peter van Lunteren

Oscar Schafer (oscar.schafer@bbc.co.uk)

2024-03-20 12:20:14

You might wish to take a look at PANNs (paper) and inference Python package. We've used this to develop and trial audio monitoring (to detect human sounds) on BBC wildlife productions. There are some links to the Google AudioSet as well, which could be used if you're looking for more raw datasets.

👍 Peter van Lunteren

Peter van Lunteren (peter@addaxdatascience.com)

2024-03-21 11:55:12

Perfect, thanks!

Finn J (finnjanson@gmail.com)

2024-04-10 18:52:20

This is fascinating would love to learn more about your work, Peter

Isaac Badu (ikebaduas7@gmail.com)

2024-03-20 11:57:53

Hello everyone my name is Isaac Badu, I am currently a first year graduate student at Wake Forest university studying Biology with a focus in Quantitative Ecology. I am currently working on a project in Habitat Fragmentation and species coexistence, specifically I am looking at 1. how the matrix (non-habitat) affects dispersal behavior and mortality during dispersal of species in fragmented landscape and 2. do species coexist or exclude each other in patches after there has been fragmentation.

I am excited to be here and I look forward to chatting with you all.

🙌 Suzanne Stathatos, Jon Van Oast, Nicolas Arrieta Larraza, Shir Bar, Ishan Nangia, Brianna Rivera, Emilio Luz-Ricca

👋 Carly Batist, Alexander Merdian-Tarko

👀 Finn J

Brianna Rivera (rivvbri@gmail.com)

2024-03-20 14:46:46

Hi everyone!

I'm excited to join this space and wanted to introduce myself. My background primarily lies in life sciences, with a focus on biology. A few years ago, I completed a data science bootcamp, and since then, I've been honing my skills through personal projects. I'm eager to join the field of AI conservation. I'm interested in exploring any volunteering opportunities available. Looking forward to chatting with you all! 😊

🙌 Jon Van Oast, Dan Morris, Isaac Badu, Shir Bar, Ishan Nangia

👋 Carly Batist, Alexander Merdian-Tarko

Kit Lewers (krle4401@colorado.edu)

2024-03-21 15:44:04

Hello I'm new to this Slack, but I wanted to share that I will be running the AI session for the TDWG-SPNHC conference this year if anyone would like to submit a talk (or a talk to any other session). If you have any questions, please feel free to DM me 🙂 :

https://www.tdwg.org/conferences/2024/sessions/#sym15-emergent-ai-contributions-to-biodiversity-[…]ion-opportunities-challenges-and-a-year-in-review

tdwg.org

SPNHC-TDWG 2024 Sessions

Organized symposia, open sessions, panels, and workshops at the 2024 SPNHC-TDWG Conference

Original URL: https://www.tdwg.org/conferences/2024/sessions/#sym15-emergent-ai-contributions-to-biodiversity-data--data-standardisation-opportunities-challenges-and-a-year-in-review

👀 Sara Beery, Ishan Nangia

🎉 Alan Stenhouse

Matt Ziegler (mattzig@cs.washington.edu)

2024-03-21 16:00:39

Hi everyone! Next week I'm presenting a webinar on "Designing Equitable Ocean Technologies," on Thursday March 28 at 3:30PT!

The topic is: new emerging technologies are starting to reshape ocean management and governance: like video/AI monitoring of fishing boats, ocean observation systems incorporating a variety of sensors, and increasingly complex fish stock models. However, we know that technologies have not always had equitable impacts: like uneven Internet/device access (the "digital divide,") and racial/gender biases in AI foundation models. So, how do we ensure that ocean management technology is fair to everyone?

https://oceannexus.org/webinar/

OCEAN NEXUS

Webinar

Welcome to Ocean Nexus—Webinar Platform for ocean equity research dialogueFree and open to public Generative AI and Ocean Technologies Meet Environmental Justice Designing Equitable Ocean Technolog…

Original URL: https://oceannexus.org/webinar/

🎉 Sara Beery, Elizabeth Campolongo, Malte Pedersen, Alessandra Sellini, David Will, Justin Kay

🌊 Alessandra Sellini, Nicolas Arrieta Larraza, Levi Cai, Justin Kay

Cameron Trotter (cater@bas.ac.uk)

2024-03-22 05:29:20

Sounds interesting, will this be recorded? It's at 10:30pm my time 😬

💡 Nicolas Arrieta Larraza

Matt Ziegler (mattzig@cs.washington.edu)

2024-04-12 02:45:11

It is recorded! https://www.youtube.com/watch?v=kJRAdUyGHzM

YouTube

Ocean Nexus (https://www.youtube.com/@ocean_nexus)

Designing Equitable Ocean Technology | Ocean Equity Learning Forum

Original URL: https://www.youtube.com/watch?v=kJRAdUyGHzM

❤️ Cameron Trotter, Erik Harden

Jarrett Blair (jarrettblair@gmail.com)

2024-03-25 17:39:41

Hi everyone! We (@Jamie Alison, @Quentin Geissmann, and others) are running ACCESS-2024, a funded and selective summer school on computational entomology (primarily imaging & computer vision). It will take place in Aarhus (Denmark) between September 30th to October 4th, 2024, and we are looking for motivated students to apply (deadline April 15th). Accommodation, food and tuition are paid for – selected candidates just need to fund their travel to and from the venue.

Additional info and application details can be found on our website https://darsa.info/ACCESS-2024/. Please circulate to interested people, or repost on X. 🙂

darsa.info

ACCESS-2024

Aarhus Comprehensive Computational Entomology Summer School, 2024

Original URL: https://darsa.info/ACCESS-2024/

X (formerly Twitter)

Quentin Geissmann @qgeissmann@ecoevo.social (@qgeissmann) on X

🚨🦟💻 Great opportunity to learn about AI for insects at our (funded) Aarhus Computational Entomology Summer School (ACCESS-2024)! Apply before April 15! Co-organised by @jamiealison30 & @TokeHoye, and contributions from international experts More at <https://t.co/BXXNCr3LZV>

Original URL: https://x.com/qgeissmann/status/1763489191290196177?s=20

❤️ Suzanne Stathatos, Shir Bar, Vinicius Amaral, Quentin Geissmann, Helena Russello, Carly Batist, Antón Álvarez, David, Joe Nangle, Magali Frauendorf, Justin Kay, Roberta Hunt

😎 Jon Van Oast, Xiaojuan Liu, Justin Kay

🐞 Kaitlyn Gaynor, Carly Batist, Joe Nangle, Kalindi Fonda, Justin Kay, Cara Appel, Alan Stenhouse

🦗 Shir Bar, David Russell, Kalindi Fonda, Justin Kay

🙌 Jamie Alison, Ben Williams, Justin Kay

Justin Kay (justinkay92@gmail.com)

2024-03-28 19:20:57

So cool!

❤️ Jarrett Blair

Chris Lang (chrislang@ucsb.edu)

2024-03-27 17:32:26

Hi all! My name is Chris, I'm a software engineer at the Benioff Ocean Science Laboratory. We have been working on an object tracking model to count different types of trash/plastics that are collected on trashwheels in Baltimore Harbor. We have a unique opportunity to compare our model results with a dumpster dive event where volunteers will be manually sorting and counting the trash. Does anyone know of any papers or studies that have done something similar, comparing computer vision models to manual surveying methods, that might be good references for this type of comparison?

👋 Sara Beery, Chris Lange

Joe Ferdinando (jgf94@cornell.edu)

2024-03-27 21:30:00

This paper from the Ocean Cleanup might be a useful starting point https://www.mdpi.com/2072-4292/13/17/3401

MDPI

Quantifying Floating Plastic Debris at Sea Using Vessel-Based Optical Data and Artificial Intelligence

Despite recent advances in remote sensing of large accumulations of floating plastic debris, mainly in coastal regions, the quantification of individual macroplastic objects (>50 cm) remains challenging. Here, we have trained an object-detection algorithm by selecting and labeling footage of floating plastic debris recorded offshore with GPS-enabled action cameras aboard vessels of opportunity. Macroplastic numerical concentrations are estimated by combining the object detection solution with bulk processing of the optical data. Our results are consistent with macroplastic densities predicted by global plastic dispersal models, and reveal first insights into how camera recorded offshore macroplastic densities compare to micro- and mesoplastic concentrations collected with neuston trawls.

Original URL: https://www.mdpi.com/2072-4292/13/17/3401

🙏 Chris Lang

Chris Lange (s2125675@ed.ac.uk)

2024-03-28 09:57:42

Hi Chris. Nice name.

😆 Chris Lang

Dan Stowell (dan.stowell@naturalis.nl)

2024-03-28 11:00:34

Postdoc job with us! Postdoctoral researcher (postdoc) in AI for acoustic biodiversity monitoring https://tiu.nu/22215 An EU-funded position using AI to monitor birds by their sound. (Please share!)

❤️ Suzanne Stathatos, Oisin Mac Aodha, Clare Price, Sara Beery, Alan Stenhouse

👀 Sam Lapp

Antón Álvarez (aalvarez@wwf.es)

2024-03-30 09:20:24

After reading this amazing paper, which frames the paradigm of application-driven ML (ADML) research—predominantly the type conducted in this Slack—and provides advice on reviewing, hiring, and teaching,
I'm left wondering how you would like, or what you see as necessary, for your conservationist collaborators to improve dataset creation, task framing, and criteria for success...
I believe many of us have encountered claims of having a well-annotated dataset, only to find out it wasn't quite so. But at the same time, sometimes weak labeling can be applied to significantly reduce the workload.
Interdisciplinary collaboration is very necessary, so are you aware of any guidelines or resources that guide conservationists to better collaborate with ML professionals? Do you think it could be a useful? How do you usually do it?

The closest thing I'm familiar with is the guideline provided by @Dan Morris in his talk at CV4Ecology 2023.

YouTube

CV4Ecology (https://www.youtube.com/@cv4ecology)

Dan Morris, Google - Getting to know your AI for Conservation Project

Original URL: https://youtu.be/K_S_X6Vlp64?t=1141

➕ Justin Kay, Mitchell Rogers, Andrew Schulz

😎 Jon Van Oast

Bernie Boscoe (boscoeb@sou.edu)

2024-03-30 12:52:14

Hi Anton, thank you for sharing that excellent paper. I am thinking along the same lines, having done a similar exploration with astronomers https://arxiv.org/pdf/2211.14401.pdf and now working with conservationists. Happy to discuss!

Aran Dasan (aran@sntech.co.uk)

2024-04-02 03:34:23

Non-expert here, setting up our first ‘applications-driven’ computer vision project. These were helpful to help me reach out to our CV-experts that we’re going to be working with 🙂 thanks for sharing!

Burooj Ghani (buroojghani@gmail.com)

2024-04-01 15:34:57

The DCASE 2024 Few-shot Bioacoustic Event Detection task is now live. Please participate and share.

https://dcase.community/challenge2024/task-few-shot-bioacoustic-event-detection

dcase.community

Few-shot Bioacoustic Event Detection - DCASE

This task focuses on sound event detection in a few-shot learning setting for animal (mammal and bird) vocalisations. Participants will be expected to create a method that can extract information from five exemplar vocalisations (shots) of mammals or birds and detect and classify sounds in field recordings. If you are …

Original URL: https://dcase.community/challenge2024/task-few-shot-bioacoustic-event-detection

👍 Oisin Mac Aodha, Benjamin Hoffman, Dan Morris, Sara Beery, Aoife Toomey, Maddie Cusimano, mimi

💡 Nicolas Arrieta Larraza

Nina Grace Baranduin (ngb34@cam.ac.uk)

2024-04-04 11:22:49

Hi everyone 👋 is anyone working with camera trap, bioacoustic or GPS data with carnivores? Or know any groups that are working with this kind of data? Specially interested in African Wild Dogs or Wolves but would be great to hear about others too

Jenna Kline (jennamkline@gmail.com)

2024-04-04 11:33:41

Hi @Nina Grace Baranduin I'm working with a group here at Ohio State studying the African Wild Dog pack at the The Wilds here in Ohio. We are gathering data with camera traps and drones.

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-04-04 12:46:27

check out the folks at WildCru!
And African Wildlife Conservation Fund.
Zambian Carnivore Programme too

Namitha Suresh (ns873@cornell.edu)

2024-04-04 15:49:56

Hi @Nina Grace Baranduin! Also a group including me at Cornell studying dholes! We're working with bioacoustic data!

Evan Eskew (eveskew@gmail.com)

2024-04-10 13:36:16

I believe the Abrahms lab at UW (https://www.abrahmslab.com/) may also be doing some of this with wild dogs?

abrahmslab.com

Abrahms Lab

Our Mission From our oceans to savannas, animals must cope with dynamic environments that are undergoing unprecedented rates of change. An understanding of how these environmental processes shape animal behavior, ecological interactions, and population persistence is urgently needed to support

Original URL: https://www.abrahmslab.com/

Maricela Abarca (myabarca@dons.usfca.edu)

2024-04-05 14:06:48

Hello everyone 👋 I'm currently a data science graduate student, with a background in ecology. Glad to see this space and community that meshes two things I am eager to learn about and advance solutions for!

One of my final projects for school involves thinking about biodiversity monitoring solutions. I'm collecting answers from folks to these questions:

What specific problems are you trying to solve ?
What are some challenges you face in solving these problems?
What resources would help you overcome these challenges?
What kinds of technology supports your goals?
What are some risks in regard to the success of your goals and how do you mitigate them?

If anyone might have some time to answer all or any combination of these with a couple sentences, that would really help me out! Feel free to DM me.

Looking forward to engaging with everyone on frontiers in AI conservation!
Maricela

🎉 Jon Van Oast, Dunrie Greiling

🌊 Ishan Nangia

👋 Aran Dasan

Sara Beery (sara.beery@gmail.com)

2024-04-08 14:24:27

https://magazine.caltech.edu/post/what-computer-vision-can-tell-us-about-the-natural-world

Caltech Magazine

What Computer Vision Can Tell Us About the Natural World — Caltech Magazine

A Caltech-led program works to give ecologists computer-vision tools to analyze large sets of data. By Brian Smith It should not be hard to find the biggest creature in the world. Yet, amid all the noise and activity across a vast ocean, even the calls of the mighty blue whale can be vanishing

Original URL: https://magazine.caltech.edu/post/what-computer-vision-can-tell-us-about-the-natural-world

❤️ Avi Sundaresan, Ruth Oliver, Oisin Mac Aodha, Eric Greenlee, Negar Sadrzadeh, Justin Kay, Omiros Pantazis, charlotte, Dylan Van Bramer (she/her), Antón Álvarez, Ishan Nangia, Mitch Fennell, Andrew Schulz, Cara Appel, Finn J, Chase Van Amburg, Tuan-Anh VU, Aoife Toomey, Alan Stenhouse, Magali Frauendorf, G. Andrew Fricker

👍 Holger Klinck, Dan Morris, Subhransu Maji, Burooj Ghani, Nico Lang

😎 Jon Van Oast

👏 Robin Sandfort, Aakash Gupta

Jes Lefcourt (jeslefcourt@gmail.com)

2024-04-09 18:51:08

Hi everyone! The EarthRanger team is hiring! We're looking for a really good Python developer / architect to join our team. For those not familiar, EarthRanger is a philanthropic initiative by the Allen Institute for Artificial Intelligence (AI2), a non-profit founded by the late Paul Allen, co-founder of Microsoft. EarthRanger is a free software solution that helps conservationists make informed operational decisions for wildlife monitoring and protection by integrating real-time data from wildlife tracking devices, ranger radios, field sensors, etc, etc, etc and providing a comprehensive view of everything going on within a protected area or other area of interest. We're working at about 500 sites in 70 countries, and adding about 1 site per day. Come join us to help protect animals and their habitats! https://www.linkedin.com/jobs/view/3877119361/?refId=Un8cRlQBSPm0Y%2FhbsMjqCg%3D%3D&trackingId=Un8cRlQBSPm0Y%2FhbsMjqCg%3D%3D

linkedin.com

Allen Institute for AI (AI2) hiring Backend Software Architect, EarthRanger (Contract Opportunity) in United States | LinkedIn

Posted 8:01:51 PM. Backend Software Architect, EarthRanger (Contract Opportunity)Who You Are:Love animals? Have you…See this and similar jobs on LinkedIn.

Original URL: https://www.linkedin.com/jobs/view/3877119361/?refId=Un8cRlQBSPm0Y%2FhbsMjqCg%3D%3D&trackingId=Un8cRlQBSPm0Y%2FhbsMjqCg%3D%3D

❤️ Kevin Rineer, Subhransu Maji, Irina Tolkova, Sara Beery, Arthur Caillau, Jennifer, Chase Van Amburg, Alan Stenhouse

👍 Dan Morris, Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2024-04-09 21:43:22

@Nate Harada

Patrick Beukema (patrickb@allenai.org)

2024-04-10 10:41:06

Is there a good meta-analysis/review paper/recent survey on classifying/understanding/predicting elephant (or similar) behavior from GPS sequences/transponders, either forecasting or more seq2seq based?

Rebecca Wilks (R.C.Wilks@sms.ed.ac.uk)

2024-04-18 09:36:49

Whilst not a review, this paper used GPS tracks to classify bull elephant behaviour into Musth/Non-Musth state via HMMs (state space models) 🐘 https://www.savetheelephants.org/wp-content/uploads/2019/06/2019_Taylor_et_al-2019JAE_reproductivetactictsofbulls.pdf

Matt Hron (matt.hron@wildlifeprotectionsolutions.org)

2024-04-10 17:23:29

Hi everyone! Wildlife Protection Solutions is hiring a full-stack software developer (Python/React) with a focus on AI/ML for wildlife conservation. See the attached doc, the post on WildLabs, or contact info@WildlifeProtectionSolutions.org for more details about the position. Thanks!

wildlabs.net

Job Opportunity - Wildlife Protection Solutions | WILDLABS

Full-Stack Software Developer (Python/React) - Specializing in AI/ML for Wildlife Conservation

Original URL: https://wildlabs.net/career-opportunity/job-opportunity-wildlife-protection-solutions

Finn J (finnjanson@gmail.com)

2024-04-10 18:47:54

Hi everyone, keen to meet and learn from you all. I'm a Data Scientist, focused primarily on Responsible AI in Healthcare and Sustainability. I'm researching public health, biodiversity and water management in Mexico (particularly Chiapas). If anyone has any connections or experience with sustainability or environmental studies in Mexico (or other Latin American regions), do reach out!

Erik Harden (raster365@gmail.com)

2024-04-10 20:53:56

Hello everyone! Glad to see there's a large community in wildlife conservation using AI. I am a senior at my university and working on a fine tuning the Mdv5 model from Megadetector to classify certain species. I already have an annotated dataset of around 3000 pictures split up into testing, training, and validation. What are the next steps to fine the model?

Thank you in advance for any help 🙂

👋 Ankita Shukla, Kevin Rineer, Sara Beery, Valentin Gabeff, Dan Morris

Dan Morris (agentmorris@gmail.com)

2024-04-10 22:11:52

If MegaDetector is working well and you want to add species classes, 99% of the time I would advise you not to fine-tune MegaDetector. Instead, consider training a separate classifier that operates on the pixels inside the MegaDetector boxes. MEWC offers a good step-by-step process for doing this:

https://github.com/zaandahl/mewc

The only cases where I would advise fine-tuning MegaDetector are the cases where MD itself isn't working well on your species, but this is a more substantial process, and you'll need to come up with bounding boxes somehow, and 3000 images is unlikely to be enough. But, FWIW, if you want to venture into the unknown, I've done this a couple times, specifically for those cases where MD struggles (in both cases with reptiles):

https://github.com/agentmorris/unsw-goannas
https://github.com/agentmorris/usgs-tegus

Those repos give some instructions about exactly what I did, but it's a pain!

Dan Morris (agentmorris@gmail.com)

2024-04-10 22:15:09

Clarifying: the "fine-tuning MD" part is easy, not because there's anything special about MD, but because fine-tuning YOLOv5 is easy. It doesn't even require any code, the YOLOv5 CLI is amazing. But creating a dataset of bounding boxes is a PITA, so I wouldn't do that unless you need to.

👍 Sara Beery

Erik Harden (hardene@sou.edu)

2024-04-10 20:56:43

Hi everyone. I am the same person as above ^. Signed into the wrong email 😅

✔️ Jon Van Oast, Kevin Rineer

Ben Weinstein (benweinstein2010@gmail.com)

2024-04-12 18:56:00

I'm starting to play around with overhead images that lack georeferencing where we want to estimate the unique number of individuals per species in a set of imagery. Consider these two images captured from a piloted flight past a bird colony and a hand-held camera. Human reviewers can use the background textures to try to estimate overlapping areas, and then manually decide whether to count the birds in that portion of the photo. Its fairly trivial using old-school computer vision to stitch them (see example with matching keypoints), but you can easily imagine that stitching together a large mosaic will naturally warp the images, especially as the plane turns, making downstream machine learning difficult. Instead of stitching then predicting, i'm thinking about this as a 're-id' problem, except for the fact that my objects are not visually different, like whales or tigers. By using the background context, we should be able to match up duplicate detections, assuming the photos are taken in rapid succession. We can then use more general deep learning methods like (https://openaccess.thecvf.com/content/ICCV2021W/DSC/papers/Bansal_Where_Did_I_See_It_Ob[…]ance_Re-Identification_With_Attention_ICCVW_2021_paper.pdf). My idea is to use our existing object detection models, which are pretty good, and then search all corresponding images for matching background context. I know others are thinking about this space (perhaps @Lasha Otarashvili, I can see a annotation tool in Scout), all thoughts welcome. I am working with my annotation team to curate a dataset, but I sense there are others out there already. Helpful data would be more airborne and fuzzy objects, and less tiger stripe camera trap data. They feel like different problems. I think this problem of unordered, non-curated sequences of airborne photos could be useful to a large monitoring audience (e.g African Savannah herds @Ben Koger, @Howard L Frederick, Coral Reefs, and other large aggregations)

<https://github.com/OpenStitching/stitching/blob/main/stitching/feature_detector.py | feature_detector.py>

<pre><code>from collections import OrderedDict import cv2 as cv import numpy as np from .stitching_error import StitchingError class FeatureDetector: """<https://docs.opencv.org/4.x/d0/d13/classcv_1_1Feature2D.html>""" DETECTOR_CHOICES = OrderedDict() DETECTOR_CHOICES["orb"] = cv.ORB.create DETECTOR_CHOICES["sift"] = cv.SIFT_create DETECTOR_CHOICES["brisk"] = cv.BRISK_create DETECTOR_CHOICES["akaze"] = cv.AKAZE_create DEFAULT_DETECTOR = list(DETECTOR_CHOICES.keys())[0] def __init__(self, detector=DEFAULT_DETECTOR, ****kwargs): self.detector = FeatureDetector.DETECTOR_CHOICES[detector](****kwargs) def detect_features(self, img, **args, ****kwargs): return cv.detail.computeImageFeatures2(self.detector, img, **args, ****kwargs) def detect(self, imgs): return [self.detect_features(img) for img in imgs] def detect_with_masks(self, imgs, masks): features = [] for idx, (img, mask) in enumerate(zip(imgs, masks)): assert len(img.shape) == 3 and len(mask.shape) == 2 if not len(imgs) == len(masks): raise StitchingError("image and mask lists must be of same length") if not np.array_equal(img.shape[:2], mask.shape): raise StitchingError( f"Resolution of mask {idx+1} {mask.shape} does not match" f" the resolution of image {idx+1} {img.shape[:2]}." ) features.append(self.detect_features(img, mask=mask)) return features @staticmethod def draw_keypoints(img, features, ****kwargs): kwargs.setdefault("color", (0, 255, 0)) keypoints = features.getKeypoints() return cv.drawKeypoints(img, keypoints, None, ****kwargs) </code></pre>

👍 Sara Beery, Howard L Frederick, Justine Boulent

Dan Morris (agentmorris@gmail.com)

2024-04-12 19:05:52

Oooh this is one of those fun threads where we can place our bets on what will actually work, then you'll tell us how wrong we were. I see three approaches here: (1) stitch then detect (using something like SAHI to divide inference into patches), (2) detect on the original images, register the images and translate the detections accordingly (without explicitly producing a mosaic), and use basically-off-the-shelf NMS to remove redundant detections, and (3) the fancy stuff you're suggesting where you explicitly search for matching context.

My bet is that object detection will be pretty robust to warping, as long as you train it on a representative distribution of warping artifact, so my bet is on (1) being the best bang for your buck, followed by (2), and I think (3) will also work, but will get the same results for lots more code.

If you have to train on data that won't see a lot of distortion from the registration, but you expect a lot of distortion at inference time, then my bet switches from (1) to (2).

I've placed my bets!

👍 Ben Weinstein, Sara Beery

Levi Cai (levi@levicai.com)

2024-04-12 19:19:47

I would personally start with some off the shelf multi-object tracking approaches. Assuming you have high-enough frame rates, these should generally work. Something like DeepSORT or whatever the current SOTA variant of that is. I do like the idea of adding a little bit of context though around each detection to help with the matching portions.

Levi Cai (levi@levicai.com)

2024-04-12 19:27:33

I see, I guess a potential issue here is the constraint of working with un-ordered data. Though assuming you had a raw video, this shouldn't really be true? Also if you are assuming you have enough overlap, then stitching as you suggested (or fancy things like Metashape or COLMAP or something) should get you an ordering, and then you can leverage existing MOT pipelines.

Ben Weinstein (benweinstein2010@gmail.com)

2024-04-12 19:37:33

not raw video, just hand held camera photos. think a series of 10-20 photographs.

Levi Cai (levi@levicai.com)

2024-04-12 20:11:33

Ah I see, so not at all high framerate. My initial thought is (1) and (2) are probably a bit more blended. I think with (2), you could align them by stitching, but only to come up with an "ordering" and treating as a sequence, and then use MOT-style matching with inflated detection regions, you might get some nice robustness properties. With (1) I'm not sure how you would handle the actual stitch without running a detector over the individual images first anyway?

I would be curious if you're able to just flat out take the inflated detections and do a DeepSORT ranking and thresholding and just see if they match up (maybe with a bunch of discrete rotations). I think this is already fairly similar to your proposed approach, but just comparing detections between all images, rather than searching through the whole image?

👍 Ben Weinstein, Sara Beery

Jose Ruiz-Munoz (jfruizmu@unal.edu.co)

2024-04-12 21:28:13

Interesting. It sounds like a special case of multi-view object segmentation

Sara Beery (sara.beery@gmail.com)

2024-04-13 18:12:59

This is assuming the animals don't move at all?

➕ Timm Haucke

Ben Weinstein (benweinstein2010@gmail.com)

2024-04-15 12:01:27

I think they are allowed to move in small areas, but obviously not in major ways. The images are usually captured in rapid sequence, like through an airborne transect as a human photographer passes. I'll keep sharing in this thread as I proceed, thanks everyone for feedback.

👍 Sara Beery

Sara Beery (sara.beery@gmail.com)

2024-04-15 12:59:35

I could imagine that this would work better for ecosystems with some types of background texture than others, very curious to hear how things go!

👍 Ben Weinstein

Howard L Frederick (simbamangu@gmail.com)

2024-04-16 04:06:46

@Ben Weinstein this is indeed exactly the kind of thing we are interested in for the Scout annotation system (and others). I think circling a large colony or herd is a different problem but figuring out adjacency and image footprint all help solve the same problem:

We often need to de-duplicate observations in adjacent images along flight path.

Sometimes we need to just track who’s been seen in one image and figure out which members of a herd were occluded by vegetation (or neighbours) in the adjacent images.

We typically at a minimum have ordered images from the aircraft (sometimes with GPS and IMU data), and usually can be confident of 40-50% overlap between image pairs.

Howard L Frederick (simbamangu@gmail.com)

2024-04-16 04:14:10

Matching adjacent oblique images with keypoints from a moving platform seems to work really poorly with SIFT / SURF - but @Hannes Naude (Innoventix, South Africa) has a tool for footprint matching between adjacent obliques that uses ‘lightglue’ and ‘superglue’ - no need for IMU data, just checks timestamps or GPS to confirm a given pair is worth trying to match.

Howard L Frederick (simbamangu@gmail.com)

2024-04-16 04:15:38

Nifty little demo of matching available here:
https://huggingface.co/spaces/Realcat/image-matching-webui

@Ben Weinstein trying the superglue match with superglue / RANSAC on your 2 images gives 155 matches.

huggingface.co

Image Matching Webui - a Hugging Face Space by Realcat

Discover amazing ML apps made by the community

Original URL: https://huggingface.co/spaces/Realcat/image-matching-webui

Howard L Frederick (simbamangu@gmail.com)

2024-04-16 04:17:54

Thread Reply:

Hannes Naude (naude.jj@gmail.com)

2024-04-16 04:39:39

Hi all.

Thanks for pulling me in to this fascinating topic @Howard L Frederick.

Yes, this sounds like a problem that might be handled well by the registration pipeline we have in Detweb (machine guided annotation and analysis tool that we originally built for ESS, but that we are currently trying to open up to more general use). It uses approach (2) in Dan's list, which works well for game surveys (both oblique and Nadir views) in the absence of stationary content in the images (visible wings, struts, wands etc. cause it to fail pretty quickly, We expect this to be fixable with some basic masking, but that is not implemented yet). I'm a little worried about scaling for the number of sightings you need to match in a dense bird colony (we use optimal assignment aka Munkres algorithm internally which scales as O(N^2)), but am keen to give it a go. Can you share a handful of representative images somewhere? I can pull it into the system and let you know how well it works or how badly it fails.

Hannes Naude (naude.jj@gmail.com)

2024-04-16 08:25:03

OK, So I quickly started a new project in Detweb, uploaded your images and (manually) annotated the birds I saw. I then launched an image registration task and it seems to handle the matching quite well.

I include a little screencast that shows how this works.

Screenshare - 2024-04-16 2:23:18 PM.webm

🙌 Howard L Frederick, Dan Morris, Petar Gyurov

Howard L Frederick (simbamangu@gmail.com)

2024-04-17 04:21:37

@Dan Morris I’m betting #2 on your list.

Ben Weinstein (benweinstein2010@gmail.com)

2024-04-17 09:46:43

Awesome, where can I read more about the tool/strategy, is it possible for me to extract the relevant part of that code, or atleast get a sense for the general steps. I can generate a few more image pairs today. I am working on an annotation workflow for my team to generate eval data.

Ben Weinstein (benweinstein2010@gmail.com)

2024-04-17 10:26:52

Thanks @Howard L Frederick, I made a discussion on that repo, let's see what the maintainer says. https://github.com/Vincentqyw/image-matching-webui/discussions/28. Do you have sample overlapping images with annotations that we can add to an eval set? just one or two.

#28 Using this resource as a python module, contributing to creating a .config file

Hello <a href="https://github.com/Vincentqyw">@Vincentqyw</a>, thanks for your work on this tool. I am a researcher at the University Florida and an open-source developer for machine learning models for biological monitoring (<a href="https://deepforest.readthedocs.io/en/latest/">https://deepforest.readthedocs.io/en/latest/</a>). Our users often have overlapping non-georeferenced images in which there are duplicate objects and we want a unique count. Example image with birds <a href="https://private-user-images.githubusercontent.com/1208492/323249012-e65b0f4d-2d74-484d-b271-9eeafd81e42f.jpg?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.ccB1m-q8HXNioli_fF-Tdh42u_VUphzC0Qkn6gwZqns">Screenshot 2024-04-10 at 2 50 51 PM</a> Screenshot of a pair of matched image from the demo. There are birds in each image that overlap taken from a plane as it passes an area. <a href="https://private-user-images.githubusercontent.com/1208492/323248670-2f99c47e-15be-40aa-9835-4f1c22b0906e.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.lG4KUWQLo7GIiyKmM8k2uI7rWjRtMGiyJRKWfSl8cVY">image</a> Our goal is to not to create a stitched image, but rather identify the overlapping detections among images to create a unique set. I can imagine writing a workflow in our python module to 1. Run our object detection modules to predict the location and classes of objects in the image. In this case, birds. 2. Call your local GUI or pipe images to the docker container and get results. 3. Remove duplicate detections in the warped images and return users the raw images with only the unique non-overlapping detections among a set of images. This is all hypothetical, but my general sense is that there is alot of desire from the automated airborne/underwater biological monitoring community, which I help organize. Questions 1. Is this repo the right choice for this project? Can I interact with the module through python or command line? 2. What do you see as the challenges here? Thanks for your time.

Category

General

👍 Jose Ruiz-Munoz

Hannes Naude (naude.jj@gmail.com)

2024-04-18 04:00:00

@Ben Weinstein Sure, I've been promising for years now to open source the code, but I haven't gotten so far (mostly because it doesn't live up to my own standards for how clean open sourced code for general consumption should be and adding features has always outranked doing the long overdue cleanup). Last night I decided to just take the leap so the uncleaned code has been pushed to github.

So the problem of re-identifying multiple sightings as corresponding to the same individual is broken up into two parts:

Figuring out how the two (for present purposes) images relate to one another. We assume that we are dealing with aerial images from a significant height so this can be well approximated with a Homography.

Given the relation between images figuring out which sightings are paired with which and which sightings are unpaired (either because an animal was spotted in image A and missed in image B, or because an incorrect sighting was made in image A which does not match to anything in image B) The user will then typically need to manually resolve the second category.

Detweb tries to do as much as possible directly in the frontend (written in React) and just push results to serverless services on AWS (DynamoDB mostly). However part 1 above requires CUDA so it is implemented as a ECS container image with all the dependencies pre-installed, which gets launched automatically when the user pushes image registration work to a SQS queue. The relevant code is in https://github.com/WildEyeConservation/Detweb/blob/develop/cdk/containerImages/lightGlueImage/code/processSQS.py and uses LoFTR to do the matching. A much cleaner example (without all kinds of AWS integration and housekeeping clutter) is in https://github.com/WildEyeConservation/Detweb/blob/develop/cdk/containerImages/lightGlueImage/code/lightglueMatch.py You seem to be familiar with traditional keypoint matching, so there will be nothing new here for you. The only change is to use SuperPoint (rather than SIFT/SURF) for keypoint detection/description and LightGlue for matching. From that point on you can use the matched points for whatever (typically throw them into RANSAC to estimate a homography).

The most relevant code for part 2 lives in the React frontend, namely https://github.com/WildEyeConservation/Detweb/blob/develop/src/useOptimalAssignment.jsx Here I use the Munkres algorithm to calculate an optimal assignment between pairs and create a new set of annotations that are displayed to the user for confirmation. This code is currently being actively worked on and is not really commented. Most of the complexity here has very little to do with the algorithms and mostly with a self-taught React developer wrangling React to try and get an acceptable UX. 😳

I'd be happy to answer any questions you may have.

👍 Ben Weinstein

Ben Weinstein (benweinstein2010@gmail.com)

2024-04-18 14:34:09

Sure, i'll respond in DM not to clutter thread.

Ben Weinstein (benweinstein2010@gmail.com)

2024-04-25 12:42:20

For those interested in following this thread, here is where I am in terms of a reproducible example. https://github.com/weecology/DoubleCounting

🙌 Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2024-04-25 17:47:41

And with an image. This is using simple serial homography among images as described by @Hannes Naude. I still want to explore the 3d approach since it makes fewer assumptions, but for far away images and rapid sequence, this is a reasonable start. I will be wrapping this into DeepForest for users to remove double counts among a series of images.

👍 Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2024-04-25 17:48:31

If others here have sample data I can run through this, i'd like to write up a blog post. @Howard L Frederick

Ben Weinstein (benweinstein2010@gmail.com)

2024-05-01 14:30:36

For those following along on me trying to reduce double counting in non-uniformly overlapping airborne photos combined with object detection, here is the first example of the success. In blue are original detections that are deemed doubles, pink are detections that are maintained. The selection rule here is 'left-hand' meaning that when two detections overlap, we choose the earlier image. I have also written 'right-hand' and 'highest score'. I am starting to doc this if it useful to others, it should only be considered a simple solution and serve as a baseline for more sophisticated approaches. https://github.com/weecology/DoubleCounting

weecology/DoubleCounting

A repo for visualizing airborne imagery, making a prediction with a deepforest model and remove double counting among overlapping images.

Language

Python

Last updated

a minute ago

Marius Miron (marius.miron@earthspecies.org)

2024-04-15 08:55:13

Dear AI for Conservation Community,

VIHAR-2024 is the fourth international workshop on Vocal Interactivity in-and-between Humans, Animals and Robots, a satellite event of Interspeech 2024. It will take place in a hybrid format and will be hosted in Kos, Greece on 6th September 2024 and online on 9th September 2024. VIHAR-2024 aims to bring together researchers studying vocalization and speech-based interaction in-and-between humans, animals and robots from a variety of different fields. VIHAR-2024 will provide an opportunity to share and discuss theoretical insights, best practices, tools and methodologies, and to identify common principles underpinning vocal behavior in a multi-disciplinary environment.
We invite original submissions of 5-page papers (with the 5th page reserved exclusively for acknowledgements and references) or 2-page extended abstracts in all areas of vocal interactivity. Accepted papers will be compiled in the VIHAR-2024 proceedings that will be published online. All papers should follow the Interspeech 2024 template.
Suggested workshop topics may include, but are not limited to the following areas:

Self-supervised learning for vocal signals

Generative audio systems for vocal interactivity

Function of vocalizations: discovery of information embedded within signals, testing for functional reference, linking vocal signals to behavior

Physiological and morphological comparisons between vocal systems in animals and humans

Vocal imitation and social learning of vocal signals

Valence and emotion in vocal signals

Inter and intraspecies comparative analyses of vocalizations

Interspecific vocal interactivity between non-conspecifics.

Speech perception and production in human-human interactions and human-robot interactions

Comparative analysis of vocal signals in vocal interactivity.

Theory development of vocal interaction interfaces

Submission instructions can be found at the EasyChair submission page: https://easychair.org/conferences/?conf=vihar2024
Important dates (AoE):

Submission deadline : 9th June 2024

Notification of acceptance: 8th July 2024

Final versions for inclusion in proceedings: 15th July 2024

Author registration closes: 31st July 2024

Workshop: 6th and 9th September 2024

This event is sponsored by Earth Species Project (https://www.earthspecies.org/ ) and supported by the VIHAR steering committee (http://www.vihar.org/ ) and the International Speech Communication Association (http://www.isca-speech.org/).

Organizing committee: Marius Miron, Yossi Yovel, Sara Keen, Eliya Nachmani, Paola Peña, Björn Schuller, Olivier Pietquin

charlotte (deshchang@gmail.com)

2024-04-15 10:00:04

For people planning to attend the North American Congress on Conservation Biology, the early bird registration deadline has been extended from today to April 26 (FMI here). For folks who are interested in social-ecological / coupled human-natural systems and NLP, please take note of our Text Analysis for Conservation symposium!

Sako Arts (sako@fruitpunch.ai)

2024-04-16 06:23:49

Hi All, we have a new cool FruitPunch AI for Conservation Challenge coming up and we are still looking for experts in bioacoustics!
https://app.fruitpunch.ai/challenge/ai-for-forest-elephants-2
In AI for Forest Elephants we aim to build an AI to detect, elephant rumbles, gunshots and vehicle noises in collaboration with Cornell's Elephant listening project where we use an enormous data set of audio recordings collected from a grid present in the Central African rainforest.
If you are an expert you can be of real help! The Challenge starts the 30th of April and will last for 10 weeks, hit the Apply as Expert button to join in!

😎 Jon Van Oast, Dan Morris

❤️ Marius Miron, Ishan Nangia, Antón Álvarez, Aakash Gupta, Ed Miller

🐘 Alexander Merdian-Tarko, Ed Miller

Tom August (tomaug@ceh.ac.uk)

2024-04-17 08:58:43

Can anyone recommend a good paper that reviews how classification scores are created, and how they should be interpreted, for a non ML audience. I'm interested in regards to interpreting output from image classifiers. Thanks in advance

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-04-20 05:40:13

https://arxiv.org/abs/2106.04972
https://towardsdatascience.com/how-to-use-confidence-scores-in-machine-learning-models-abe9773306fa
https://towardsdatascience.com/aleatoric-and-epistemic-uncertainty-in-deep-learning-77e5c51f9423
https://blog.paperspace.com/aleatoric-and-epistemic-uncertainty-in-machine-learning/
https://arxiv.org/abs/1910.09457

arXiv.org

Understanding Softmax Confidence and Uncertainty

It is often remarked that neural networks fail to increase their uncertainty when predicting on data far from the training distribution. Yet naively using softmax confidence as a proxy for uncertainty achieves modest success in tasks exclusively testing for this, e.g., out-of-distribution (OOD) detection. This paper investigates this contradiction, identifying two implicit biases that do encourage softmax confidence to correlate with epistemic uncertainty: 1) Approximately optimal decision boundary structure, and 2) Filtering effects of deep networks. It describes why low-dimensional intuitions about softmax confidence are misleading. Diagnostic experiments quantify reasons softmax confidence can fail, finding that extrapolations are less to blame than overlap between training and OOD data in final-layer representations. Pre-trained/fine-tuned networks reduce this overlap.

Original URL: https://arxiv.org/abs/2106.04972

Medium

How to use confidence scores in machine learning models

As humans, machine learning models sometimes make mistakes when predicting a value from an input data point. But also like humans, most…

Reading time

14 min read

Original URL: https://towardsdatascience.com/how-to-use-confidence-scores-in-machine-learning-models-abe9773306fa

Medium

Aleatoric and Epistemic Uncertainty in Deep Learning

How they differ from each other and how to deal with them with TensorFlow Probability

Reading time

4 min read

Original URL: https://towardsdatascience.com/aleatoric-and-epistemic-uncertainty-in-deep-learning-77e5c51f9423

Paperspace Blog

Introduction to Uncertainty in Machine Learning Models: Concepts and Methods - part 1

In part 1 of this introductory series on uncertainty in ML models, we introduce several proven concepts and methods for identifying two types of uncertainty and evaluating them with statistical methodologies.

Written by

Adrien Payong

Filed under

Theory, Deep Learning, Machine Learning

Original URL: https://blog.paperspace.com/aleatoric-and-epistemic-uncertainty-in-machine-learning/

arXiv.org

Aleatoric and Epistemic Uncertainty in Machine Learning: An Introduction to Concepts and Methods

The notion of uncertainty is of major importance in machine learning and constitutes a key element of machine learning methodology. In line with the statistical tradition, uncertainty has long been perceived as almost synonymous with standard probability and probabilistic predictions. Yet, due to the steadily increasing relevance of machine learning for practical applications and related issues such as safety requirements, new problems and challenges have recently been identified by machine learning scholars, and these problems may call for new methodological developments. In particular, this includes the importance of distinguishing between (at least) two different types of uncertainty, often referred to as aleatoric and epistemic. In this paper, we provide an introduction to the topic of uncertainty in machine learning as well as an overview of attempts so far at handling uncertainty in general and formalizing this distinction in particular.

Original URL: https://arxiv.org/abs/1910.09457

Dan Morris (agentmorris@gmail.com)

2024-04-21 12:13:15

If you're looking for a general tutorial, Eric's links look great. If you're looking for something specific to conservation, I don't know of anything that's really a tutorial, but this paper is about confidence scores in the context of camera traps, so the introduction maybe isn't far from what you're looking for:

https://www.biorxiv.org/content/10.1101/2023.11.10.566512v2.full

Ghazi Randhawa (muhammadghazirandhawa@gmail.com)

2024-04-20 12:36:29

Are there any people here or who know of research groups who work at the intersection of AI & Data Science, regulatory Economics\Law, and Conservation topics?

charlotte (deshchang@gmail.com)

2024-04-23 18:30:21

Hey Ghazi - I’m sure that others can chime in with better examples, but I know of a few teams that fulfill at least 2 out of the 3 dimensions (AI/DS + Conservation/Environment in particular).

Lea Berrang-Ford’s collaboration

A multi-institution food systems project led by Jaron Porciello

A project on mapping climate impacts from the Mercator institute

At the risk of tooting my own horn, I’ve co-led a project with The Nature Conservancy using LLMs to categorize and map evidence for natural climate solutions and their impacts on people and nature. The first paper in this body of work is currently under review.

All of these projects largely focus on using NLP (and LLMs / transformer models) in particular to examine published (and sometimes gray lit) datasets at a large scale. I haven’t seen research groups working across AI + conservation + law in particular, but I’d be curious to hear about examples.

Nature

A systematic global stocktake of evidence on human adaptation to climate change

Nature Climate Change - Determining progress in adaptation to climate change is challenging, yet critical as climate change impacts increase. A stocktake of the scientific literature on implemented...

Original URL: https://www.nature.com/articles/s41558-021-01170-y

Nature

Accelerating evidence-informed decision-making for the Sustainable Development Goals using machine learning

Nature Machine Intelligence - Evidence syntheses produced from the scientific literature are important tools for policymakers. Producing such evidence syntheses can be highly time- and...

Original URL: https://www.nature.com/articles/s42256-020-00235-5

Nature

Machine-learning-based evidence and attribution mapping of 100,000 climate impact studies

Nature Climate Change - Evidence is growing on the impacts of climate change on human and natural systems. A two-step attribution approach—machine-learning-assisted literature review coupled...

Original URL: https://www.nature.com/articles/s41558-021-01168-6

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-04-22 10:19:20

https://x.com/lauransotomayor/status/1781739493579522500

X (formerly Twitter)

LauraNSotomayor:face_with_cowboy_hat::bat: (@lauransotomayor) on X

🌳Open database for accessing, contributing, analyzing, and visualizing remote sensing-based tree mortality data at global scale 🪵<https://t.co/5U77AZ02S4> 📡Integrating Earth observation, :female_technologist::skin_tone_2:machine learning & 🥾ground-based data sources. ~Citation: #EGU24 <https://t.co/IxjGHWerIq>

Original URL: https://x.com/lauransotomayor/status/1781739493579522500

😎 Timm Haucke, Sara Beery, David Russell, Maricela Abarca

🎉 Jon Van Oast, Sam Lapp, Clemens Mosig, Alan Stenhouse

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-04-23 11:12:21

Today, Meta and World Resources Institute are launching a global map of tree canopy height at a 1-meter resolution, allowing the detection of single trees at a global scale. In an effort to advance open source forest monitoring, all canopy height data and artificial intelligence models are free and publicly available.

https://sustainability.fb.com/blog/2024/04/22/using-artificial-intelligence-to-map-the-earths-forests/

Meta Sustainability

Using Artificial Intelligence to Map the Earth’s Forests - Meta Sustainability

An open source, global canopy height dataset and a foundational AI model for a more accountable carbon market.

Written by

10updianne, Jamie Tolan, Camille Couprie, John Brandt, Justine Spore, Tobias Tiecke, Tracy Johns, Patrick Nease

Est. reading time

7 minutes

Original URL: https://sustainability.fb.com/blog/2024/04/22/using-artificial-intelligence-to-map-the-earths-forests/

🌳 Shir Bar, Arthur Caillau, Catherine Breen, Patrick Beukema, Omiros Pantazis, Maricela Abarca, Burak Ekim, Enis Berk Çoban, Ben Weinstein, Jose Ruiz-Munoz, Timm Haucke, Prabath Gunawardane, Risa Shinoda, Clemens Mosig, mimi, Alexander Merdian-Tarko, Toryn Schafer, Ishan Nangia, Casey Youngflesh, Chase Van Amburg, Jane Wu, Eric Greenlee, Alan Stenhouse, Edward Bayes

🎉 Jon Van Oast, Tiziana Gelmi Candusso, Jane Wu

🙌 Danica Stark, Jane Wu, Arky

Nate Harada (nharada1@gmail.com)

2024-04-23 11:29:18

This is huge! The DinoV2 backbone is super powerful and having one that’s pre-trained on satellite and aerial data is amazing. It will certainly generalize well beyond trees.

🙌 Carly Batist, Edward Bayes

💯 Carly Batist, Maricela Abarca

Ben Weinstein (benweinstein2010@gmail.com)

2024-04-23 14:41:08

@Nate Harada do you want to show a fine-tuning example? We could rig up some bird data from https://zenodo.org/records/5033174

Nate Harada (nharada1@gmail.com)

2024-04-23 15:02:45

yeah was gonna try and play with this when I have some time

👍 Ben Weinstein

Sonny Burniston (sonnyburniston@yahoo.co.uk)

2024-04-23 16:04:44

Quite interested in this proposed use case above. What would the outputs of this proposed fine tuning look like?

Catherine Breen (cbreen@uw.edu)

2024-04-24 07:10:37

Thank you for sharing Carly! I am also curious in working with fine-tuned applications!

Nate Harada (nharada1@gmail.com)

2024-04-26 16:38:56

Hmm so I’m seeing a developing opinion that the actual tree height map isn’t that good, curious if people who actually need to do canopy height estimation have opinions on that specific task. I’m looking more specifically at using the backbone only as a starting point for other tasks, for example few-shot classification for wildlife or tree species detection.

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-04-26 17:21:05

@Nate Harada could you link to places you’ve seen those opinions? or is just in-passing type convos you’ve heard

Nate Harada (nharada1@gmail.com)

2024-04-26 17:54:42

Some was word of mouth, but a few tweets (obv take w a grain of salt cause some are competitors):

https://twitter.com/AdPscual/status/1783515862365831623
https://twitter.com/TC_Chakraborty/status/1783839579457614165
https://twitter.com/arjenvrielink/status/1783795483585970442

I have no skin in the game for this particular application, just sharing what I’ve seen/heard. Would be interested to hear people finding the model useful for their problems! I’d love if the backbone was broadly useful, personally.

X (formerly Twitter)

AdPscual (@AdPscual) on X

A fancy global 1-m res CHM product was released this week but I will not be using it. Shocking problems with artifacts and clouds.

Original URL: https://twitter.com/AdPscual/status/1783515862365831623

X (formerly Twitter)

TC (Tirthankar Chakraborty) (@TC_Chakraborty) on X

Would be nice to get more independent checks. Found strange shifts compared to visual imagery, which they do mention, as well as some discontinuities between tiles. Below is one example (of many) over Richland, WA, USA. A massive undertaking regardless @Meta @WorldResources

Original URL: https://twitter.com/TC_Chakraborty/status/1783839579457614165

X (formerly Twitter)

Arjen Vrielink (@arjenvrielink) on X

We had a look at the newest ~snake oil~ (pardon my French, can't help the vitriol) map of a supposedly global, thrilling, AI driven nanometer tree height map by @Meta and @WorldResources. This is what we found, a short thread:

Original URL: https://twitter.com/arjenvrielink/status/1783795483585970442

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-04-26 18:36:13

interesting feedback!

Depanshu Sani (depanshus@iiitd.ac.in)

2024-04-24 09:18:52

Hey everyone🎊 I'm glad to be a part of this community!
I am a Ph.D. student affiliated with the Vision Lab at IIIT Delhi, India. I just realized that this might be a good space to showcase our recent work on AI for Agriculture, recently accepted at IEEE WACV 2024 as an oral presentation.

SICKLE: A Multi-Sensor Satellite Imagery Dataset Annotated with Multiple Key Cropping Parameters
🌾 What is SICKLE? 📡
It is a first-of-its-kind dataset constituting a time-series of multi-resolution imagery from 3 distinct satellites: Landsat-8, Sentinel-1 and Sentinel-2. Our dataset constitutes multi-spectral, thermal and microwave sensors during the January 2018 − March 2021 period in the Cauvery Delta region of Tamil Nadu, India.

🌾 What are the key features of the dataset? 📡

The dataset consists of images from multiple satellites having a variety of sensors.

Each agricultural plot is annotated with multiple cropping parameters, for example, crop type, crop yield and phenology dates, also allowing multi-task learning.

The annotations are available at 3 scales, i.e. 30m, 10m and 3m, also allowing high-resolution inference from low-resolution satellite data.

The dataset is organized in a way that can be readily used by researchers from multiple domains, including agronomy, remote sensing and machine learning.

Each temporal sequence in the dataset is constructed by considering the cropping practices followed by farmers primarily engaged in paddy cultivation.

Tasks that can be performed: semantic segmentation, crop phenology date prediction, yield prediction, panoptic segmentation, synthetic band generation, image super-resolution, cross-satellite sensor fusion, etc.

🌾 Why is this dataset important? 📡

Monitoring the cropping pattern is a composition of multiple tasks, for instance, crop classification followed by yield estimation. SICKLE is the first work to provide annotations of multiple cropping parameters for the same set of plots.

Heterogeneous farming is followed in the study region having different growing seasons. Moreover, the primary crop (i.e. paddy) is grown 2-3 times in a year. Thus, the straight-forward way to use a time-series of satellite images (say all images from Jan-Dec) is not suitable. We present a novel strategy for preparing time-series data over a seasonal temporal window that is consistent with the regional cropping standards that are typically followed by the farmers for crop production.

Majority of the plots in this study region are small farms, with more than 95% plots having an area of less than or equal to 1 acre and an average size of 0.38 acres, thus making prediction tasks more challenging for low and medium resolution satellite images.

In case of any queries, please reach out to me at depanshus@iiitd.ac.in. Also checkout our short YouTube video to get a high-level overview of our work.
If you think this work might be beneficial for you, please visit our official website.

YouTube

Depanshu Sani (https://www.youtube.com/@depanshu-sani)

SICKLE Dataset | WACV 2024 | Oral Presentation

Original URL: https://www.youtube.com/watch?v=2p4BDVLrmdw

👋 Konstantin Klemmer, Dan Morris, Sara Beery, Holly Houliston, Risa Shinoda, Eric Greenlee, Don Cosseboom

🙌 Ishan Nangia, charlotte, Don Cosseboom

Hugo Magaldi (magaldi.hugo@gmail.com)

2024-04-26 04:01:02

[ML questions for Species Classifier]

Hi everyone ! I'm building a species classifier out of camera traps images and I am looking for advice on data related questions, as I'm new to the conservation field 🙂

is there common practice to deal with class imbalance in the training data ? I'm using weights so far so that each class has equal ponderation

same, is there a standardized way to remove pseudo-duplicates (consecutive images from a same video clip/sequence where the animal has barely moved) ? I'm experimenting with FiftyOne

Thank you in advance!

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-04-26 09:07:11

@Dan Morris

Depanshu Sani (depanshus@iiitd.ac.in)

2024-04-26 11:26:45

@Hugo Magaldi I am also new to the conservation area, so I'll give my opinion from the perspective of a core-ML/CV person.

You can try out focal loss instead of cross-entropy loss. The focal loss can simply be thought of as a variation to the cross-entropy loss, where a higher weightage is given to the samples that are hard to classify (in this case, the samples that are under-represented). It is a pretty standard and easy approach to deal with class-imbalance.

I am not sure if I understood the term pseudo-duplicate correctly. This might be because of my lack of awareness about the FiftyOne. But I am assuming that by pseudo-duplicate you mean that the same image is replicated in the dataset multiple times but with slight variations, such as cropping or rotating the image. If this is the case, I'd suggest you to first analyze the dataset. If you find that the class imbalance is just because of these pseudo-duplicates and without these duplicates the dataset would have been somewhat balanced, then you might not even have to employ the focal loss. Instead, you can implement a class-specific augmentation technique. By this I mean that for the under-represented classes in your dataset, you can virtually add more samples for such classes by applying random augmentations. A simple example of this could be that if you have two classes in your dataset, say X and Y, and only because of these pseudo-duplicates 80% of the samples in the dataset belong to X. Then in such a case, you can achieve a balanced dataset by randomly cropping, rotating and flipping the images that belong to class Y, such that the two classes have the same proportion of images.

I hope this helps. Let me know in case you have any further questions.

👍 Hugo Magaldi

Dan Morris (agentmorris@gmail.com)

2024-04-26 11:53:16

Re: pseudo-duplicates... do you mean images from the same sequence/burst that are just similar, i.e. where an animal hasn't moved between images, or literal image copies? If you mean images from the same sequence/burst where the animal hasn't moved, my recommendation would be to leave them in, unless you're really limited by training cycles.

Re: rare classes... I don't have a very useful technical answer; class weighting seems reasonable to me, and more machine-learning-y people than me can weigh in on weighting schemes and loss functions. But I'll give an unsolicited non-technical answer: I think sometimes we worry too much about performance on rare classes in a way that is detrimental to overall system performance. If you are training a real-time classifier whose only job is to report the instance of rare classes, by all means, weight those heavily. Similarly, if you have your sights set on full automation, that's... ambitious, but then yes, you need to perform well on rare classes and it might be worth paying a price in accuracy on common classes.

But for the more common scenario where you're training a classifier to help someone process a gazillion camera trap images, more often than not the benefit you provide to the user will be entirely dominated by having high precision and adequate recall on the most common classes, even if their focal species are rare, and the marginal cost of having the user review the rare classes manually is very small. I'm not telling you not to focus on rare class performance, just reminding you to consider the use case and to evaluate whether hurting performance on common classes is worth the lift to performance on rare classes (there's no free lunch!).

👍 Sara Beery, Hugo Magaldi, Shir Bar

Hugo Magaldi (magaldi.hugo@gmail.com)

2024-04-26 12:22:33

@Depanshu Sani Thank you for your answer. I will try out focal loss. By pseudo duplicates I meant very similar images coming from the same sequence. But indeed data augmentation can be useful for underrepresented classes.

Hugo Magaldi (magaldi.hugo@gmail.com)

2024-04-26 12:39:50

@Dan Morris Thank you for your insights. Taking into account the time for user review, much smaller for rare classes, is a good point. The example I had in mind is a classifier for 5 species of monkeys, with 1 specie making up for 90% of the dataset. The other 4 are not rare species, just less observed, and I still aim for a good accuracy on all 5 classes.

Dan Morris (agentmorris@gmail.com)

2024-04-27 19:54:01

Boring (but significant) update from the LILA-verse:

LILA data is now hosted on GCP, Azure, and AWS. All three copies are identical.

The only action item is for anyone who has URLs lying around that point to LILA data on Azure: all Azure URLs have changed for reasons that aren't very interesting. See dataset pages for new URLs. The old URLs will stop working any day now.

Huge thanks to the Google Cloud Public Datasets Program, the Microsoft AI for Good Lab, and Source Cooperative, who are providing hosting on GCP, Azure, and AWS, respectively.

Multi-cloud hosting is hopefully helpful for anyone working on this data on the cloud (since now with high probability the data is hosted on the cloud you're working on), but just as important, this helps with long-term stability, i.e. this data should be available on the cloud for the foreseeable future, even if any one of these hosting options disappears.

@Sara Beery and @Timm Haucke are also maintaining an offline copy at MIT, so that just in case by some catastrophe LILA loses support from all three cloud providers at the same time, it won't vanish.

And last but not least, there's also a huge pile of hard drives under my desk with all of LILA on them, so I can still work on camera trap stuff after the zombie apocalypse.

If anyone wants to see all of LILA described in one super-long markdown file, the Source Cooperative documentation is here:

https://beta.source.coop/repositories/agentmorris/lila-wildlife/description/

Thanks to GCPD, AI4G, SC, and Sara/Timm!

🙌 Justin Kay, Viktor Domazetoski, Timm Haucke, Bernie Boscoe, Felipe Parodi, Neha Hulkund, Shir Bar, Ștefan Istrate, Mitchell Rogers, Scott Smith, Jason Holmberg (Wild Me), Hugo Magaldi, Aniruddha Saha, Sara Beery, Maricela Abarca, Chase Van Amburg, Dylan Van Bramer (she/her), Matt Hron, Fagner Cunha, Holly Houliston, Carly Batist, Nico Lang, Elizabeth Campolongo, Enis Berk Çoban, Suzanne Stathatos, Malte Pedersen, Jose Ruiz-Munoz, Mitch Fennell, Toryn Schafer, Lauren Harrell, Yuval Boss, Casey Clifton, Edward Bayes, Antón Álvarez, Sam Lapp

:male_zombie: Devis Tuia, Sara Beery, Valentin Gabeff, Mitch Fennell, Aakash Gupta

🎉 Jon Van Oast

🙌:skin_tone_3: Alan Stenhouse

👍 Piotr Tynecki, Sepand Dyanatkar

Dan Morris (agentmorris@gmail.com)

2024-04-28 18:03:05

Also, in case anyone was wondering how MegaDetector does on zombies, here are MD results for more or less the first hit for "zombie apocalypse" on Google images:

😂 Mitchell Rogers, Burak Ekim, Shir Bar, Carly Batist, Elizabeth Campolongo, Mitch Fennell, Alan Stenhouse, Piotr Tynecki

🧟 Jon Van Oast, Sara Beery

Sara Beery (sara.beery@gmail.com)

2024-04-30 14:59:33

https://twitter.com/sarameghanbeery/status/1785382847135760472

X (formerly Twitter)

FathomVerse | mobile game

Online citizen science platforms like iNaturalist and Macaulay Library contain a wealth of images but are hard to search using text. We are looking for ideas so we can develop the next generation of AI models that can help users to better search these image collections.

Original URL: https://twitter.com/sarameghanbeery/status/1785382847135760472

❤️ Ankita Shukla, Oisin Mac Aodha, Shir Bar, Sonny Burniston, Elizabeth Campolongo, Timm Haucke, Gustavo Perez, charlotte, Alessandra Vidal Meza, Gabriel Manso, Dylan Van Bramer (she/her), Omiros Pantazis, Alan Stenhouse, Yuerou Tang, Andrew Schulz, Sam Heinrich, Rebecca Wilks

😎 Jon Van Oast, Timm Haucke

🙌 Maricela Abarca

👍 Piotr Tynecki

Oisin Mac Aodha (macaodha@caltech.edu)

2024-04-30 15:03:02

For those that don't use twitter:

Online citizen science platforms like iNaturalist and Macaulay Library contain a wealth of images but are hard to search using text. We are looking for ideas so we can develop the next generation of AI models that can help users to better search these image collections.

We are currently building a new dataset that consists of wildlife images combined with text descriptions. We want this dataset to be representative of the real world interesting questions that citizen scientists, ecologists, researchers, etc. want to answer.

As we get started, we hope to collect ideas from the community about what types of image searches they would be interested to perform. If this is of interest, please fill out this short form:
https://forms.gle/CmRf826AkJPkvhjX7

PS You can fill it out more than once.

❤️ Sara Beery, Jon Van Oast, Elizabeth Campolongo, charlotte, Dylan Van Bramer (she/her), Omiros Pantazis, Konstantin Klemmer, Andy Viet Huynh, Kevin Rineer, Kakani Katija

Kakani Katija (kakani@mbari.org)

2024-05-01 12:26:44

FathomVerse is now available to download for free on the App Store and Google Play!

Download on the App Store
Get it on Google Play

FathomVerse is a cozy, community science experience that will take you on an amazing journey through the depths of the ocean. Not only that, it also helps improve AI models that scientists use to discover ocean life. FathomVerse allows anyone with a smartphone or tablet to take part in ocean exploration and discovery.

🎮 Play minigames to interact with real images collected by researchers and discover ocean animals.

🔎 Hone your skills and learn how to visually identify 40+ groups of ocean animals.

❣️Save your favorite images and curate a personal gallery.

Join us on an ambitious journey to create a game that connects the power of community with cutting-edge technology for the benefit of ocean life. Dive deeper into the FathomVerse community on Instagram, TikTok, and join our Discord.

Thanks to lots of members of the AI for Conservation community, including @Sara Beery @Oisin Mac Aodha @gvanhorn @Tanya Berger-Wolf @Genevieve Patterson for input and support.

Let us know what you think!

Onwards and downwards!

FathomVerse

FathomVerse is a mobile game designed to inspire a new wave of ocean explorers and improve AI for ocean life.

Original URL: https://fathomverse.game

App Store

‎FathomVerse

‎Are you ready to embark on a journey of exploration and discovery? Introducing FathomVerse, an immersive mobile game, that offers a visually stunning experience of the ocean world and contributes to cutting-edge ocean research. By playing FathomVerse, you can help train the AI that researchers use t…

Original URL: https://apps.apple.com/us/app/fathomverse/id6469854247

play.google.com

FathomVerse - Apps on Google Play

Join a new wave of ocean explorers improving the AI used to discover ocean life.

Original URL: https://play.google.com/store/apps/details?id=org.mbari.fathomverse&pcampaignid=web_share

🦈 Ben Weinstein, Elizabeth Campolongo, Lauren Harrell, Timm Haucke, Levi Cai, charlotte

🐟 Oisin Mac Aodha, Enis Berk Çoban, Gracie Ermi, Shir Bar, Lauren Harrell, Timm Haucke, Levi Cai, Cameron Trotter

❤️ Sergei Nozdrenkov, gvanhorn, Lauren Harrell, Timm Haucke, Tarun, Levi Cai

🦀 Chris Lange, Lauren Harrell, Timm Haucke, Levi Cai, Maricela Abarca

🐠 Mitchell Rogers, Levi Cai

👍 Piotr Tynecki

PG (premsgill@gmail.com)

2024-05-01 14:02:52

Hi All. Great to part of this community. I'm curious to hear what version control and backup workflow people have in this community when we may be dealing with large assets beyond code (e.g., imagery, 3D assets, video) that Git isn't suited for.

Patrick Beukema (patrickb@allenai.org)

2024-05-01 14:43:02

Our team uses git-lfs for test cases for computer vision use cases because I like the github repo being the single source of truth for a given project. We store our data on GCP typically (in buckets) — and we do version control that (Google will do that for you albeit for an additional cost). For metadata/annotations we either release via a bucket, or on GH directly if they are small enough. We have also looked at (but don't currently use) dvc.org which we found was not optimal at the time for larger datasets (TB/PB), but that may have changed since.

One useful abstraction that is worth thinking about is decoupling the human/machine annotations of your data from the raw data. In our cases, we get raw data from a lot of public sources (like NASA/earth data/ESA), and it may be most convenient for users to access the raw data from those APIs directly (although not always)

👍 PG

Elizabeth Campolongo (e.campolongo479@gmail.com)

2024-05-03 13:40:51

We use Hugging Face as a model and dataset repository and archive (they partnered with DataCite to provide DOIs). Larger files are stored through git-LFS (at no cost), and it has a lot of nice integration features through huggingface_hub for loading models and datasets, and even running demos of various models and projects. This also makes it easier for integration with code in GitHub; you can also load assets through URLs. The UI isn't quite as nice as GitHub for active development (PRs and branches have less clear, easily accessible features since they're merged into one), but overall it's been working well for us.

huggingface.co

imageomics (HDR Imageomics Institute)

Org profile for HDR Imageomics Institute on Hugging Face, the AI community building the future.

Original URL: https://huggingface.co/imageomics

Patrick Beukema (patrickb@allenai.org)

2024-05-01 14:39:30

Hey all, question for the group about measuring daylight. We are currently training some models for maritime intelligence that would benefit from the amount of daylight at that point on earth (mostly on the ocean/high seas). Also we need it to run at reasonable scale cheaply (100M inferences/day on a small CPU ideally). Does anyone know what the best library is for that? I came across https://astral.readthedocs.io/en/latest/ which seems pretty decent but I don’t really know this space. (Also this will be used in other contexts — like on land for wildlife monitoring, if that is relevant().

Chris Doehring (chrisdo@earthranger.com)

2024-05-01 14:52:25

You might look at Ephem too.

PyPI

ephem

Compute positions of the planets and stars

Original URL: https://pypi.org/project/ephem/

🙏 Patrick Beukema

Siddharth Gupta (emailsiddha@gmail.com)

2024-05-01 19:34:51

Hello!! I'm a high school student super interested in this area - what's the best way for me to get involved?

Tom Ratsakatika (trr26@cam.ac.uk)

2024-05-02 08:03:55

Hi everyone! Has anyone implemented DeepFaune's classification model in their own project? I'm currently trying to use their model for a automated camera trap-based alert system for bears and wild boar in Romania. It would be extremely helpful to see how others have managed to get the pre-trained models to work in their own pipeline, if anyone is willing to share their code!

Ref: https://plmlab.math.cnrs.fr/deepfaune/software/-/tree/master

GitLab

Files · master · DeepFaune / DeepFaune Software · GitLab

PLMlab - La forge by Mathrice

Original URL: https://plmlab.math.cnrs.fr/deepfaune/software/-/tree/master

Tom Ratsakatika (trr26@cam.ac.uk)

2024-05-02 08:06:06

@Piotr Tynecki - any insights in addition to your posts hugely appreciated! https://wildlabs.net/discussion/successfully-integrated-deepfaune-video-alerting-system

wildlabs.net

Successfully integrated deepfaune into video alerting system | WILDLABS

Hi all, I've successfully integrated deepfaune into my Video alerting full-features security system StalkedByTheState. The yellow box around the image represents the zone of interest that I defined and the blue crosses represents matching the category of interest in the zone, in this case wolves. This will generate a new event into the system that can result in multiple actions, such as generating a video alert or flashing flood lights or other actions to try and deter the wolf from coming further if this was outside a farm for example.<https://github.com/hcfman/sbts-installStalkedByTheState> is an advanced security system that an monitor multiple cameras and trigger video alerts or other actions when it gets a match. The AI computer vision is abstracted behind a network interface which made it a simple task to integrate deepfaune.Deepfaune first uses megadetector to detect the presence of either a person, an animal or a car. I've joined this two steps and present the API in the form of a traditional yolo API so it appears to detect and classify in one step.This announcement is a bit of a pre-release really. The code is working but I still have to release this in a usable for on github.StalkedByTheState is designed to be able to run locally on NVidia's AI on the edge Jetson servers.It's now possible to define a json file defining what animals you wish to detect and generate video alerts for them. Or even invoke local control automatically from the Jetson server directly. There is support for local relay control, inputs and outputs.I still have to complete the installation script to support the new Jetson Orin series computers, it's working on them, but I've manually installed it at the moment. When this is complete I'll make a new announcement.

Original URL: https://wildlabs.net/discussion/successfully-integrated-deepfaune-video-alerting-system

Sara Si-Moussi (sara.si-moussi@univ-grenoble-alpes.fr)

2024-05-02 11:18:19

@Vincent Miele CNRS

Ed Miller (ed@hypraptive.com)

2024-05-02 16:41:07

Are you working with @Thijs and Hack the Planet?

✅ Thijs

Piotr Tynecki (piotr@tynecki.pl)

2024-05-06 05:23:10

@Tom Ratsakatika I will share the post update this week.

👍 Tom Ratsakatika

Piotr Tynecki (piotr@tynecki.pl)

2024-05-10 01:40:00

@Tom Ratsakatika did you saw this reply? I shared the code snippet in PM.

Tom August (tomaug@ceh.ac.uk)

2024-05-03 12:29:05

€2,000 travel grants are available for researchers interested in insect monitoring using automated cameras and computer vision and all the topics that surround these. Grants are available to anyone from a COST member country (https://www.cost.eu/about/members/) which is Europe + some others.

Grants cover travel, accommodation, food, and drink

You cover the awesome science

Find out more: www.insectai.eu

🙌 Casey Clifton, Carly Batist, Ishan Nangia, Antón Álvarez

👍 Piotr Tynecki

John Dziak (dziakj1@gmail.com)

2024-05-06 21:59:06

I wanted to pass this along in case anyone was interested:
"OCTO is pleased to announce that it will host:

Webinar: Netting the Future: AI's Role in Sustainable Fisheries Across the Indo-Pacific
Presented by: Stuart J. Green of Blue-Green Advisors and Farid Maruf of USAID-SUFIA-TS, Tetra Tech
Date/Time: Tuesday, May 28, 9 am US EDT/6 am US PDT/1 pm UTC/2 pm BST/3 pm CEST/8 pm WIB
Description: Artificial Intelligence (AI), Advanced Analytics (AA), and Machine Learning (ML) can be transformational in promoting fair, legal, and sustainable fisheries management across the Indo-Pacific region. This webinar will delve into the key findings of the recent USAID report "Applying AI/AA/ML in Promoting Fair, Legal and Sustainable Regional Fisheries Management in the Indo-Pacific Region." This webinar will explore emerging technological solutions that show potential in overcoming barriers to sustainable fisheries management and enhancing monitoring, analysis, and enforcement mechanisms. These innovative technologies have the potential to revolutionize fisheries management, ensuring ecological sustainability and economic viability for coastal communities.
Hosted by: OCTO
Register: https://us02web.zoom.us/webinar/register/WN_afY6fSTyRCamUa3bYnc3Kw "

👍 Justin Kay

Peter van Lunteren (peter@addaxdatascience.com)

2024-05-07 02:54:03

Does anyone know of an equivalent of MegaDetector for drone imagery?

I'm asking because the combination of working with MegaDetector as a feature extractor to locate the animals in camera trap images and then sending the crops through a project-specific classifier has proven very valuable. I was wondering if a similar pipeline of an object detector and classifier is also the standard for camera trap imagery. What is the current state of species identification models for drone imagery, and are there any open-source models I can experiment with?

After some very-very-limited tests, it looks like MegaDetector v5a actually is pretty good in locating the animals in drone images (if the images are in colour and not in infrared, and if they are properly zoomed in). Just wondering if I'm going in the right direction here...

Thanks in advance!

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-05-07 04:01:13

For drone videos, we worked on software for video annotation. It has an inbuilt detector for initial detection and then uses a tracker to track the animals across consecutive images. ( https://github.com/robot-perception-group/smarter-labelme / http://doi.org/10.1007/978-3-031-44981-9_12 ) maybe that could help in your case. It comes pre-trained with the MS-COCO object detection classes but once you have a few hundred images annotated you can re-train the detector including those, which should make it much better.

robot-perception-group/smarter-labelme

Video / Image Annotation (Polygon, Semantic mask, Classification) with Python and Re3

Stars

Language

Python

🙌 Peter van Lunteren

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-05-07 04:03:05

don't hesitate to send me a message if you have questions about installation or usage

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2024-05-07 06:35:10

Hi @Peter van Lunteren, we are actually also working on such kind of things within our research project BAMBI (https://www.bambi.eco/). Currently, we are still finishing the labelling work. But within the next months we plan to release some animal detection models trained on some thousands of individual animals, visible in ten of thousands of video frames 🙂 Maybe there is some way of collaboration there 🙂

🙌 Peter van Lunteren

Sara Beery (sara.beery@gmail.com)

2024-05-07 11:08:11

The team at WildMe has been working on this as well, @Jason Holmberg (Wild Me)

🎉 Jon Van Oast, Jason Holmberg (Wild Me)

🙌 Peter van Lunteren

Ben Weinstein (benweinstein2010@gmail.com)

2024-05-07 12:31:25

@Peter van Lunteren you can train multi-class models easily through our API as well. A general 'animal' detector is something we have discussed, but not taken up. https://deepforest.readthedocs.io/en/latest/. We have found starting from our 'bird' model you can train to other animal like deer in just a few annotations.

🙌 Peter van Lunteren

👍 Piotr Tynecki

Peter van Lunteren (peter@addaxdatascience.com)

2024-05-09 02:54:14

Thanks for the tips - this is great! I'll play around with the options and get in touch if I have specific questions. This slack channel is very valuable 🙂

❤️ Sara Beery

Steve Haddock (haddock@mbari.org)

2024-05-13 19:56:00

@Danelle Cline 👀

✅ Danelle Cline

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2024-05-07 06:42:40

Hi everyone,

some of you have probably heard of the workshop series on "Camera Traps, AI and Ecology". The workshop was held at the University of Jena (https://inf-cv.uni-jena.de/camtrap-ws/) last year in a hybrid format, respectivley online the times before (https://camtrapai.github.io/indexold.html).

This year we are going to continue to host the workshop in a hybrid format. Like that, we are pleased to invite you to the fourth international workshop on "Camera Traps, AI and Ecology," which will be held this coming September on the Hagenberg campus of the University of Applied Sciences Upper Austria (FH Oberösterreich). We are currently accepting submissions and greatly value your contributions!

This workshop is part of the international series dedicated to using artificial intelligence to monitor (wild) animals and address ecological issues. The event aims to bring together experts from various fields, including data providers such as nature parks and conservation areas, scientists such as ecologists and conservationists, and AI and Data Science experts.

We aim to promote exchange between these communities, link ecological data and AI methods, and initiate new interdisciplinary projects. The workshop will be held in a hybrid format (live and online), including invited lectures, paper presentations, and discussions.

We would be delighted to welcome you as participants at this workshop. You can participate as a listener or actively present your submission! Accepted publications will be published in the form of workshop proceedings (including DOI). In addition, we would appreciate it if you could forward this invitation to your network.

Here are the details of the workshop:
Date: 05.09. – 06.09.2024 (Paper Deadline: 28.06.)
Homepage: https://camtrap2024.fh-ooe.at/
Location: Hagenberg im Mühlkreis, Austria or Online
There will be no registration fees for participating on-site or online!

The journey to the picturesque Mühlviertel and Linz offers additional advantages during the workshop. From September 4th to 8th, the Ars Electronica Festival, the world's largest festival at the intersection of art, technology, and society (which will be part of our social event), as well as the Linzer Klangwolke on September 7th, will take place.

If you have any further questions or need additional information, please do not hesitate to contact me directly or via camtraps2024@fh-hagenberg.at .

We look forward to welcoming you to this workshop!

With best regards,
Christoph

On behalf of the rest of the organisers:
David Schedl (FH Upper Austria)
Paul Bodesheim (University of Jena)
Tilo Burghardt (University of Bristol)

camtrapai.github.io

Camera Traps, AI, and Ecology - International Workshop Archive

Original URL: https://camtrapai.github.io/indexold.html

camtrap2024.fh-ooe.at

Phil Trans 2024 Towards a toolkit for global insect biodiversity monitoring

Homepage of the Camera Traps, AI, and Ecology Workshop 2024

Original URL: https://camtrap2024.fh-ooe.at/

🙌 Jennifer, Irina Tolkova, Sara Beery, Majid Mirmehdi, Dan Morris, Talia Speaker, Malika Nisal Ratnayake

😎 Jon Van Oast

❤️ Otto Brookes

👍 Piotr Tynecki

Tom August (tomaug@ceh.ac.uk)

2024-05-08 04:39:43

Yesterday the special issue on automated insect monitoring came out in Philosophical Transactions of the Royal Society. There are papers on computer vision, acoustic monitoring, radar and molecular methods. It also has a lot of opinion pieces and perspective pieces that are relevant to all of these technologies. For example dealing with the needs of insect monitoring in low and middle income countries, data sharing, citizen science and long term monitoring.
https://royalsocietypublishing.org/toc/rstb/current

Roel van Klink is compiling a folder with all PDF's. Feel free to download everything here, and share the link:
https://www.dropbox.com/scl/fo/zjsc6oyghabhs5lngl14o/AK9MFw0GQawiMRDKYVa65Fs?rlkey=b5uofp8ogg37og2nc0jwofty1&dl=0
[text taken from a post by Roel in another Slack workspace]

Dropbox

Shared with Dropbox

Original URL: https://www.dropbox.com/scl/fo/zjsc6oyghabhs5lngl14o/AK9MFw0GQawiMRDKYVa65Fs?rlkey=b5uofp8ogg37og2nc0jwofty1&dl=0

👍 Georgia Atkinson, Shir Bar, Sara Beery, Thijs van der Plas, Justin Kay, Carly Batist, Ishan Nangia, Talia Speaker, Benjamin Hoffman, Piotr Tynecki, Isabel Fenton, Morgan Langley, Malika Nisal Ratnayake

Molly Blank (mblank@naturalstate.org)

2024-05-08 13:13:52

Hi, everyone - I understand that introductions are encouraged for newcomers. 👋 Delighted to have been pointed this incredible resource (thanks, @Dan Morris!).

:flag_ke: I’m working at Natural State, a Kenya-based non-profit developing biodiversity survey platforms that scale across landscapes to facilitate more types of conservation funding.
🥾 Our approach to AI in conservation in the near-term requires experts in landscapes and people on the ground, so we’re keen to use tools that partner well with human-in-the loop process (e.g. sorting empty images, annotation tooling for building training data).
📷 I work primarily in Subsaharan landscapes with ongoing projects in central/Northern Kenyan rangelands, and succulent Karoo. Our bread and butter data types are camera trap images, bioacoustic recordings, and oblique aerial images with remote sensing in the works.
💥 Fun fact: we're finalists in the upcoming Rainforest XPRIZE to showcase some of our data management methods designed with our Kenyan field team.

Looking forward to learning with you all!

😎 Jon Van Oast, Jason Holmberg (Wild Me), Timm Haucke

🙌 Nicolas Arrieta Larraza, Dan Morris, Shir Bar, Bernie Boscoe, Ishan Nangia, Talia Speaker, Jason Holmberg (Wild Me), Kalindi Fonda, Timm Haucke, Patrick Beukema

👋:skin_tone_2: Cara Appel

👋 Timm Haucke, Dhruvin Vora, Viktor Domazetoski, Maricela Abarca, Darshana Salvi

👍 Piotr Tynecki, Kishore Panaganti

Dan Morris (agentmorris@gmail.com)

2024-05-08 13:41:08

And in a bizarre confluence of universes, Molly's father taught me everything I know about op-amps (which still isn't a lot, but whatever I know, it's thanks to Molly's dad).

🌊 Kalindi Fonda

😎 Jason Holmberg (Wild Me)

Ben Williams (ben.williams.20@ucl.ac.uk)

2024-05-14 05:59:43

I'm on the lookout for labs experienced in digital twinning, especially those that have incorporated computer vision techniques. If anyone is aware of any such labs, particularly those that have or are interested in applying these methods to conservation efforts, then please do share 😊

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-05-14 10:05:25

this might be of interest - https://biodt.eu/

BioDT

BIODT | Digital twinning to predict how biodiversity responds to global change

Digital twinning to predict how biodiversity responds to global change

Original URL: https://biodt.eu/

🙏 Ben Williams

Joe Nangle (joe.nangle@portfoliot.com)

2024-05-14 11:25:18

Take a look at this as well.

Brian Mayton gave our team of the tour and sensor network, and it's pretty impressive stuff! Possibly more audio-oriented than your computer vision focus, but maybe still of interest?

MIT Media Lab

Project Overview ‹ Doppelmarsh: Cross-reality environmental sensor data browser – MIT Media Lab

Doppelmarsh is a cross-reality sensor data browser built for experimenting with presence and multimodal sensory experiences. Built on evolving terrain data fro…

Original URL: https://www.media.mit.edu/projects/doppelmarsh-cross-reality-environmental-sensor-data-browser/overview/

🙏 Ben Williams

Ben Williams (ben.williams.20@ucl.ac.uk)

2024-05-14 13:07:06

Thank you both, fascinating projects! Any other recommendations welcomed

Gerald Maschmann (gerald_maschmann@fws.gov)

2024-05-15 13:22:16

👋 Hi everyone!

Gerald Maschmann (gerald_maschmann@fws.gov)

2024-05-15 13:22:41

I’m a fish biologist with the US Fish and Wildlife Service in Fairbanks, Alaska. I run projects that use underwater video cameras that record salmon and other fish species passing through fish weirs. Technicians then view the video recordings and ID and count the passing fish. We use this data to estimate escapement and spawning abundance of salmon, which is then used by fish managers to manage the various salmon fisheries.

I’ve started to explore the possibilities of using A.I. to ID and count my fish for me. So, I’m here to learn more about what A.I. can do and what it can’t do. AND possibly connect with someone who could start putting together a software package that we could use.

I look forward to connecting with you. Thanks

👋 Declan, Justin Kay, Don Cosseboom, Ben Weinstein, Malte Pedersen, Timm Haucke, Sara Beery, Risa Shinoda, Enis Berk Çoban, Shir Bar, Subhransu Maji, Maricela Abarca

🙌 Ben Williams, Mohamed Elhoseiny

Ben Weinstein (benweinstein2010@gmail.com)

2024-05-15 13:30:30

https://www.frontiersin.org/articles/10.3389/fmars.2023.1200408/full, but also, https://arxiv.org/pdf/2207.09295 @Justin Kay

Frontiers

Wild salmon enumeration and monitoring using deep learning empowered detection and tracking

Pacific salmon have experienced declining abundance and unpredictable returns, yet remain vital to livelihoods, food security, and cultures of coastal communities around the Pacific Rim, creating a need for reliable and timely monitoring to inform sustainable fishery management. Currently, spawning salmon abundance is often monitored with in-river video or sonar cameras. However, reviewing video for estimates of salmon abundance from these programs requires thousands of hours of staff time, and data are typically not available until after the fishing season is completed. Computer vision deep learning can enable rapid and reliable processing of data, with potentially transformative applications in salmon population assessment and fishery management. Working with two First Nations fishery programs in British Columbia, Canada, we developed, trained, and tested deep learning models to perform object detection and multi-object tracking for automated video enumeration of salmon passing two First Nation-run weirs. We gathered and annotated more than 500,000 frames of video data encompassing 12 species, including seven species of anadromous salmonids, and trained models for multi-object tracking and species detection. Our top performing model achieved a mean average precision (mAP) of 67.6%, and species-specific mAP scores > 90% for coho and > 80% for sockeye salmon when trained with a combined dataset of Kitwanga and Bear Rivers’ salmon annotations. We also tested and deployed a ...

Original URL: https://www.frontiersin.org/articles/10.3389/fmars.2023.1200408/full

➕ Justin Kay

Gerald Maschmann (gerald_maschmann@fws.gov)

2024-05-15 13:51:46

This is great, thank you!

Anna Willoughby (arwill19@gmail.com)

2024-05-15 14:14:58

One of the graduate students at my school, Christian Swartzbaugh (css36162@uga.edu), is using AI to count and track individuals of two different fish species in experimental tanks. He's advised by Stacy Lance (https://www.lancelab.org/). I'm sure you could reach out to either of them. Attached is the abstract from our annual symposium where he presented on it.

Gerald Maschmann (gerald_maschmann@fws.gov)

2024-05-15 14:18:45

Thank you!

Timm Haucke (timm@haucke.xyz)

2024-05-15 14:30:37

I‘m involved in a similar project primarily counting river herring, led by Robert Vincent at MIT (rvincent@mit.edu). Feel free to reach out!

❤️ Justin Kay, Sara Beery

Dan Morris (agentmorris@gmail.com)

2024-05-15 14:51:20

[shamelessly pasting my reply to a similar message on the help_needed channel]

I have been maintaining a list of all the publicly-available models for finding fish in imagery that I'm aware of:

https://github.com/agentmorris/agentmorrispublic/blob/main/fish-datasets.md#publicly-available-models-for-fish-detection

@Filippo Varini and I have also been keeping track of public datasets suitable for training models like this:

https://github.com/filippovarini/filippo_public/blob/master/fish-datasets.md

👍 Gerald Maschmann, Sara Beery, Filippo Varini, Malte Pedersen

Gerald Maschmann (gerald_maschmann@fws.gov)

2024-05-15 14:52:21

Thanks Timm, I would appreciate that. Should I reach out to you or Robert?

Timm Haucke (timm@haucke.xyz)

2024-05-15 15:08:18

@Gerald Maschmann feel free to reach out to Robert first, he knows best who to include in the conversation

Thor Veen (thor@aeria.ai)

2024-05-15 15:34:53

Hi @Gerald Maschmann, we started the work on a full stack analysis pipeline for the project described in Frontiers by Atlas et al. Happy to chat what we are doing ed.

David Russell (davidrussell327@gmail.com)

2024-05-15 16:04:24

Another option is VIAME which is a toolkit that's used by a number of the NOAA fisheries offices. I was briefly involved in the development a few years back and am happy to answer basic questions or connect you with the people still working on it at Kitware.

David Russell (davidrussell327@gmail.com)

2024-05-15 16:08:24

One important features is it's designed specifically to make it easy to train models for your own tasks.

Gerald Maschmann (gerald_maschmann@fws.gov)

2024-05-15 17:02:27

Ok, thank you! This is great. I will definitely be in touch. Right now I'm in the middle of planning for the salmon season, so it might be later this summer or even this fall before I circle back.

Filippo Varini (fppvrn@gmail.com)

2024-05-15 19:58:18

Hi @Gerald Maschmann, I advice you to also check out the new MBARI open source models on hugging face. Here.

You can play around with a few examples in this interactive dashboard

huggingface.co

FathomNet (FathomNet)

FathomNet is an open-source, expert annotated image database that can be used to train, test, and validate state-of-the-art artificial intelligence algorithms to help us understand our ocean and it...

Original URL: https://huggingface.co/FathomNet

huggingface.co

Megalodon - a Hugging Face Space by FathomNet

Discover amazing ML apps made by the community

Original URL: https://huggingface.co/spaces/FathomNet/megalodon

❤️ Tarun

Chris Lange (s2125675@ed.ac.uk)

2024-05-16 07:22:07

Here is some recent work that focuses on detecting fish in sonar imagery using AI, and improving performance at new locations you haven't trained on. The authors are very knowledgeable and have been working on salmon detection. Might be a good idea to reach out to them.

https://arxiv.org/abs/2207.09295

https://aldi-daod.github.io/

arXiv.org

The Caltech Fish Counting Dataset: A Benchmark for Multiple-Object Tracking and Counting

We present the Caltech Fish Counting Dataset (CFC), a large-scale dataset for detecting, tracking, and counting fish in sonar videos. We identify sonar videos as a rich source of data for advancing low signal-to-noise computer vision applications and tackling domain generalization in multiple-object tracking (MOT) and counting. In comparison to existing MOT and counting datasets, which are largely restricted to videos of people and vehicles in cities, CFC is sourced from a natural-world domain where targets are not easily resolvable and appearance features cannot be easily leveraged for target re-identification. With over half a million annotations in over 1,500 videos sourced from seven different sonar cameras, CFC allows researchers to train MOT and counting algorithms and evaluate generalization performance at unseen test locations. We perform extensive baseline experiments and identify key challenges and opportunities for advancing the state of the art in generalization in MOT and counting.

Original URL: https://arxiv.org/abs/2207.09295

aldi-daod.github.io

Align and Distill: Unifying and Improving Domain Adaptive Object Detection

Original URL: https://aldi-daod.github.io/

Tom Wye (Fishial.ai) (twye@fishial.ai)

2024-06-04 11:25:45

@Gerald Maschmann check out fishial.ai there a demo at https://portal.fishial.ai/search/by-fishial-recognition the system can be used to collect and label image of fish. The segmentation model is spot on.. If you want to talk email support@fishial.ai

portal.fishial.ai

Fishial.AI | Enabling Fishial Recognition

We are building the largest open-source fish species image library labeled for AI machine learning and seek the development of a highly accurate open-source AI model that can identify fish species world wide!

Original URL: https://portal.fishial.ai/search/by-fishial-recognition

Robert Dawes (robert.dawes@bbc.co.uk)

2024-05-20 08:40:06

Does anyone have any insights into what are the current best classifications networks to use for animal species identification? We currently fine tune an inception v3 network - but we haven't refreshed our approach for 3 or 4 years so I'm wondering if there are now better architectures and base models around.

(This is all assuming that successful detection has already localised your animal in the image)

Robert Dawes (robert.dawes@bbc.co.uk)

2024-05-20 08:43:10

Incidentally, on a side note, I've had a play with chatGPT4o and the initial results looks pretty impressive. I asked it how it was getting the results and it said (if you can trust it) that it was making use of the text description that you might see in a spotter's guide. Which is an interesting approach for making performance of AI results more human understandable.

Sara Beery (sara.beery@gmail.com)

2024-05-20 09:10:13

@Zhongqi Miao has been looking at this recently 🙂

Nicolas Arrieta Larraza (n.arrieta.larraza@gmail.com)

2024-05-20 10:15:45

Hi hi! Is there any limitation/constraint you would need to take into account? (model size, inference speed,...)

Nowadays, I would say ResNet-like architectures are good enough 👍.

Nevertheless, in my personal opinion, good performance comes down to other factors such as: good quality data, loss function choice and a suitable data pipeline, among others.💡

👍 Robert Dawes

Nate Harada (nharada1@gmail.com)

2024-05-20 11:17:01

Suggest a pretrained transformer like DinoV2 or SigLIP unless you have a ton of labeled data or strict latency requirements

👍 Robert Dawes, Gaspard Dussert, Valentin Gabeff

Sara Beery (sara.beery@gmail.com)

2024-05-20 11:25:59

Data choices still almost always matter more than architecture

✅ Nicolas Arrieta Larraza, Bernie Boscoe, Piotr Tynecki, Edward Bayes, Kyra Swanson

👍 Robert Dawes

Piotr Tynecki (piotr@tynecki.pl)

2024-05-20 13:23:34

@Robert Dawes You can train and benchmark YOLO-based architectures (YOLOv8, YOLOv9, YOLO-NAT, etc.) with Vision Transformers (ViT) like RT-DETR, or try a novel approach based on Visual Language Models, such as CogVLM or Large Multimodal Models like PaliGemma.

All of these approaches will provide interesting and accurate results for your issue, but as @Sara Beery said, the key to your focus should be the gold standard dataset and smart + eco-based data stratification before launching fine-tuning stage.

👍 Robert Dawes

Robert Dawes (robert.dawes@bbc.co.uk)

2024-05-20 15:34:54

Interesting to hear about these transformer based approached (which of course didn't exist when we created our original pipeline)

Robert Dawes (robert.dawes@bbc.co.uk)

2024-05-20 15:38:11

Have you got any tips to ensure our data is good as possible? Or any suggest papers that have followed a good approach?

Robert Dawes (robert.dawes@bbc.co.uk)

2024-05-20 15:39:23

And in answer to the performance issues, we previously used Inception v3 because it was very quick and cheap - and didn't require much compute. This time round we have scope to be a bit more flexible.

Zhongqi Miao (zhongqi.miao@berkeley.edu)

2024-05-20 15:48:46

It really depends on how you define "better". I personally is still using resnet18 for most of my classification projects because i don't need the extra 2% to 5% improvement from the architectures. And the majority issues are from data itself like imbalance and noise, which can't be solved by model architectures. Ultimately, we will rely on human in the loop, so as long as we have a good calibration pipeline, it is applicable. So what is in your mind that can make your current pipeline better? That Would be a good starting point.

👍 Nicolas Arrieta Larraza

Piotr Tynecki (piotr@tynecki.pl)

2024-05-20 15:50:25

@Robert Dawes One good tip is to collect, store and process the data in Camtrap DP standard (paper), especially if you have data from many sources from many research groups.

Robert Dawes (robert.dawes@bbc.co.uk)

2024-05-21 05:40:20

@Zhongqi Miao - our current pipeline has classifiers for a couple of domains which we've been using for a few years. We've now got some work that would involve expanding to look at other domains/locations and could also bring in some new training sets. So we felt that if we're going to do some more training we ought to try and at least think about what model we're training.

What "better" means could actually be open for debate. It might be just better classification, but it could also mean being quicker to retrain with new data or being as cheap as possible.

Robert Dawes (robert.dawes@bbc.co.uk)

2024-05-21 05:42:42

At the very least we should compare something like resnet to our current model to see what improvements or issues there are

Zhongqi Miao (zhongqi.miao@berkeley.edu)

2024-05-21 12:50:07

@Robert Dawes I see! I think in your case, it is more of a domain generalization task. First of all, in terms of a place to check new models, https://paperswithcode.com/sota/image-classification-on-imagenet, this leaderboard is always a good place to start, and you will see almost all models on the leaderboard are transformer models with huge parameters and huge pretraining or some contrastive pretrainings. Transformers are also notoriously hard to fine-tune. So I don't know whether these model will be useful for your project. However, these models, because they are trained on large data, are usually good at domain generalization on general domain data. But for camera traps, I have doubts, because the domains are so different from general images. You can still see resnet-152 on the leaderboard though.

paperswithcode.com

Papers with Code - ImageNet Benchmark (Image Classification)

The current state-of-the-art on ImageNet is OmniVec(ViT). See a full comparison of 980 papers with code.

Original URL: https://paperswithcode.com/sota/image-classification-on-imagenet

Zhongqi Miao (zhongqi.miao@berkeley.edu)

2024-05-21 12:51:25

On the other hand, since it is a domain generalization tasks, I think searching for domain generalization targeted methods might be a more effective way. Even collecting enough training data from different domain can be an option I think

Robert Dawes (robert.dawes@bbc.co.uk)

2024-05-21 13:04:37

I’ve wondered in the past how good a benchmark ImageNet is for wildlife classification given that most objects in there don’t have the characteristics of wildlife - deformation etc

Robert Dawes (robert.dawes@bbc.co.uk)

2024-05-21 13:07:01

I - like most people - have fine tuned networks trained on ImageNet. But I wonder if that has some flaws as an approach.

Robert Dawes (robert.dawes@bbc.co.uk)

2024-05-21 13:07:20

Thanks very much for the notes above

Zhongqi Miao (zhongqi.miao@berkeley.edu)

2024-05-21 13:12:38

Exactly. Therefore those leaderboard is only a place for people to know what's the state of the art classification models are. And usually, as long as finetuned well, better models on the leaderboader can yield better results on animals. But the tricky part is "finetuned well". And like you said, imagenet pretrained models And wildlife imagery have domain discrepancies. Therefore out of the box, no imagenet models work on camera trap images. However, all the low level visual features are the same, they are all edges, textures, lines, And shapes. Imagenet is good because it covers most of these low level features. And finetuning is basically a way to transfer the combinations of these lowlevel feature to other domains like wildlife imagery. And this is also why tramlnsformers trained on huge pretraining data work even better because these data cover more lowlevel features and possible combinations

👍 Bernie Boscoe

Antonio Ferraz (antonio.a.ferraz@jpl.nasa.gov)

2024-05-21 20:12:04

Hi everyone, who is attending the World Biodiversity Forum? Lacey Hughey (Smithsonian), Talia Speaker (WildLabs and WWF) and myself (NASA JPL), we are thinking in organizing side events (an informal meeting and social in a bar), around the topic of animal movement science and applications. Let’s us know if you are interested or know someone that might be interested in joinning us.

❤️ Sara Beery, Caleb Robinson, Robin Sandfort, Kalindi Fonda, Talia Speaker

Kalindi Fonda (kalindi.fonda@gmail.com)

2024-05-22 06:23:43

Oh wow this is interesting.

Would you recommend going to the event as a "private person"?

Robin Sandfort (sandfort@wildbiologie.org)

2024-05-22 07:33:13

Which of the evenings are you looking at? I might be coming.
Greetings from Austria,
Robin

❤️ Talia Speaker

Talia Speaker (talia.speaker@wildlabs.net)

2024-05-22 12:27:21

Yay Robin! We're thinking a social following the Apero event on the Monday evening and a meeting Wednesday around/after lunch, but still finalizing

✅ Robin Sandfort

Sam Lapp (sam.lapp@pitt.edu)

2024-05-24 18:35:47

Hi all, I’m looking for opinions on PyTorch vs Pytorch+Lightning

specifically, I’m debating whether the OpenSoundscape python library should move from internally using raw Pytorch to using Pytorch Lightning. My thinking is that many implementation details (such as correctly and optimally scaling training/inference across devices, implementing mixed precision or batch gradient accumulation, integration with many different logging GUIs, etc) have been implemented by the PyTorch experts so re-implementing may be worse than switching to Lightning. I’m curious to hear opinions/thoughts/votes or specific details to take into consideration.

⚡ Suzanne Stathatos

Suzanne Stathatos (suzanne.stathatos@gmail.com)

2024-05-24 18:45:36

Lightning’s modularization is honestly really nice. I know other frameworks (i.e. lightly) include tutorials and documentation in both pytorch and pytorch+lightning, which is definitely more work from a documentation perspective, but is also really nice from a user perspective to not be forced to use one or the other.

I know @Markus Marks has thoughts on this too

👍 Sam Lapp

Josafat-Mattias Burmeister (josafat-mattias.burmeister@web.de)

2024-05-25 09:24:25

I have worked with both plain PyTorch and PyTorch Lightning in different projects. PyTorch Lightning offers many great features such as automatic device handling, support for various metrics logging platforms, scaling of training across multiple devices, early stopping, checkpointing, etc. In my experience, this speeds up development, especially at the start of a project. It also helps to increase the maintainability of the code, as you have to write, test and maintain less of the code yourself. However, the learning curve for PyTorch Lightning is also steeper. For people who have little developer experience with deep learning, I would therefore recommend gaining experience with plain PyTorch first. For me, this helped to understand what happens at which point in a training or inference pipeline. Another disadvantage of PyTorch Lightning is that you have less control over the training and inference pipeline, at least if you are working with the standard Lightning Module. For projects with many non-standard steps in training or inference, plain PyTorch is the better choice in my opinion.

👍 Sam Lapp, Valentin Gabeff

Ben Weinstein (benweinstein2010@gmail.com)

2024-05-28 12:14:59

We use lightning for DeepForest and its really nice to package together a single class, it allows us to store a config file

m = main.deepforest()

and then be able to do things like

m.create_trainer()

the only awkward piece is sometimes a pytorch lightning trainer wants the module specified explicitly, so it can look alittle silly.

m.trainer.fit(model=m)

but otherwise it really saves the user from needing to know things and would allow you to swap out models and architecture in the future without the user needing to care about it. If your goal is to make user experience easier, I would use lightning, if your goal is to give maximum flexibility for just your team, plain pytorch. Here is the example of our lightning subclass https://github.com/weecology/DeepForest/blob/7de55c1a941ab53dc06158cb84f8fd3ca5ca5b0a/deepforest/main.py#L22

👍 Sam Lapp

Sam Lapp (sam.lapp@pitt.edu)

2024-05-30 10:24:23

thanks for the exmaples!

Sam Lapp (sam.lapp@pitt.edu)

2024-05-30 10:37:51

@Ben Weinstein this example reminds me of something I’ve wondered with PyTorch Lightning - the API requires a separate trainer object from the LightningModule object, which led me to add self.trainer = Trainer(… to the subclass of LightningModule as you have in this codebase. But this leads to the strange syntax of self.trainer.fit(self, … . seems like that has worked out OK for you? I haven’t been able to figure out if that’s what other PL users are doing or if there are issues with it.

Ben Weinstein (benweinstein2010@gmail.com)

2024-05-30 12:22:14

I think being clear in docs that this is the desired format is a reasonable trade for being able to have a single config file that goes in and the user gets to call

self.create_trainer()

or some kind of wrapper and not need to specify the many many arguments.

Ben Weinstein (benweinstein2010@gmail.com)

2024-05-30 12:22:29

Of all the complaints we get, that has never been one of them.

👍 Sam Lapp

Ashley Kim (hugaskim@gmail.com)

2024-05-25 12:38:28

Hi everyone, my name is Ashley! I am a rising junior at UC Berkeley interested in tackling challenges within the intersections of biodiversity and AI. I was inspired towards this field by Google’s Wildlife Insights initiative and was so glad to have found the talk with Sara Beery on the tinyML YouTube channel as a result. Joining the Slack, I am so excited to see a wide variety of interesting topics sparking conversation but also feel a bit overwhelmed. As a beginner hoping to build a robust understanding, what subjects should I look at first?

👋 Sara Beery, Omiros Pantazis, Shir Bar, Mitchell Rogers, Jason Holmberg (Wild Me), Dan Morris, Ishan Nangia, Andy Viet Huynh

Filippo Varini (fppvrn@gmail.com)

2024-05-26 18:34:12

Hi all, I am trying to convince an audience of conservationists that is new and skeptical about AI that it can drastically enhance their efforts.
Can anyone suggest me a paper or graph showing how models like the Megadetector boosted conservation efforts ?

Dan Morris (agentmorris@gmail.com)

2024-05-26 20:34:56

This paper is not related to MegaDetector, but it's the most comprehensive assessment I've seen of the benefits of AI to a camera trap project (it evaluates the use of the eVorta system wrt time, money, and carbon):

Smith J, Wycherley A, Mulvaney J, Lennane N, Reynolds E, Monks CA, Evans T, Mooney T, Fancourt B. Man versus machine: cost and carbon emission savings of 4G-connected Artificial Intelligence technology for classifying species in camera trap images. Preprint. 2024.

A couple of papers related to MegaDetector have estimated results as a speedup fraction to processing time... YMMV, they are all asking very different questions:

Fennell M, Beirne C, Burton AC. Use of object detection in camera trap image identification: Assessing a method to rapidly and accurately classify human and animal detections for research and application in recreation ecology. Global Ecology and Conservation. 2022 Jun 1;35:e02104.

Henrich M, Burgueño M, Hoyer J, Haucke T, Steinhage V, Kühl HS, Heurich M. A semi-automated camera trap distance sampling approach for population density estimation. Remote Sensing in Ecology and Conservation. 2023.

Mitterwallner V, Peters A, Edelhoff H, Mathes G, Nguyen H, Peters W, Heurich M, Steinbauer MJ. Automated visitor and wildlife monitoring with camera traps and machine learning. Remote Sensing in Ecology and Conservation. 2023.

[The last one doesn't explicitly talk about time, but because it's a rare scenario where full automation is plausible (because it doesn't involve species classification), time is implicit.]

But the way I usually look at this is to set aside all the papers that are about the use of AI for camera traps... any paper about AI for conservation (including papers written by me, or you, or most folks on this Slack) is biased toward a positive result, even if only through the review process. After all, we're here because we're excited about the potential of AI in this space!

Instead the case I would make is just "here's a bunch of people using AI who have no reason to if it didn't save them time". For that argument, I can only give you the MegaDetector-centric view, but FWIW, this is the reason I keep a list of MegaDetector users who agree to be included (or publish their use of MegaDetector):

https://github.com/agentmorris/MegaDetector/blob/main/README.md#who-is-using-megadetector

I want a new potential user to be able to look at that list, ping someone they at least sort of know, and ask about their experiences. Almost everyone on that list just had a job to get done, and had no inherent motivation to use AI.

Along the same lines, on the list I maintain re: papers about AI and camera traps:

https://agentmorris.github.io/camera-trap-ml-survey/#papers-with-summaries

...I have a tag called "ecology paper". That's not related to the journal it was published in, rather that tag is meant to capture "this person just had a job to do related to ecology, and wasn't inherently interested in AI, they just used it as a tool to do their work". (I realize I have not made it very easy to search for that tag... note to self.) Those are IMO better arguments in favor of AI than quantitative claims made in a paper about the use of AI: if those authors didn't find that AI was useful, they just wouldn't have used it.

💚 Burak Ekim, Justin Kay

👍 Ed Miller

Justin Kay (justinkay92@gmail.com)

2024-05-27 03:05:03

Avoiding the term “AI” and using something more specific like “automated image processing” could also help. “AI” comes with a lot of baggage these days 😅

👍 Lukas Picek, Ed Miller, Dan Morris, Carly Batist, Mitch Fennell, michele volpi

💯 Jon Van Oast

Filippo Varini (fppvrn@gmail.com)

2024-05-27 12:19:26

Thank you!!

Romain Lefèvre (romain.adrien.lefevre@protonmail.com)

2024-05-27 11:28:50

👋 Hello everyone!

I'm Romain, a scientific researcher in the Behavioural Ecology Group at the University of Copenhagen, led by Elodie Briefer (https://www.behavioural-ecology-group.com/). My passion lies in leveraging machine learning to decode animals' emotions, and I'm thrilled to meet you and join this Slack channel! 😊

Beyond my research, I also specialize in web design, aiming to enhance online presence and professionalism. Recently, I had the pleasure of designing the AI & Bioacoustics 101 online workshop website, and I could not be happier to introduce it to you.

Understanding and protecting wildlife has always been a challenge. Traditional methods of monitoring animal behavior and communication can be time-consuming and invasive. This is where AI comes in—revolutionizing how we decode and interpret animal sounds, leading to significant advancements in conservation and welfare.

But how can we enhance animal welfare and conservation using artificial intelligence?

Join us at the Bioacoustics & AI 101 workshop on September 25th - 26th, where we’ll explore how AI can transform bioacoustics, making it more efficient and accessible. Whether you're a researcher, biologist, or simply curious about the intersection of technology and nature, this workshop offers invaluable insights.

Workshop Highlights:

Demystifying AI: Break down complex AI concepts into simple, accessible language.

Hands-on learning: Train your own deep learning network to classify sounds—no coding required!

Cutting-edge research: Explore the latest advancements in AI-powered bioacoustics.

Featured Speakers:

Morten Goodwin: Professor of AI, University of Agder

Marie Roch: Head of the MAR Lab, San Diego State University

Anna Zamansky: Head of the Tech 4 Animals Lab, University of Haifa

Yossi Yovel: Head of the Bat Lab, Tel Aviv University

Léo Papet: Co-founder of Biophonia

Colleen Reichmuth: Head of the Pinniped Lab, University of California Santa Cruz

Key Dates:

Abstract submission deadline: July 19th, 2024

Acceptance notification: August 16th, 2024

Registration closes: September 13th, 2024

➡️ Don't miss this exciting opportunity to learn and connect with experts in the field. Register now for free and embark on an insightful journey into the future of AI in bioacoustics: https://aibioacoustics101.com/

Looking forward to virtually seeing you there! 🙂

Best,

Romain

Bioacoustics & AI 101

Bioacoustics & AI 101 | Home

Join the Bioacoustics & AI 101 online workshop, Sept 25-26, 2024. Explore cutting-edge AI applications in sound analysis, and learn to demystify AI.

Written by

Romain Lefèvre

Time to read

2 minutes

Original URL: https://aibioacoustics101.com/

😎 Jon Van Oast, Chris Lange

Juan Sebastián Cañas (jscanass@gmail.com)

2024-05-27 12:55:08

Hi everyone! Today we opened an unlabeled dataset of PAM, check here for more information 🐸 https://x.com/jscanass/status/1795129045815812382 https://zenodo.org/records/11244814

Zenodo

Unlabeled AnuraSet: A dataset for leveraging unlabeled data in machine learning models for passive acoustic monitoring

The Unlabeled AnuraSet (U-AnuraSet) is an extension of the original AnuraSet dataset. It consists of soundscape recordings from passive acoustic monitoring conducted in Brazil. The recording sites are identical to those in the original AnuraSet. Each site comprises 2,666 one-minute raw audio files of unlabeled data. The U-AnuraSet is publicly available to encourage machine learning researchers to explore innovative methods for leveraging unlabeled data in the training of models aimed at solving problems such as anuran call identification. If you find the Unlabeled AnuraSet useful for your research, please consider citing it as follows: Cañas, J.S., Toro-Gómez, M.P., Sugai, L.S.M., et al. A dataset for benchmarking Neotropical anuran calls identification in passive acoustic monitoring. Sci Data 10, 771 (2023). <https://doi.org/10.1038/s41597-023-02666-2>

Original URL: https://zenodo.org/records/11244814

🎉 Dan Morris, Maddie Cusimano, Marius Miron, Andy Viet Huynh

😎 Jon Van Oast, Chris Lange, Andy Viet Huynh

👏 Yuval Mendelson, Maddie Cusimano, Andy Viet Huynh

Sara Beery (sara.beery@gmail.com)

2024-05-29 17:04:43

Come join us for an AI for Conservation meetup in Seattle this June!! Co-located with CVPR 2024 and co-hosted by my students @Justin Kay
and @Timm Haucke, we welcome anyone interested in the intersection of AI and the environment, ecology, biodiversity, etc!!

🌍 Oisin Mac Aodha, Gabriel Tseng, Subhransu Maji, Nino Migineishvili, Dan Morris, Omiros Pantazis, Ted Schmitt, Justin Kay, Risa Shinoda, Lukas Picek, Dante Wasmuht, Thijs van der Plas, Chris Lange, Hugo Magaldi, Nico Lang, Julia Chae, Vincent Lostanlen, Gustavo Perez, Shir Bar, Lasha Otarashvili, Madeleine Grunde-McLaughlin, Neha Hulkund, Benjamin Algreen Adler, Andy Viet Huynh, Anastasios Angelopoulos

🎉 Jon Van Oast, Joe Nangle, gvanhorn, Carly Batist, Mitchell Rogers, Omiros Pantazis, Michael Procko, Justin Kay, Risa Shinoda, Bernie Boscoe, Cara Appel, Lukas Picek, Malte Pedersen, Thijs van der Plas, Valentin Gabeff, Julia Chae, Vincent Lostanlen, Gustavo Perez, Shir Bar, Talia Speaker, Lasha Otarashvili, Jason Holmberg (Wild Me), Madeleine Grunde-McLaughlin, Neha Hulkund, Tarun, Christoph Praschl, Andy Viet Huynh, Rowan Converse, Anastasios Angelopoulos

👍 Luke Sheneman, Andy Viet Huynh, Anastasios Angelopoulos

🙌 Neha Hulkund, Andy Viet Huynh, Anastasios Angelopoulos, Malika Nisal Ratnayake

🙂 Erika Barthelmess (she/hers)

Dan Morris (agentmorris@gmail.com)

2024-05-29 18:07:17

Being in Seattle after 7pm is getting close to my bedtime, but I'm so excited to see all of the AI4C folks that I'll make an exception to my "90-year old farmer" schedule.

😅 Lukas Picek, Sara Beery, Justin Kay, Mitch Fennell, Lauren Harrell, Timm Haucke, Neha Hulkund

Thijs van der Plas (vdplasthijs@gmail.com)

2024-05-30 05:58:48

Fantastic, thank you for organising! 🙂

❤️ Sara Beery

Hugo Magaldi (magaldi.hugo@gmail.com)

2024-05-30 09:26:19

Would have loved to join you and meet the community but paris-seattle makes for a long commute !

❤️ Sara Beery

Masato Hagiwara (hagisan@gmail.com)

2024-05-30 13:23:20

Hi @Sara Beery as someone working on animal communication and based in Seattle, this sounds great! I’d love to join, but can I just show up or would have to register?

👍 Mitchell Rogers, Sara Beery

Sara Beery (sara.beery@gmail.com)

2024-05-31 09:02:33

Just show up! It's very casual

Lauren Harrell (laurenaharrell@gmail.com)

2024-05-31 09:03:04

I’m sorry I can’t be there (will be in Zurich)! But this sounds awesome!

Masato Hagiwara (hagisan@gmail.com)

2024-05-31 12:15:01

@Sara Beery Great, looking forward to seeing everyone!

Eric Cunningham (ejcunningham@gmail.com)

2024-05-31 16:04:14

Amazing, for organizing this! I’ll plan to join + looking forward to meeting you all 🙂

Matt Weldy (matthewjweldy@gmail.com)

2024-06-07 11:53:47

Sounds like fun. I'll pop across the ferry.

Marius Miron (marius.miron@earthspecies.org)

2024-06-05 05:47:21

Dear AI for Conservation Community,

The submission deadline for VIHAR-2024 has been extended to June 30th, 2024.

VIHAR-2024 is the fourth international workshop on Vocal Interactivity in-and-between Humans, Animals and Robots, a satellite event of Interspeech 2024. It will take place in a hybrid format and will be hosted in Kos, Greece on 6th September 2024 and online on 9th September 2024. VIHAR-2024 aims to bring together researchers studying vocalization and speech-based interaction in-and-between humans, animals and robots from a variety of different fields. VIHAR-2024 will provide an opportunity to share and discuss theoretical insights, best practices, tools and methodologies, and to identify common principles underpinning vocal behavior in a multi-disciplinary environment.
We invite original submissions of 5-page papers (with the 5th page reserved exclusively for acknowledgements and references) or 2-page extended abstracts in all areas of vocal interactivity. Accepted papers will be compiled in the VIHAR-2024 proceedings that will be published online. All papers should follow the Interspeech 2024 template.
Suggested workshop topics may include, but are not limited to the following areas:

Self-supervised learning for vocal signals

Generative audio systems for vocal interactivity

Function of vocalizations: discovery of information embedded within signals, testing for functional reference, linking vocal signals to behavior

Physiological and morphological comparisons between vocal systems in animals and humans

Vocal imitation and social learning of vocal signals

Valence and emotion in vocal signals

Inter and intraspecies comparative analyses of vocalizations

Interspecific vocal interactivity between non-conspecifics.

Speech perception and production in human-human interactions and human-robot interactions

Comparative analysis of vocal signals in vocal interactivity.

Theory development of vocal interaction interfaces

Ethics of vocal interactivity

Submission instructions can be found at the EasyChair submission page: https://easychair.org/conferences/?conf=vihar2024
Important dates (AoE):

Submission deadline: 30th June 2024

Notification of acceptance: 21st July 2024

Final versions for inclusion in proceedings: 30th July 2024

Author registration closes: 21st August 2024

Workshop: 6th and 9th September 2024

Jarrett Blair (jarrettblair@gmail.com)

2024-06-06 08:00:03

Hi everyone,

The third edition of the Computational Entomology Webinar will be running on June 20th at 1300 UTC. The overall theme of this edition of the webinar is processing liquid-preserved invertebrate specimens. Please see the flyer attached to this message for the speaker list, and you can find out more about the webinar (including registration) on our WILDLABS event page (https://tinyurl.com/CE-WebinarIII)

wildlabs.net

Computational Entomology Webinar III: Processing liquid samples | WILDLABS

Webinar about computer vision and robotic tools to automatically process insect specimens stored in liquid samples

Original URL: https://tinyurl.com/CE-WebinarIII

❤️ Talia Speaker, Sara Beery, Shir Bar, Joseph Dimos, Ishan Nangia

Indu (indup2@illinois.edu)

2024-06-07 10:29:27

As a note, we’re currently looking for participants with high ML expertise (e.g., have taken a course on ML and/or have experience working with a ML system, often use and study ML) and high birding expertise (e.g., have taken a course on birding and/or have experience in birding, often conduct bird-watching and study birding)

Devis Tuia (devis.tuia@epfl.ch)

2024-06-07 09:40:16

Hello everyone, I didn’t see an announcement for this on this slack (sorry if I missed it), so I take the liberty to advertise without the blessing of the organisers 😄 (@Mohamed Elhoseiny, @Sara Beery, etc): at ECCV 2024 in Milan there will be a workshop about Computer Vision for Ecology! I guess many people in this slack will be excited (I am!). Paper deadlines in July / August: https://cv4e.netlify.app

cv4e.netlify.app

CV4E

Workshop on computer vision for ecology at ECCV 2024

Original URL: https://cv4e.netlify.app

🌍 Oisin Mac Aodha, Ishan Nangia, Nico Lang, Vincent Lostanlen, Mohamed Elhoseiny, Yseult Hb, Don Cosseboom, Sara Beery, Shir Bar, Robin Zbinden, Nicolas Arrieta Larraza, Fagner Cunha, Andrew Schulz, Mélisande Teng, Gustavo Perez, Valentin Gabeff, Julia Chae, Juan Sebastián Cañas

🙌 Justin Kay, Dan Morris, Nico Lang, Omiros Pantazis, Vincent Lostanlen, Don Cosseboom, Sara Beery, Robin Zbinden, Cameron Trotter, Gustavo Perez, Julia Chae, Juan Sebastián Cañas

❤️ Sara Si-Moussi, Vincent Lostanlen, Jon Van Oast, Don Cosseboom, Sara Beery, Robin Zbinden, Julia Chae, Juan Sebastián Cañas

🌏 Chris Lange, Julia Chae, Juan Sebastián Cañas

Oisin Mac Aodha (macaodha@caltech.edu)

2024-06-07 09:42:33

Very cool!

Urs (urs.waldmann@uni-konstanz.de)

2024-06-11 06:59:04

Will there be a page limit for submissions? I did not find any information on the website. Maybe I missed something. Thanks!

Devis Tuia (devis.tuia@epfl.ch)

2024-06-11 07:22:12

@Mohamed Elhoseiny?

Benno Simmons (benno.simmons@gmail.com)

2024-06-11 06:10:02

Fully-funded PhD opportunity with me and Tatsuya Amano at University of Exeter! Come and do the first research into responsible AI for biodiversity monitoring, developing ways to ensure these AIs are safe, unbiased and accountable. Generous research and travel costs, plus the opportunity to spend time in Australia at University of Queensland. Deadline June 28 https://www.findaphd.com/phds/project/towards-responsible-ai-systems-for-automated-biodiversity-monitoring-centre-for-ecology-and-conservation-quex-phd-studentship/?p172597
PLEASE REPOST/SHARE WIDELY!

👍 Burooj Ghani, Dylan Van Bramer (she/her), Omiros Pantazis, Remi Gosselin, Dan Morris, Don Cosseboom, Sara Beery, Connor Levenson

Nico Lang (nila@di.ku.dk)

2024-06-12 03:00:13

Hi,
We are hosting the PhD course on “SSL4EO: Self-Supervised Learning for Earth Observation” from July 01-05, 2024 at the University of Copenhagen, Denmark. We would like to invite PhDs and other researchers interested in the topic of self-supervised learning and/or Earth observation to participate and engage with fellow researchers, and experts at SSL4EO.

Event Details:
Title: SSL4EO: Self-Supervised Learning for Earth Observation
Dates: July 01-05, 2024
Location: Øster Voldgade 10, 1350 Copenhagen, Denmark
Credit points: 2.5 ECTS
More details: https://ankitkariryaa.github.io/ssl4eo/

Registration: https://eventsignup.ku.dk/ssl4eo/signup (Early registration is recommended, space is limited.)

Agenda: Our PhD course will cover a range of topics related to self-supervised learning in the context of Earth observation. Our list of invited speakers includes:

Randall Balestriero, META AI
Marc Russwurm, Wageningen University
Konstantin Klemmer, Microsoft Research
Bruno Sánchez-Andrade Nuño, Clay
Jan Dirk Wegner, UZH/ETH Zürich
Xiaoxiang Zhu, Technical University of Munich

Contact Information: For any inquiries or additional information, feel free to reach out to us {ak, stefan.oehmcke, nila}@di.ku.dk.

We look forward to welcoming you to the first iteration of SSL4EO: Self-Supervised Learning for Earth Observation!

Best regards,
Stefan Oehmcke, Nico Lang, and Ankit Kariryaa

ankitkariryaa.github.io

SSl4EO

SSL4EO: Self-Supervised Learning for Earth Observation

Original URL: https://ankitkariryaa.github.io/ssl4eo/

🙌 Thijs van der Plas, Oisin Mac Aodha, Yuru Jia

Paul Allin (allinpaul@gmail.com)

2024-06-12 05:10:27

Hi everyone, very excited about a new research project using ML and airborne small aperture radar to detect snares. Looking for a good MSc student, CVs need to be in by then end of July. Please feel free to share!

😎 Jason Holmberg (Wild Me)

Benjamin Algreen Adler (benjamin@algreenadler.com)

2024-06-13 09:00:00

Many of you may have already seen this, but the Bezos Earth Fund's $100M AI for Climate and Nature grand challenge is now accepting submissions, and biodiversity conservation is a focus area in the first round. Eligibility is limited to academic institutions and U.S. 501c3 nonprofits, but I imagine many of you have ideas and could partner with an organization to submit.

Feel free to shoot me a note if you have questions, or join the webinar on June 20th.

AI for Climate and Nature Grand Challenge

Home - AI for Climate and Nature Grand Challenge

Original URL: https://www.aiforclimateandnature.org/

AI for Climate and Nature Grand Challenge

Potential applicants can learn more about the AI for Climate and Nature Grand Challenge at the June 20 virtual information session. The team will present an overview of the Grand Challenge and answer questions submitted in advance. A video recording and summary of questions and answers will be published following the session. A global $100 […]

Written by

aifnac

Est. reading time

1 minute

Original URL: https://www.aiforclimateandnature.org/virtual-information-session/

🎉 Catherine Breen, mimi

👍 Cathy Atkinson

Sara Beery (sara.beery@gmail.com)

2024-06-16 05:19:55

Come look at birds with us next Thursday morning before CVPR!!

https://x.com/sarameghanbeery/status/1802269672575783202

X (formerly Twitter)

E14 · NE Clark Rd, Seattle, WA 98105

Are you coming to @CVPR? Do you like birds? Do you like walks? @oisinmacaodha @Grant_Van_Horn @juliachae_ @cl_appel and I are hosting the FOURTH #CVBirdWalk this Thursday morning! All are welcome to join - birding experience and/or CVPR registration not necessary 😁

Original URL: https://x.com/sarameghanbeery/status/1802269672575783202

🐦 Nicolas Arrieta Larraza, Justin Kay, Urs, Shir Bar, Mitchell Rogers, Andrew Schulz, Oisin Mac Aodha, Fagner Cunha, Jonathan Roberts, Malte Pedersen, Ben Seleb, Thijs van der Plas, Julia Chae, Brian Geuther, Mike Rowling

❤️ Lukas Picek, Joseph Dimos, Omiros Pantazis, Justin Kay, Ted Schmitt, Julia Chae, Talia Speaker

🎉 Jon Van Oast, Julia Chae

🦉 Angela Zhu

Jes Lefcourt (jeslefcourt@gmail.com)

2024-06-16 06:37:40

For those who are driving, I suggest parking here: https://maps.app.goo.gl/WTjD63LQeL4RMfL29

google.com

★★★★★ · Parking lot

Original URL: https://maps.app.goo.gl/WTjD63LQeL4RMfL29

❤️ Sara Beery, Ted Schmitt, Julia Chae

Dan Morris (agentmorris@gmail.com)

2024-06-18 12:29:37

Make all of us who didn't make it this morning jealous... what awesome birds did you see?

Sara Beery (sara.beery@gmail.com)

2024-06-18 12:33:30

On Thursday!! No birds this morning, alas

Dan Morris (agentmorris@gmail.com)

2024-06-18 14:30:49

Ah, my bad. I mentally collapsed all of CVPR into "Monday" and "the rest of CVPR". I'll ask again for a bird report on Thursday.

👍 Sara Beery, Izzy Zhu

Sara Beery (sara.beery@gmail.com)

2024-06-19 13:21:51

For those joining tomorrow, if you're coming over from the conference center with us we will meet at 7 in front of Arch (7th and Pike) and then take the light rail. Otherwise we will collectively meet at the UW Boathouse at 7:30!

👍:skin_tone_2: Cara Appel

👍 Nino Migineishvili, Julia Chae, Pia Bideau, Brian Geuther, Mathias Günther, Xiaojuan Liu, Andrew Schulz

🎉 Riley Knoedler

👍:skin_tone_3: Jess Tam

Mathias Günther (mathias.guenther@uni-konstanz.de)

2024-06-19 20:22:06

Will you leave at 7 sharp? Or is there a little time to pick up some breakfast to go?

Ted Schmitt (teds@allenai.org)

2024-06-17 13:22:43

I plan to meet you there at 7:30.

🎉 Ben Weinstein, Sara Beery

Julia Chae (chaenayo@mit.edu)

2024-06-18 00:31:15

Introducing the first Computer Vision for Ecology (CV4E) Workshop at #ECCV2024 in Milan 🌳 If your work combines computer vision and ecology, submit a paper and join us! The workshop will cover applications across diverse ecological systems, featuring exciting speakers, discussions, and challenges (TBA). Thank you @Devis Tuia for promoting us earlier this month 🙂

Deadlines: July 15 (Proceedings) / Aug 15 (Non-Archival)
More details: https://cv4e.netlify.app/submit/

Also please follow our twitter page for future updates!
https://x.com/CV4E_ECCV/status/1802921207286862026

X (formerly Twitter)

CV4E Workshop @ ECCV (@CV4E_ECCV) on X

Introducing the first Computer Vision for Ecology (CV4E) Workshop at #ECCV2024 in Milan 🌳 If your work combines computer vision and ecology, submit a paper and join us! Deadlines: July 15 (Proceedings) / Aug 15 (Non-Archival) More details: <https://t.co/L1GtZnb6fq>

Original URL: https://x.com/CV4E_ECCV/status/1802921207286862026

🎉 Justin Kay, Catherine, Timm Haucke, Nina van Tiel, Alba Márquez-Rodríguez, Catherine Villeneuve, Dan Morris, David Russell, Sara Beery, Neha Hulkund, Andrew Schulz, Gustavo Perez, Mohamed Elhoseiny, Edward Amoah Idun

🙌 Catherine, Timm Haucke, Cameron Trotter, Georgia Atkinson, Levi Cai, Nicolas Arrieta Larraza, Catherine Villeneuve, Sara Beery, Neha Hulkund, Gustavo Perez, Mohamed Elhoseiny

Julia Chae (chaenayo@mit.edu)

2024-06-27 15:35:46

To everyone interested in submitting, we have just updated the website with more submission details such as the paper formatting requirements! Please see https://cv4e.netlify.app/submit/

Also, the paper deadlines are still July 15th and Aug 15th for proceedings and non-archival respectively 🙂 Looking forward to everyone's works

CV4E

Submit a paper or abstract | CV4E

Overview The workshop invites papers for archival (proceedings) and non-archival tracks. Additionally, abstracts may be submitted to the non-archival track. The review process will be conducted on the Conference Management Tool (CMT).

Original URL: https://cv4e.netlify.app/submit/

😎 Timm Haucke, Sara Beery, Subhransu Maji, Devis Tuia, Robin Zbinden, Malte Pedersen, Gustavo Perez

Anuj Gore (anujgore23@gmail.com)

2024-06-21 06:59:52

Hi guys, maybe some of you have probably seen this before but I came across this really interesting paper which demonstrates that African elephants address each other with individually specific name-like calls. Here’s the link if anyone is curious: https://fermatslibrary.com/s/african-elephants-address-one-another-with-individually-specific-name-like-calls#email-newsletter

Fermat's Library

Fermat&#39;s Library | African elephants address one another with individually specific name-like calls annotated/explained version.

Fermat&#39;s Library is a platform for illuminating academic papers.

Original URL: https://fermatslibrary.com/s/african-elephants-address-one-another-with-individually-specific-name-like-calls#email-newsletter

🙂 Peter van Lunteren

charlotte (deshchang@gmail.com)

2024-06-24 19:56:49

Hi folks! For anyone attending NACCB, if you’re interested in NLP for conservation, come join our symposium tomorrow at 8am in the Great Hall North. We have a great set of speakers and are excited to have a panel discussion at the end. 👋

👍 Dan Morris, Tiziana Gelmi Candusso, Matt Ziegler

Dan Morris (agentmorris@gmail.com)

2024-06-24 20:47:09

I'll be there!

💯 charlotte

❤️ charlotte

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-06-25 02:07:21

You have so many cool talks, I might have to jump in and out as I have some colleagues with overlapping talks but looking forward to catch as much as I can of your symposium!

❤️ charlotte

💯 charlotte

David Stein (davidstein94@googlemail.com)

2024-06-25 05:14:35

Hello everyone! I'm a PhD student at TU Dresden, Germany, working on bird species classification with data from the audio domain, in particular audiomoths.

I would like to draw your attention to the half-day online workshop "Machine Learning for Biodiversity Monitoring", which we're going to host on August 12th. Apart from audio and birds species, we are happy to have contributions covering other data modalities and different animal families.

I'm attaching a call for participation for further information. If you would like to participate or if you have any questions feel free to DM me or to send an e-mail to David.Stein1@tu-dresden.de.

Best wishes,
David

🤩 Alba Márquez-Rodríguez, Lucy Dimitrova, Ilyass Moummad, Yanting Teng

👀 Valerie, Aude Vuilli

👍 Chris Lange, Vincent Lostanlen, Alexander Merdian-Tarko, Mohamed Elhoseiny

😎 Jon Van Oast

🙂 Gijs M. Gerrits

Jonah Fox (jonahfox@gmail.com)

2024-06-27 07:00:41

Hi - I'm doing a project for a client - it's very loose right now - but I wanted to get a demo of some kind of species identification (birds or flora or similar) using photographs. Can anyone recommend a good place to start - or a half decent open source model ? There's a few on Kaggle, but it's a bit hard to know if they are good/easy to get into.

Amee Assad (aa3628@columbia.edu)

2024-07-04 17:25:21

Hi! Check out fine-grained classification problems
https://paperswithcode.com/task/fine-grained-image-classification
CUB and NABirds datasets are a place to begin your search

paperswithcode.com

Papers with Code - Fine-Grained Image Classification

Original URL: https://paperswithcode.com/task/fine-grained-image-classification

❤️ Jonah Fox

Jonah Fox (jonahfox@gmail.com)

2024-07-05 08:31:39

that's an amazing website ty so much !

🙂 Amee Assad

Patrick Beukema (patrickb@allenai.org)

2024-06-27 11:30:29

Hi all, we are working on a project that requires annotating ecosystems in remote sensing data to create maps and ultimately a segmentation style model to automatically classify pixels. It does not strike us as a simple annotation effort. Has anyone here completed that type of project before — specifically a distributed/many-expert annotation effort at pixel (or polygon) level granularity? And if so could you comment on:

1) what tools/annotation software you found most effective (third party or open-source),
2) if there are any existing guidelines/recommendations/suggestions for how to do this most effectively
3) any papers focused on the actual data/annotation/processing (besides just the modeling work)?

Jonathan Sauder (sauder@mit.edu)

2024-06-27 16:55:40

Hi, we've been building a large semantic segmentation dataset (of coral reefs) with the help of multiple experts. We're using CVAT (https://github.com/cvat-ai/cvat) which we're hosting on one of our own machines. CVAT works great, and allows configuring in a way to have projects with users and assign them sets of images etc. We also use CVAT's 'Segment Anything' support (runs on the GPU on our machine), which helps a lot.

cvat-ai/cvat

Annotate better with CVAT, the industry-leading data engine for machine learning. Used and trusted by teams at any scale, for data of any scale.

Website

<https://cvat.ai>

Stars

11740

🙏 Patrick Beukema

Patrick Beukema (patrickb@allenai.org)

2024-06-27 17:04:08

Thank you!

Ishan Nangia (ishannangia.123@gmail.com)

2024-06-29 14:20:31

We had used labelme for easily annotating degradation indicators in a degraded tropical forest. Super easy to use tool. Straightforward. Allowed for offline usage (one of our annotators had limited access to internet). But doesnt have a lot of the modern features that come with cloud-based or self hosted softwares.

https://github.com/labelmeai/labelme

For working with multiple annotators, we didnt have them labelling the same images. Also, every annotator was considered equal to the other in terms of their ability. However these two docs might be interesting reads:

https://docs.cleanlab.ai/stable/tutorials/multiannotator.html
https://docs.aws.amazon.com/sagemaker/latest/dg/sms-annotation-consolidation.html

cleanlab

Estimate Consensus and Annotator Quality for Data Labeled by Multiple Annotators

This 5-minute quickstart tutorial shows how to use cleanlab for classification data that has been labeled by multiple annotators (where each example has been labeled by at least one annotator, but ...

Original URL: https://docs.cleanlab.ai/stable/tutorials/multiannotator.html

docs.aws.amazon.com

Consolidate Annotations - Amazon SageMaker

An annotation is the result of a single worker's labeling task. Annotation consolidation combines the annotations of two or more workers into a single label for your data objects. A label, which is assigned to each object in the dataset, is a probabilistic estimate of what the true label should be. Each object in the dataset typically has multiple annotations, but only one label or set of labels.

Original URL: https://docs.aws.amazon.com/sagemaker/latest/dg/sms-annotation-consolidation.html

🙏 Patrick Beukema

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2024-06-28 07:16:58

Hi everyone!

I already messaged you some weeks ago concerning the forth edition of the workshop series on "Camera Traps, AI and Ecology", which will be held this coming September on the Hagenberg campus of the University of Applied Sciences Upper Austria (FH Oberösterreich). We are currently accepting submissions and greatly value your contributions!

I wanted to give you a small update: Considering various requests and, understanding that additional time may benefit the full development of potential papers, we have decided to extend the deadline for the paper submission. The new deadline is now set for the 19th of July, 2024.

The website https://camtrap2024.fh-ooe.at/ has been updated with the new deadline. Feel free to utilize this extra time, and we look forward to your active contribution to the workshop! We would appreciate if you could forward this information to anyone, who might be interested in the workshop too! Online registration is available from today on the homepage!

We are also happy to already be able to announce the preliminary program (https://camtrap2024.fh-ooe.at/program/) including keynotes by outstanding fellows such as Cliodhna Quigley (University of Vienna), Robin Sandfort (Capreolus e. U.) and Stefano Mintchev (ETH Zurich), as well as our social event at the Ars Electronica Festival (https://camtrap2024.fh-ooe.at/venue/#socialevent).

In case of any queries or further information, you can reach us at camtraps2024@fh-hagenberg.at.

With best regards,
Christoph

On behalf of the rest of the organisers:
David Schedl (FH Upper Austria)
Paul Bodesheim (University of Jena)
Tilo Burghardt (University of Bristol)

camtrap2024.fh-ooe.at

Workshop on Machine Vision for Earth Observation and Environment Monitoring

Homepage of the Camera Traps, AI, and Ecology Workshop 2024

Original URL: https://camtrap2024.fh-ooe.at/

Camera traps, AI, and Ecology 2024

Program

Programm of the workshop

Original URL: https://camtrap2024.fh-ooe.at/program/

Camera traps, AI, and Ecology 2024

Venue

Detailed information of the venue

Original URL: https://camtrap2024.fh-ooe.at/venue/#socialevent

❤️ Sara Beery, Mélisande Teng, Sowbaranika, Edward Amoah Idun, Sofía Miñano, Otto Brookes, Chris Lange

🙂 Robin Sandfort

👍 Robin Sandfort

Keiller Nogueira (keillernogueira@gmail.com)

2024-07-01 14:42:35

Hi
********************************************************************************************************************************

2nd Workshop on Machine Vision for Earth Observation and Environment Monitoring
in conjunction with the British Machine Vision Conference (BMVC) 2024

Glasgow, Scotland, UK

https://mveo.github.io/

********************************************************************************************************************************

IMPORTANT DATES

Paper submission deadline: Friday, 16 August 2024
Notification of Acceptance: Monday, 9 September 2024
Camera-ready Paper Due: Monday, 16 September 2024
Workshop: Thursday, 28 November 2024

********************************************************************************************************************************

Following the success of the previous edition held in 2023, the Workshop on Machine Vision for Earth Observation and Environment Monitoring (MVEO) is back. MVEO aims to foster collaboration and idea exchange among the Computer Vision, Remote Sensing and Environmental Monitoring communities, both nationally and internationally, promoting interdisciplinary research, encouraging innovative computer vision approaches for automated interpretation of Earth observation and other correlated data, and enhancing knowledge within the vision community for this rapidly evolving and highly impactful area of research.

********************************************************************************************************************************

TOPICS include but are not limited to:

- Methods: Data-centric machine learning; remote sensing data + language processing (such as Large Language Models) models; open-set, open-world, and open long-tailed recognition; multi-resolution, multi-temporal, multi-sensor, multi-modal approaches; generative models (GANs, stable diffusion, etc); self-, weakly, semi-, and unsupervised approaches; human-in-the-loop and active learning; etc.

- Tasks: Classification; object detection; segmentation (universal, semantic, panoptic, and/or instance); data augmentation and improvement; deep fake; domain adaptation and concept drift; super-resolution; explainability and interpretability; multi and hyperspectral, optical and radar image processing; and so on.

- Applications: Disaster relief; urban planning; sustainable and intelligent agriculture; coast, sea, and marine monitoring; pollution monitoring and air/water quality analysis; circular economy; Cultural Heritage documentation and preservation; climate change; sustainable development goals; geoscience; phenological studies; and so on.

********************************************************************************************************************************

ORGANIZERS

Keiller Nogueira, University of Stirling, UK
Jan Boehm, University College London, UK
Ronny Hänsch, German Aerospace Center (DLR), Germany
Chunbo Luo, University of Exeter, UK
Diego Marcos, Junior Professor, Inria, Universite de Montpellier, France
Paolo Russo, Assistant Professor, Sapienza University of Rome, Italy
Fabiana Di Ciaccio, Assistant Professor, University of Florence, Italy
Ahmed Emam, PhD researcher, University of Bonn, Germany

mveo.github.io

Original URL: https://mveo.github.io/

🙌 Aria Ma, Claydson Bezerra, megan perra, Wanjohi Christopher, An Yu, Diego Marcos, Edward Amoah Idun, Robin Zbinden, Nina van Tiel, Quentin Geissmann

👍 Steve Murphy, Jason Holmberg (Wild Me)

❤️ Alfredo Lozada Fuentes, Kalindi Fonda, Chris Lange

👀 Valerie

Patrick Beukema (patrickb@allenai.org)

2024-07-05 12:00:02

Hi all, During CVPR, Sara swung by AI2 and gave a lecture on a new benchmark dataset for natural world imagery called INQUIRE.

Abstract: Natural world images collected by communities of enthusiast volunteers provide a vast and largely uncurated source of data. For instance, iNaturalist has over 180 million images tagged with species labels, already contributing immensely to research such as biodiversity monitoring and having been cited in over 4,000 scientific papers. Yet, these images are also known to contain a wealth of ""secondary data"" captured unintentionally or otherwise included in images and not properly reflected in image labels. Although this data contains crucial insights into interactions, animal social behavior, morphology, habitat, co-occurrence, and many more questions, the costly, time-consuming, or expert-dependent analysis needed to extract such information prevents breakthroughs. Advances in deep learning methods for language and computer vision have the potential to enable the efficient and automated processing techniques needed to unlock the ""hidden treasure"" in such datasets– being able to directly search large image collections for these concepts would enable richer analyses that span beyond species identification.

We posted the talk here: https://www.youtube.com/watch?v=TnOrlfRgmv4. A heartfelt thank you to Sara and her team for their visit. It was truly a pleasure to finally meet many of you in person after a year of virtual interactions.

YouTube

Allen Institute for AI (https://www.youtube.com/@allenai)

AI-enabled scientific discovery in natural world imagery

Original URL: https://www.youtube.com/watch?v=TnOrlfRgmv4

❤️ Sara Beery, Oisin Mac Aodha, Malte Pedersen, Shir Bar, Negar Sadrzadeh, Justin Kay, Ted Schmitt, Brian Geuther, Timm Haucke, Dan Morris, Omiros Pantazis, Jess Tam, charlotte, Mitchell Rogers, Julia Chae, Neha Hulkund, Yuval Mend, Jason Holmberg (Wild Me), Nico Lang, Chris Lange, Gracie Ermi, Michael Bunsen

Julia Chae (chaenayo@mit.edu)

2024-07-09 15:21:03

ECCV 2024 Workshop Extended Deadline + Updated Requirements! We are extending the archival deadline to July 22nd. We are also now accepting Full Length Papers (14 pages) for the archival track, in addition to Short-form Papers (7 pages).

Please see https://cv4e.netlify.app/submit/ for final details

CV4E

Submit a paper or abstract | CV4E

Original URL: https://cv4e.netlify.app/submit/

🙌 Justin Kay, Neha Hulkund, Jason Holmberg (Wild Me), Andy Viet Huynh, Gustavo Perez, Nina van Tiel, Robin Zbinden, Subhransu Maji, Sara Beery, Juan Sebastián Cañas, Thijs van der Plas, Mohamed Elhoseiny

🎉 Jon Van Oast, Andy Viet Huynh, Robin Zbinden

Ed Miller (ed@hypraptive.com)

2024-07-10 22:44:22

Hi All, I spent my 4-week sabbatical from Arm helping researchers in Ecuador accelerate their camera trap analysis using AI. We started with with simple to use application like EcoAssist and are working toward an Ampere-based Biodiversity Server. Read more here: https://www.linkedin.com/pulse/bringing-bearid-ecuador-sabbatical-story-ed-miller-tkh5c/?trackingId=t7Ylo0FcRt6YGRlOJYNQDg%3D%3D

linkedin.com

Bringing BearID to Ecuador (a sabbatical story)

One of the many perks of working at Arm is the sabbatical program. This is described on the Life at Arm page as: After four years of working with us, you can take an extended paid four-week sabbatical to really hit pause or pursue a passion project.

Written by

Ed Miller

Reading time

7 min read

Original URL: https://www.linkedin.com/pulse/bringing-bearid-ecuador-sabbatical-story-ed-miller-tkh5c/?trackingId=t7Ylo0FcRt6YGRlOJYNQDg%3D%3D

😎 Jason Holmberg (Wild Me), Sara Beery

🐻 Jason Holmberg (Wild Me), Jess Tam, Joe Nangle, Chris Lange, Matthias Zuerl, Aakash Gupta, Peter van Lunteren, Alexander Merdian-Tarko

❤️ Avi Sundaresan, Bernie Boscoe, Arthur Caillau, Yseult Hb, Joe Nangle, Christoph Praschl, Jose Ruiz-Munoz, Christine Laney

Ben Weinstein (benweinstein2010@gmail.com)

2024-07-10 23:41:21

@Ed Miller i've worked with a number of lodges in Ecuador, I forwarded your post to them, you may hear from Maquipucuna/Santa Lucia lodge where there are long standing camera trap programs and plenty of andean bears.

👍 Ed Miller

Kara Watts (kdwwatts@gmail.com)

2024-07-12 13:59:05

@Ed Miller that is such an amazing project. You are truly living my dream. Do you have any advice on how to make the kind of connections that led to your involvement in the BearID project? I transitioned from a career in animal behavior research to focus on software development with ML and am looking for more project opportunities in animal science.

Ed Miller (ed@hypraptive.com)

2024-07-15 12:02:45

@Kara Watts I made my initial connections on https://www.wildlabs.net/. You can dig around the various forums there to look for ideas and collaborators. You can also start working on something that interests you and reach out there for help!

wildlabs.net

Open-Canopy: A Country-Scale Benchmark for Canopy Height Estimation at Very High Resolution

Original URL: https://www.wildlabs.net/

🙌 Kara Watts

Kara Watts (kdwwatts@gmail.com)

2024-07-15 12:03:31

@Ed Miller Thanks for the advice!

👍 Ed Miller

Jennifer Turliuk (jenn.turliuk@gmail.com)

2024-07-11 08:50:54

Hi folks! I am involved with running the climate and energy hackathon this fall at MIT. It brings hundreds of students from across the nation together to work on real climate challenges with real companies (last year: Google, Crusoe, Schneider Electric, Foothill Ventures and more). If you're interested in having MIT students work on a problem important to your company / the climate, send me a DM or email (jturliuk@mit.edu). We're looking for potential challenge partners/sponsors for this year's hackathon. Here is last year's website for more info: https://www.mitenergyhack.org/

❤️ Christoph Praschl, Wanjohi Christopher, Sara Beery, Abhi Ravivarma

🙌 Joe Nangle, Wanjohi Christopher, Sara Beery

Felipe Montealegre-Mora (felimomo@berkeley.edu)

2024-07-12 18:59:54

Hey folks! I'm a beginning postdoc at the Data Science for the Environment center at Berkeley, and I've been trying to get the 'lay of the land' in terms of open source packages for camtrap image classification.

Do you know of any open source GUIs for classifying camtrap images?

I'm interested in both computer-vision-powered backends as well as just GUIs that make the manual classification task easier. I've found EcoAssist (thanks @Peter van Lunteren for the cool piece of software!) but not all that much more - are there any other packages that could be helpful in this?

👋 Sara Beery, Bernie Boscoe, Andy Viet Huynh, Timm Haucke, Chris Lange, Enis Berk Çoban

👋:skin_tone_3: Jess Tam

👍 Peter van Lunteren, Maddie Cusimano

Sara Beery (sara.beery@gmail.com)

2024-07-12 19:06:52

@Dan Morris has a pretty nice list somewhere I'm sure 🙂

🙌 Felipe Montealegre-Mora

🎉 Jon Van Oast

Dan Morris (agentmorris@gmail.com)

2024-07-13 09:51:02

As always, I live for making lists, or in this case, even better, a list of links into another list...

Within the list I maintain re: tools related to AI and camera traps:

https://agentmorris.github.io/camera-trap-ml-survey/

...there are a few relevant sections. The first is the list of image review systems that use AI in some way, which in 2024 is almost all of them:

https://agentmorris.github.io/camera-trap-ml-survey/#camera-trap-systems-using-ml

If a tool is open-source, there should be a link to the source in the description of that tool.

There is a second section that lists no-AI-at-all tools, but as I look at this section now, I think "no AI at all" is almost synonymous with "deprecated". I.e., even the tools that have been updated in the last few years are no longer in use as far as I know, at least I haven't heard of anyone using them recently, with the possible exception of Camera Base (which hasn't been updated in a while AFAIK, but I still hear about folks using it):

https://agentmorris.github.io/camera-trap-ml-survey/#manual-labeling-tools-people-use-for-camera-traps

Last but not least, there's a section for tools that aren't specific to camera traps that people sometimes use for reviewing camera trap images; among these, Exif Pro is the most widely used AFAIK and the only one that's open-source:

https://agentmorris.github.io/camera-trap-ml-survey/#non-camera-trap-specific-labeling-tools-that-people-use-for-camera-trap-data

I have no real data to back this up, but my sense is that among desktop tools, Timelapse has the largest "market share" on Windows, and Exif Pro has the largest "market share" on MacOS, although Windows is still far more widely used than MacOS for reviewing camera trap images.

https://agentmorris.github.io/camera-trap-ml-survey/#manual-labeling-tools-people-use-for-camera-traps

💚 Felipe Montealegre-Mora, Jason Holmberg (Wild Me), Tiziana Gelmi Candusso, Matthias Zuerl, Emilio Luz-Ricca

Sara Beery (sara.beery@gmail.com)

2024-07-13 10:25:23

See??? @Dan Morris always coming in clutch with the lists 🤩

🙌 Felipe Montealegre-Mora, Jason Holmberg (Wild Me), Tiziana Gelmi Candusso, Robin Sandfort, Emilio Luz-Ricca

Felipe Montealegre-Mora (felimomo@berkeley.edu)

2024-07-13 12:50:06

thanks so much you both, this was so helpful!!

Louis Moreau (luis.omoreau@gmail.com)

2024-07-16 11:05:18

Hi! I'm not sure if this has been shared here already.
Here is a nice dataset of aerial imagery combined with height and density data for forests
https://arxiv.org/abs/2407.09392v1?utm_source=tldrai

arXiv.org

Estimating canopy height and canopy height change at meter resolution from satellite imagery has numerous applications, such as monitoring forest health, logging activities, wood resources, and carbon stocks. However, many existing forest datasets are based on commercial or closed data sources, restricting the reproducibility and evaluation of new approaches. To address this gap, we introduce Open-Canopy, the first open-access and country-scale benchmark for very high resolution (1.5 m) canopy height estimation. Covering more than 87,000 km$^2$ across France, Open-Canopy combines SPOT satellite imagery with high resolution aerial LiDAR data. We also propose Open-Canopy-$Δ$, the first benchmark for canopy height change detection between two images taken at different years, a particularly challenging task even for recent models. To establish a robust foundation for these benchmarks, we evaluate a comprehensive list of state-of-the-art computer vision models for canopy height estimation. The dataset and associated codes can be accessed at <https://github.com/fajwel/Open-Canopy>.

Original URL: https://arxiv.org/abs/2407.09392v1?utm_source=tldrai

🌴 Chris Lange

Nicholas (nichseemail@gmail.com)

2024-07-16 15:23:05

Thank you for sharing Louis. This is helpful

Sara Olsson (sara@edgeimpulse.com)

2024-07-17 10:41:00

Hi all, I've published a blog post on rapidly labeling camera trap data using ChatGPT for species identification and a simple object detection model to get the bounding boxes. With this setup, I could run through a set of camera trap recordings and automatically obtain 520 labeled images with detected animals, with only some minutes of manual work by controlling the output, and batch relabel samples of the same species but got different spellings, e.g. "Alaska moose" and "Alaskan moose".

While presented in Edge Impulse, this approach can be replicated outside the platform using Python scripts for example. I’d love for you to give it a read and share your thoughts on its potential usefulness in real-world applications.

video: https://www.youtube.com/watch?v=_Ek1MmZIvtE&t=1s
and blog post: https://www.edgeimpulse.com/blog/adaptive-camera-trap-gpt-4o/

YouTube

Edge Impulse (https://www.youtube.com/@EdgeImpulse)

Adaptive Camera Traps: Identifying Unknown Species with GPT-4o

Original URL: https://www.youtube.com/watch?v=_Ek1MmZIvtE&t=1s

Edge Impulse

Improving Camera Traps to Identify Unknown Species with GPT-4o

When deploying AI models for wildlife monitoring to remote cameras in the field, a common challenge is dealing with unexpected animal species that were not accounted for during training. Here's an approach to improve that.

Written by

Sara Olsson

Filed under

edge ai

Original URL: https://www.edgeimpulse.com/blog/adaptive-camera-trap-gpt-4o/

🙌 Alex Bucknall, Aude Vuilli, Enis Berk Çoban, Jose Ruiz-Munoz, Elizabeth Campolongo, Carly Batist, Jason Holmberg (Wild Me), Jenna Kline, Talia Speaker, Dante Wasmuht, Roberta Hunt, Sinan Robillard, Louis Moreau, Alexander Merdian-Tarko

Louis Moreau (luis.omoreau@gmail.com)

2024-07-24 05:26:12

Great work @Sara Olsson!

Kalindi Fonda (kalindi.fonda@gmail.com)

2024-07-17 20:44:12

Hello! Did you also get the message that Slack will start deleting the old content? (edit: I don't think this question is relevant for this Slack space, as it's currently active, but the next question on better availability of old data/conversations is still relevant).

Is there a way for us to protect some of the conversations from here? Is there any option to upgrade, figure out a different way or are we ok for this to be a more on the go conversation platform (the 90 day cutoff is unideal already)? 🌱
Thank you for being a beautiful community.

Jon Van Oast (jon@wildme.org)

2024-07-17 20:53:23

Jon Van Oast (jon@wildme.org)

2024-07-17 20:53:57

it links to this page which does seem to imply there is a paid plan which solves the problem.

Slack Help Center

Usage limits for free workspaces

If you’re using a free workspace, there are specific usage limits to be aware of. Keep reading to learn more about message, file, data retention, and app installation limits. Tip: Upgrading t...

Original URL: https://slack.com/help/articles/115002422943-Usage-limits-for-free-workspaces

Jon Van Oast (jon@wildme.org)

2024-07-17 20:55:32

( or we could move to something open source like matrix. 🙂 )

Kalindi Fonda (kalindi.fonda@gmail.com)

2024-07-17 21:22:37

Actually on second review of my Slack deletion notice emails it might be only for Slack channels that haven't had any activity recently. However, it's still a bit unideal that we can't tap into all the old content of this channel.

Do people encounter the issue where they run into messages they can't see often? What are the usecases where this happens?

Steve Haddock (haddock@mbari.org)

2024-07-18 16:13:02

@Kalindi Fonda Free slack teams already have a limit where you can't access messages over 90 days old. However Slack has been hanging on to those messages behind the scenes, probably to incentivize you paying to unlock them and access your full history. This new policy is that after August 26, they will delete the archived messages that are older than a year, so even if you switched to a paid plan, you couldn't access your older team history.
There was a discussion of changing platforms for this slack, but ... it took place more than 90 days ago..! ;^)
(Hope this explanation was not too mansplainy — their change from keeping 10,000 messages to only 90 days really impacted a lot of science communities, so I've been following the topic.)
https://www.yahoo.com/tech/slack-delete-chats-files-free-105338864.html

Yahoo Tech

Slack to delete chats and files from free workspaces at end of August

Slack users who don't want their chat history and files to be deleted will have to pay for it, the company has announced. The corporate chat platform says it will begin deleting messages and files older than a year from free workspaces from August 26 onwards - unless the users subscribe to a paid plan. In June users began receiving emails from Slack that "free workspace content older than one year will be deleted." Free users will also only have access to the past 90 days of messages and files,

Original URL: https://www.yahoo.com/tech/slack-delete-chats-files-free-105338864.html

Kalindi Fonda (kalindi.fonda@gmail.com)

2024-07-19 04:38:24

Thanks @Steve Haddock, yes I've been part of quite a few communities when the previous switch (to 90 messages) happened, so I remember the impact 😢

Kalindi Fonda (kalindi.fonda@gmail.com)

2024-07-19 04:40:22

So are all slack communities affected? (I got a couple emails, I guess it's for slack channels where I am the owner/admin). There is the option to export the conversations, is this something the admins would be interested?

Kalindi Fonda (kalindi.fonda@gmail.com)

2024-07-17 20:47:07

Also would anyone be up for another online meetup? We did one about about a "year" ago on and it was lovely. Next week?

🎉 Jon Van Oast, Levi Cai, Enis Berk Çoban, Sinan Robillard, Sandra Gómez Gálvez

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2024-07-18 02:17:44

Hi everyone!

This is a friendly reminder that the paper submission deadline for the upcoming International Workshop on Camera Traps, AI, and Ecology is quickly approaching. Submissions are due by July 19, 2024 (Anywhere on Earth).

Workshop Details:

Date: September 05. - 06., 2024

Location: Hagenberg Campus, University of Applied Sciences Upper Austria (FH Oberösterreich) and Online

Keynote Speakers:

Cliodhna Quigley (University of Vienna)

Robin Sandfort (Capreolus)

Stefano Mintchev (ETH Zurich)

Claudia Probst and Georg Schneider (University of Applied Sciences Upper Austria)

Tutorial Sessions by:

Swarovski Optics

SmartMultiCopters

Don't miss this opportunity to join esteemed researchers and industry leaders to explore the latest advancements at the intersection of camera traps, AI, and ecology.

Looking forward to your participation!

Kind regards,

David Schedl, Christoph Praschl, Paul Bodesheim, and Tilo Burghardt (the 2024 Organization Team)

camtrap2024.fh-ooe.at

Taxonomy Mapping for Camera Trap Datasets - LILA BC

Homepage of the Camera Traps, AI, and Ecology Workshop 2024

Original URL: https://camtrap2024.fh-ooe.at

👍 Oisin Mac Aodha, Sara Beery, Peter van Lunteren

Dan Stowell (dan.stowell@naturalis.nl)

2024-07-22 03:32:55

The Bioacoustics "Stack Exchange" (Q&A site) is looking for more moderators! Would you like to help maintain a pleasant Q&A community? (Time commitment: tiny.) https://bioacoustics.meta.stackexchange.com/questions/213/announcing-a-pro-tempore-election-for-2024 (By the way: Stack Exchange is a site that doesn't delete messages after 90 days 😉 )

😓 Sara Beery, Timm Haucke

😁 Vincent Lostanlen

🚀 Leonardo Viotti, Sam Lapp

Sam Lapp (sam.lapp@pitt.edu)

2024-07-22 16:47:44

for those of us unfamiliar with the “moderator” role, is there a description somewhere?

👍 Abhi Ravivarma

Brian Geuther (Brian.Geuther@gmail.com)

2024-07-22 12:16:32

The group I'm in will be hosting their 3rd short course on using machine learning for behavior quantification: https://www.jax.org/education-and-learning/education-calendar/2024/10-October/shor[…]-on-the-application-of-machine-learning-for-automated-quantific
While our typical audience is more geared towards neuroscientists and geneticists interested in learning fundamentals of the adopting machine learning tools for their in-laboratory experiments, there may be some interest in this conservation community. While this years schedule is still being finalized, last years talk titles and speakers are present to provide a gist of what type of content will be presented.

👍 Sara Beery, Daniel Grzenda, Gözde Cilingir, Elizabeth Campolongo

👋 Daniel Grzenda

Peter van Lunteren (peter@addaxdatascience.com)

2024-07-24 07:44:43

🚨 New LILA BC dataset alert!

Thanks to the guys at Desert Lion Conservation, we got a new dataset consisting of 65k images and 200 videos of Namibian fauna. Labels are provided for 46 categories, primarily at the species level. The labels are mapped into the shared taxonomy.

@Dan Morris thanks for your help!

More info: https://lila.science/datasets/desert-lion-conservation-camera-traps/
MD results: https://lila.science/megadetector-results-for-camera-trap-datasets/

LILA BC

All the camera trap datasets use their own category names… have they been mapped to a common taxonomy? Yes! This .csv file defines a mapping from every category in every camera trap dataset on LILA to the iNaturalist taxonomy. The goal is to let users train classifiers that span datasets, e.g. to be able to...

Est. reading time

2 minutes

Original URL: https://lila.science/taxonomy-mapping-for-camera-trap-data-sets/

LILA BC

Desert Lion Conservation Camera Traps - LILA BC

Overview This dataset contains 65,959 images and 199 videos from the Desert Lion Conservation Project in Northern Namibia. Desert Lion Conservation is a small non-profit organisation dedicated to the conservation of desert-adapted lions. Their main focus is to collect important baseline ecological data on the lion population and to study their behaviour, biology, and adaptation...

Written by

lilawp

Est. reading time

2 minutes

Original URL: https://lila.science/datasets/desert-lion-conservation-camera-traps/

LILA BC

MegaDetector results for camera trap datasets - LILA BC

I was planning to run MegaDetector on some of these datasets… any chance you’ve already done this? Yes! We’ve run MegaDetector versions 5a and 5b on every camera trap dataset on LILA, except for Snapshot Serengeti, for which we post results from MDv4, because we’re still trying to sort out this very odd issue. Note...

Est. reading time

4 minutes

Original URL: https://lila.science/megadetector-results-for-camera-trap-datasets/

👍 Timm Haucke, Dan Morris, Dante Wasmuht, Shir Bar, Viktor Domazetoski, Nathan Fox, Robin Zbinden, Valentin Gabeff

🙌 Elizabeth Campolongo, Alexander Merdian-Tarko

😎 Jon Van Oast

👍:skin_tone_2: Cara Appel

👏 Victor Anton

Amrita Gupta (agupta375@gatech.edu)

2024-07-25 19:58:48

Cross-posting here, ESA anyone? https://aiforconservation.slack.com/archives/CM1JPL18R/p1721939084318559

Amrita Gupta (https://aiforconservation.slack.com/team/UMDLQ8P24)

Hello all! Who is going to be at ESA in Long Beach in a couple of weeks? I'd love to catch up with familiar faces and meet new folks as well!

Original URL: https://aiforconservation.slack.com/archives/CM1JPL18R/p1721939084318559

🙋‍♂️ Joe Nangle, Sara Beery, Nathan Fox, Tarun

👋 charlotte, Leonardo Viotti, Sam Lapp, Sara Beery

Sam Lapp (sam.lapp@pitt.edu)

2024-07-27 16:51:46

is there a conservation tech meetup planned?

👀 Elly Knight, Sara Beery, Tessa Rhinehart

Sara Beery (sara.beery@gmail.com)

2024-07-28 18:08:00

There should be!! But I've been slammed and haven't planned anything. How about meeting for beers Monday evening at 7ish somewhere near the convention center?

Sara Beery (sara.beery@gmail.com)

2024-07-28 18:12:17

Ok, lets do it. ESA Conservation Tech Meetup at 7pm Monday August 5th, at Altar Society Brewing

https://maps.app.goo.gl/QwBKQPofRKQJbGLj7

🙌 Sam Lapp, Amrita Gupta, Tessa Rhinehart, charlotte, Nathan Fox

😎 Jon Van Oast, Tessa Rhinehart, charlotte

:dad_parrot: charlotte

🍻 Alan Stenhouse

Sara Beery (sara.beery@gmail.com)

2024-07-28 18:12:38

Help spread the word?

🫡 Amrita Gupta

Sara Beery (sara.beery@gmail.com)

2024-07-28 18:15:29

https://x.com/sarameghanbeery/status/1817685293249740829

X (formerly Twitter)

https://wildlabs.net/event/ai-conservation-meetup-esa-2024

Coming to @ESA_org in Long Beach next week? After the massive success of the AI for Conservation meetup at #CVPR24 we're doing it again :) ESA Conservation Tech Meetup at 7pm Monday August 5th, at Altar Society Brewing! See you there fellow geeks 😜 <https://t.co/uhUtktEeDp>

Original URL: https://x.com/sarameghanbeery/status/1817685293249740829

Sara Beery (sara.beery@gmail.com)

2024-07-28 18:28:04

wildlabs.net

AI for Conservation Meetup at ESA 2024 | WILDLABS

Come meet fellow folks interested in AI/Conservation Tech

Original URL: https://wildlabs.net/event/ai-conservation-meetup-esa-2024

❤️ Talia Speaker

Sam Lapp (sam.lapp@pitt.edu)

2024-08-04 12:07:00

Hope to see many of you tomorrow!

Reposting details:

ESA Conservation Tech Meetup at 7pm Monday August 5th, at Altar Society Brewing

https://maps.app.goo.gl/QwBKQPofRKQJbGLj7

google.com

Altar Society Brewing and Coffee Co. · 230 Pine Ave, Long Beach, CA 90802

★★★★★ · Brewpub

Original URL: https://maps.app.goo.gl/QwBKQPofRKQJbGLj7

👋 Joe Nangle, Matt Weldy, Sara Beery, Tessa Rhinehart, charlotte

Sara Beery (sara.beery@gmail.com)

2024-08-06 10:48:42

It was great to see everyone last night!!!

🙌 Sam Lapp

Nathan Fox (foxnat@umich.edu)

2024-08-06 12:00:38

Unfortunately, due to some nightmare travel disruptions I wasn't able to get here in time for last nights social, if anyone wants to connect over a coffee or beer over the next few days, feel free to message me!

😞 Sara Beery

Alexander Merdian-Tarko (alexander.merdian-tarko@posteo.de)

2024-08-01 08:31:03

Hi everybody 👋

I've been around here on the AI for Conservation Slack for a while and have been following what's going on. It's really cool to see what's happening in the space and the exciting things people are working on. I have the growing wish to become a part of this and contribute. So I feel now could be a good time to introduce myself.

I'm currently a Data Scientist at UNICEF Germany but would like to transition into a data role in conservation. This space is so important and growing but I feel that data jobs are still somewhat scarce or hard to get - especially for career switchers. Maybe someone here can help me out or point me to a specific opportunity/organization/resource. The people from this community I was already in touch with were so friendly and helpful!

Here's what I'm looking for

technical role where I can use my Data Science, Machine Learning and Remote Sensing skills to help nature and people (I'm quite open domain-wise or regarding the specific context)

team with nice people to learn from

full or part-time, contract opportunities for a limited amount of time could also be interesting

hybrid or remote in Western/Northern Europe

Here's what I can offer:

5+ years experience in Data Science and Machine Learning (e. g. Python, R, SQL) and 2+ years experience in GIS and Remote Sensing (e. g. Google Earth Engine)

experience with land cover mapping in KAZA in collaboration with WWF Germany's Space+Science team

experience with movement ecology in collaboration with Max Planck Institute of Animal Behavior (MoveApps) and Okavango Research Insitute

I'm highly motivated to contribute to relevant societal and environmental issues, such as conservation and climate, and curious about learning new tools and diving into novel domains

I'd be glad to hear about potential opportunities and general feedback. Feel free to connect on LinkedIn or check out my personal website.

All the best and greetings from Cologne :flag_de:
Alex

🙌 Peter van Lunteren, Malte Pedersen, Christoph Praschl, Timm Haucke, Sara Beery, Carly Batist, Nanticha Ocharoenchai (Lyn), Amanda Bullington, charlotte

🙌:skin_tone_3: Alan Stenhouse

Sara Beery (sara.beery@gmail.com)

2024-08-01 10:57:31

Interested in a PhD or Masters at the intersection of AI and Ecology? This research area is growing rapidly, and it can be hard to figure out which research groups where are doing what! Come join our infosession to hear from PIs worldwide about their research and goals!!

(and if you're a PI and want to join, ping me!)

🌍 Oisin Mac Aodha, Lukas Picek, Brian Geuther, Dan Morris, Sonny Burniston, Nico Lang, Burooj Ghani, Dylan Van Bramer (she/her), Shir Bar, Meredith Palmer, Alexander Merdian-Tarko, David Russell, Catherine Villeneuve, Omiros Pantazis, Yuerou Tang, Andrew Schulz, Shravan Ambudkar, Robin Zbinden, Ana Maria Quintero

❤️ Jon Van Oast, Filip Dorm, Braden Charles DeMattei, Dylan Van Bramer (she/her), Catherine Villeneuve, Danilo Ortelli, Tessa Rhinehart, Shravan Ambudkar, Vanesa Reyes, Nathan Fox, Asa DeHaan, Taiki Sakai - NOAA Affiliate, Nora Gourmelon, Loyani Loyani, Talia Speaker, Tuan-Anh VU, Ana Maria Quintero, Remi Gosselin, Emilio Luz-Ricca

Brian Geuther (Brian.Geuther@gmail.com)

2024-08-01 11:04:52

Do you mind if we cross-post this elsewhere?

Sara Beery (sara.beery@gmail.com)

2024-08-01 11:06:36

Please please do!!!

✅ Brian Geuther

Sara Beery (sara.beery@gmail.com)

2024-08-01 11:06:41

Spread the word widely

👀 Brittany Aguilar

Autumn Nguyen (ngoc54n@mtholyoke.edu)

2024-08-15 09:14:14

@Sara Beery Do we have a list of PIs, as well as their research groups and labs and institutions, who will be joining the session? I hope to do some research into the work that those PIs and groups do before the meeting, so that I can make the most of our time together during the session!

Sara Beery (sara.beery@gmail.com)

2024-08-15 10:24:40

Good idea if we do this again! I will share the videos on youtube after, so that will be a more permanent resource, but because there has been some flux about which faculty will join live vs record videos for later I won't be able to share a list ahead of time

❤️ Autumn Nguyen

Autumn Nguyen (ngoc54n@mtholyoke.edu)

2024-08-15 10:58:22

got it, thanks Prof Beery!

Kalindi Fonda (kalindi.fonda@gmail.com)

2024-08-19 07:09:22

Oh no I missed this! Has it been recorded?

Sara Beery (sara.beery@gmail.com)

2024-08-20 07:39:25

Yes, recordings should all be up hopefully this week

Sara Beery (sara.beery@gmail.com)

2024-08-02 16:44:27

Submit your abstract to an AGU2024 session titled "Scalable Biodiversity Assessment with Geospatial Foundation Models and Ecosystem Modeling."

We invite biodiversity assessment contributions from learning strategies to computational demands, from pre-training dataset construction to downstream applications, from benchmarks to evaluation metrics.

Join us to discuss the emerging possibilities and limitations of Geospatial Foundation Models for scalable biodiversity assessment.

-----------------
Session Abstract

Satellite-based ecosystem monitoring of forests provides a continuous and consistent assessment of afforestation, reforestation, and land use conversions on a global scale. However, to estimate a spatial distribution of different species and ecosystem biodiversity, hyperlocal ecosystem observations need to be fused with remote sensing imagery. As such, biodiversity assessments require a strong collaboration between domain experts, machine learning researchers, software engineers and local communities to verify and calibrate the increasingly AI-based prediction models. This includes multi-modal geospatial foundation models which offer a method to combine various modalities (eDNA, text, audio, and image) and overcome the training data sparsity required to generalize models across distinct geographic areas. In this session, we invite presentations that discuss the challenges of collecting and fusing relevant datasets, training, and validating AI models to recognize key parameters, and scaling solutions to restore ecosystem biodiversity on regional and global scales.

View Session Details: https://lnkd.in/d8cMtdni
The session viewer and abstract submission system is open here: https://lnkd.in/dS6QMRna.
The AGU24 abstract submission deadline is Wednesday, 7 August at 11:59 PM EDT.

👀 Alan Stenhouse, Sonny Burniston

Eelke (eelke@lumax.ai)

2024-08-04 09:27:17

eerlijk

➕ Nate Harada, Santiago Ruiz Guzman

🙌 Nate Harada

💯 Nate Harada

Sara Beery (sara.beery@gmail.com)

2024-08-06 20:22:52

I have some availability at ESA tomorrow morning if anyone wants to meet and chat! DM me :)

😎 Jason Holmberg (Wild Me)

Alex Rood (alex.rood@wildlabs.net)

2024-08-07 13:22:00

Hi everyone! My name is Alex, I'm on the community team at WILDLABS, the global conservation technology community. It's nice to meet you all!

This week, WILDLABS is having our 9th annual #Tech4Wildlife Photo Challenge, where the conservation tech community shares photos and videos of how they're using technology for wildlife. It's a great way to celebrate the sector, connect with peers, and see what everyone is working on. We'd love to see some submissions from the AI community 😄

To participate, just share photos and videos of your AI conservation work with the hashtag #Tech4Wildlife and tagging us @WILDLABSNET on X, LinkedIn, and/or Instagram. I hope to see some of your submissions!

🙌 Josiah Hester, Ed Miller, Dan Morris, Jason Holmberg (Wild Me), Carly Batist, Shawn Johnson, Aude Vuilli, Alexander Merdian-Tarko, Talia Speaker

😎 Jon Van Oast, Sara Beery, Jason Holmberg (Wild Me)

🐾 Alan Stenhouse

Ben Weinstein (benweinstein2010@gmail.com)

2024-08-08 13:18:41

The Joint Statistic Meeting is here in Portland and @Toryn Schafer and I just had a nice conversation on how to get more integration between the statistics community and AI4Ecology. @Ben Augustine and others jump in. Just starting a thread here in case others want to be involved (@Eric Orenstein). Areas of intersection include 1) multi-observer processes that use both human and automated observers, 2) finding creative solutions to move toward hierarchical models that capture uncertainty, 3) Going beyond cross-validation and boot strapping as a way of assessing prediction confidence intervals. 4) Merging automated detections with existing time-series of human observations. If there are others in the community interested in solving these kinds of challenges, see this funding opportunity. https://new.nsf.gov/funding/opportunities/computational-data-enabled-science-engineering. I think many of us would be happy to support proposals that bring greater quantitative rigor into using the outputs of these models. I'd like to start a quick literature thread here to bring community knowledge together. https://www.biorxiv.org/content/10.1101/2023.02.20.529272v1.abstract, https://www.mdpi.com/2504-446X/8/2/54, https://zslpublications.onlinelibrary.wiley.com/doi/full/10.1002/rse2.356, if others can drop papers in here they have authored, or know of.

NSF - National Science Foundation

Computational and Data-Enabled Science and Engineering in Mathematical and Statistical Sciences (CDS&E-MSS)

Original URL: https://new.nsf.gov/funding/opportunities/computational-data-enabled-science-engineering

MDPI

Detection Probability and Bias in Machine-Learning-Based Unoccupied Aerial System Non-Breeding Waterfowl Surveys

Unoccupied aerial systems (UASs) may provide cheaper, safer, and more accurate and precise alternatives to traditional waterfowl survey techniques while also reducing disturbance to waterfowl. We evaluated availability and perception bias based on machine-learning-based non-breeding waterfowl count estimates derived from aerial imagery collected using a DJI Mavic Pro 2 on Missouri Department of Conservation intensively managed wetland Conservation Areas. UASs imagery was collected using a proprietary software for automated flight path planning in a back-and-forth transect flight pattern at ground sampling distances (GSDs) of 0.38–2.29 cm/pixel (15–90 m in altitude). The waterfowl in the images were labeled by trained labelers and simultaneously analyzed using a modified YOLONAS image object detection algorithm developed to detect waterfowl in aerial images. We used three generalized linear mixed models with Bernoulli distributions to model availability and perception (correct detection and false-positive) detection probabilities. The variation in waterfowl availability was best explained by the interaction of vegetation cover type, sky condition, and GSD, with more complex and taller vegetation cover types reducing availability at lower GSDs. The probability of the algorithm correctly detecting available birds showed no pattern in terms of vegetation cover type, GSD, or sky condition; however, the probability of the algorithm generating incorrect false-positive detections was best explained by vegetation cover types with features similar in size and shape to the birds. We used a modified Horvitz–Thompson estimator to account for availability and perception biases (including false positives), resulting in a corrected count error of 5.59 percent. Our results indicate that vegetation cover type, sky condition, and GSD influence the availability and detection of waterfowl in UAS surveys; however, using well-trained algorithms may produce accurate counts per image under a variety of conditions.

Original URL: https://www.mdpi.com/2504-446X/8/2/54

❤️ Timm Haucke, Esther Rolf, Toryn Schafer, Shir Bar, Sara Beery, Jason Holmberg (Wild Me), Justin Kay, Casey Youngflesh, Elizabeth Campolongo, Subhransu Maji, Josh Hewitt, Gustavo Perez, Emilio Luz-Ricca, Alan Stenhouse

Toryn Schafer (tschafer@tamu.edu)

2024-08-08 13:48:22

It may be outdated, but I believe @Ben Augustine had done a lit search related to thrust 2 for a Powell Center working group 4 years ago

Justin Kitzes (justin.kitzes@pitt.edu)

2024-08-08 17:25:01

Our group is very interested in this area, would be great to chat about it - particularly area 1 and adding variations of marked abundance models with uncertain marks

➕ Sara Beery, Subhransu Maji

👍 Justin Kay, Toryn Schafer

Sara Beery (sara.beery@gmail.com)

2024-08-08 18:41:52

My group is interested as well, particularly @Timm Haucke and @Justin Kay

➕ Justin Kay, Timm Haucke, Subhransu Maji

👍 Toryn Schafer

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2024-08-08 19:32:02

Wild Me is definitely interested in this. I will bring cookies.

👍 Justin Kay

🍪 Joe Nangle, Alan Stenhouse

Casey Youngflesh (caseyyoungflesh@gmail.com)

2024-08-09 11:28:45

This is great. I’d be very interested in chatting more about uncertainty, data fusion, and hierarchical models

❤️ Toryn Schafer

👍 Justin Kay

Dan Sheldon (sheldon@cs.umass.edu)

2024-08-09 11:53:36

Hi folks, we’re also interested in this topic at UMass. We have a couple related papers (1, 2), and would be glad to chat with folks. @Gustavo Perez @Subhransu Maji @gvanhorn

arXiv.org

DISCount: Counting in Large Image Collections with Detector-Based Importance Sampling

Many modern applications use computer vision to detect and count objects in massive image collections. However, when the detection task is very difficult or in the presence of domain shifts, the counts may be inaccurate even with significant investments in training data and model development. We propose DISCount -- a detector-based importance sampling framework for counting in large image collections that integrates an imperfect detector with human-in-the-loop screening to produce unbiased estimates of counts. We propose techniques for solving counting problems over multiple spatial or temporal regions using a small number of screened samples and estimate confidence intervals. This enables end-users to stop screening when estimates are sufficiently accurate, which is often the goal in a scientific study. On the technical side we develop variance reduction techniques based on control variates and prove the (conditional) unbiasedness of the estimators. DISCount leads to a 9-12x reduction in the labeling costs over naive screening for tasks we consider, such as counting birds in radar imagery or estimating damaged buildings in satellite imagery, and also surpasses alternative covariate-based screening approaches in efficiency.

Original URL: https://arxiv.org/abs/2306.03151

arXiv.org

Human-in-the-Loop Visual Re-ID for Population Size Estimation

Computer vision-based re-identification (Re-ID) systems are increasingly being deployed for estimating population size in large image collections. However, the estimated size can be significantly inaccurate when the task is challenging or when deployed on data from new distributions. We propose a human-in-the-loop approach for estimating population size driven by a pairwise similarity derived from an off-the-shelf Re-ID system. Our approach, based on nested importance sampling, selects pairs of images for human vetting driven by the pairwise similarity, and produces asymptotically unbiased population size estimates with associated confidence intervals. We perform experiments on various animal Re-ID datasets and demonstrate that our method outperforms strong baselines and active clustering approaches. In many cases, we are able to reduce the error rates of the estimated size from around 80% using CV alone to less than 20% by vetting a fraction (often less than 0.002%) of the total pairs. The cost of vetting reduces with the increase in accuracy and provides a practical approach for population size estimation within a desired tolerance when deploying Re-ID systems.

Original URL: https://arxiv.org/abs/2312.05287

👍 Justin Kay, Subhransu Maji, Gustavo Perez

🙌 Ben Weinstein, Subhransu Maji, Gustavo Perez

Ben Weinstein (benweinstein2010@gmail.com)

2024-08-09 12:49:47

@Dan Sheldon @Gustavo Perez, we have airborne modeling work for BOEM (https://www.fisheries.noaa.gov/inport/item/67243), i'd be interested in trying a lightweight versus of this for their biodiversity surveys. I will DM you once I read closely.

fisheries.noaa.gov

Gulf of Mexico Marine Assessment Program for Protected Species | InPort

As part of the Gulf of Mexico Marine Assessment Program for Protected Species (GOMMAPPS), the NOAA Southeast Fisheries Science Center (SEFSC) and the U.S. Fi...

Original URL: https://www.fisheries.noaa.gov/inport/item/67243

👍 Dan Sheldon, Gustavo Perez

Elly Knight (ecknight@ualberta.ca)

2024-08-09 16:33:58

Also interested! I spend a lot of time thinking about detection probability on the acoustic side of things, esp. with respect to human & AI integration. Would love to be kept in the loop (although we’re in Canada, so not eligible for NSF)

Nathan Jacobs (jacobsn@wustl.edu)

2024-08-12 10:06:13

My group is interested. We've put out a few of papers that could potentially fit:

BirdSAT: Cross-View Contrastive Masked Autoencoders for Bird Species Classification and Mapping (WACV 2024): https://arxiv.org/pdf/2310.19168

LD-SDM: Language-Driven Hierarchical Species Distribution Modeling (still a preprint): https://arxiv.org/pdf/2312.08334

PSM: Learning Probabilistic Embeddings for Multi-scale Zero-shot Soundscape Mapping (will be published at ACM MM 2024): https://openreview.net/pdf?id=qnW0LQXY5L

Nathan Fox (foxnat@umich.edu)

2024-08-12 11:05:05

Would also be interested in being involved! Very relevant to my current research.

Toryn Schafer (tschafer@tamu.edu)

2024-08-13 13:28:35

I am interested in submitting an invited session proposal for the next JSM focusing on the thrusts 2 and 3. The JSM 2025 theme is "Statistics, Data Science, and AI Encriching Society". We could do a paper session or panel. Here is more information: https://ww2.amstat.org/meetings/jsm/2025/invitedsessions.cfm#invited

ww2.amstat.org

Invited Sessions

2025 Joint Statistical Meetings (JSM) is the largest gathering of statisticians held in North America. Attended by more than 5,000 people, meeting activities include oral presentations, panel sessions, poster presentations, continuing education courses, an exhibit hall (with state-of-the-art statistical products and opportunities), career placement services, society and section business meetings, committee meetings, social activities and networking opportunities.

Original URL: https://ww2.amstat.org/meetings/jsm/2025/invitedsessions.cfm#invited

👀 Nathan Fox, Sara Beery, Emilio Luz-Ricca

Rowan Converse (rowanconverse@unm.edu)

2024-08-14 13:47:03

Would also like to be kept in the loop!

Matt Morrissette (matt.morrissette@wildlifeprotectionsolutions.org)

2024-08-09 13:56:34

👋 Hi everyone!

👋 Elizabeth Campolongo, Nino Migineishvili, Sowbaranika, Sara Beery, Stephanie O'Donnell, Dan Morris, Ekaterina Nepovinnykh, Mitchell Rogers, Shawn Johnson, Don Cosseboom, Millie Chapman, Robin Zbinden, Céline Angonin, Jennifer, Alexander Merdian-Tarko, Talia Speaker

David Rolnick (dsrolnick@gmail.com)

2024-08-12 22:56:43

Climate Change AI is excited to announce the 2024 edition of our Innovation Grants program!

Our 2024 Innovation Grants program will fund year-long projects at the intersection of climate change and machine learning, offering up to USD 150K per project, with a total of up to USD 1.4M available. We are grateful for the support of Quadrature Climate Foundation, Google DeepMind, and Global Methane Hub, and for fiscal support from the Canada Hub of Future Earth.

For the Main Track, example subject areas include, but are not limited to:

Machine learning (ML) to aid mitigation approaches in sectors such as agriculture, buildings and cities, heavy industry, power and energy systems, transportation, and forestry.

ML applied to societal adaptation to climate change, including disaster prediction, management, and relief.

ML for climate and Earth science, ecosystems, and natural systems.

ML for R&D of low-carbon technologies such as electrofuels and carbon capture.

ML approaches in behavioral and social science related to climate change, including climate finance, economics, justice, and policy.

Projects addressing AI governance in the context of climate change or assessing the greenhouse gas emissions impacts of AI.

In addition to the Main Track, this year’s program also features two Special Tracks:

Special Track on Methane: Focusing on methane-related climate change mitigation in the short/medium term.

Special Track on Dataset Gaps: Emphasizing the creation of datasets or simulators, with potential support from a Google DeepMind researcher.

The submission deadline is September 15, 2024. More information, including eligibility criteria, is available at https://www.climatechange.ai/calls/innovation_grants_2024
We will also hold informational webinars on July 30, 2024, at 9am ET/1pm UTC (register) and August 15, 2024, at 12pm ET/4pm UTC (register). Recordings will be available following the live webinars.

💙 Patrick Beukema, Subhransu Maji, Oisin Mac Aodha, Stephanie O'Donnell, Gustavo Perez, David Russell, Evan Eskew

👀 Stephanie O'Donnell, Carly Batist, Sara Beery, Talia Speaker

💚 Alan Stenhouse

Dan Stowell (dan.stowell@naturalis.nl)

2024-08-14 05:58:22

Hi folks. What's a good URL/citation for "COCO camera trap format" please? Web search today gives me a link to a defunct branch of megadetector.
Is it maybe just here?

Sara Beery (sara.beery@gmail.com)

2024-08-14 08:56:02

We introduced it in the Recognition in Terra Incognita paper, in the appendix, so you could cite that I guess?

Sara Beery (sara.beery@gmail.com)

2024-08-14 08:57:07

But @Dan Morris looks like we should update the links on LILA for the format

👍 Dan Stowell

Dan Morris (agentmorris@gmail.com)

2024-08-14 10:09:45

I have a specific section that I link to for this:

https://github.com/agentmorris/MegaDetector/tree/main/megadetector/data_management#coco-camera-traps-format

Sara correctly points out that a few format links on LILA have become stale. Good catch, I will fix that in a few days, and also probably replace with a shortlink.

👍 Dan Stowell, Sara Beery, Elizabeth Campolongo

🎉 Jon Van Oast

Dan Morris (agentmorris@gmail.com)

2024-08-27 20:42:08

I've fixed this error in all the places it occurred on LILA, and you can use the following shortlink now to refer to the CCT format specification:

https://lila.science/coco-camera-traps

Thanks again @Dan Stowell for catching this.

GitHub

MegaDetector/megadetector/data_management/README.md at main · agentmorris/MegaDetector

MegaDetector is an AI model that helps conservation folks spend less time doing boring things with camera trap images. - agentmorris/MegaDetector

Original URL: https://lila.science/coco-camera-traps

Timm Haucke (timm@haucke.xyz)

2024-08-14 15:10:50

Probably everyone here is painfully aware of Slack's limitation of message history to the past 90 days. Luckily, we've now found a way to make an archive of older messages available here: http://beerylab.csail.mit.edu/AIforConservationArchive/ (props to slack-export-viewer!). This archive should include all messages from public channels only, back until the inception of AI for Conservation. Please let me know if there is anything missing or something not supposed to be in there. Since exporting the messages from Slack, converting and uploading them is a manual process, we can't guarantee updating the archive super regularly. I hope this still helps people find useful links / papers / resources that they have been unable to dig up!

😎 Jon Van Oast, Shir Bar, Malte Pedersen, Gustavo Perez, Justin Kay, Brian Geuther, David Russell, Sara Beery, Chris Lange, Juan Sebastián Cañas, Varshani Brabaharan, Rebecca Wilks

🙌 Carly Batist, Stephanie O'Donnell, Avi Sundaresan, Justin Kay, Nathan Fox, Taiki Sakai - NOAA Affiliate, Sara Beery, Tarun, Viktor Domazetoski, Alessandra Vidal Meza, Aamir Ahmad, Elizabeth Campolongo, Enis Berk Çoban, Kalindi Fonda, Emilio Luz-Ricca, Marion Richardot, Victor Anton, Adrien Fontvielle

🙌:skin_tone_2: Cara Appel

🙌:skin_tone_3: Alan Stenhouse

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2024-08-15 03:23:30

Thanks! btw, Slack recently increased the 90 days to 365.

Timm Haucke (timm@haucke.xyz)

2024-08-15 08:36:34

Interesting, I’m still seeing 90 days, but let’s hope this change is indeed rolled out!

Autumn Nguyen (ngoc54n@mtholyoke.edu)

2024-08-15 09:12:30

thank you so much Timm!

🙌 Timm Haucke

Sara Beery (sara.beery@gmail.com)

2024-08-14 16:07:33

May be a bit late for this, but just saw the following in California:

Sara Beery (sara.beery@gmail.com)

2024-08-14 16:07:43

CALL FOR EXPERTS
AI Tools for Addressing Conservation and Biodiversity

Dear Colleagues,

The California Council on Science and Technology (CCST) is seeking nominations for an Expert Briefing on artificial intelligence tools for addressing conservation and biodiversity challenge—see details below. Please send nominations (including self-nominations) for experts who can give background on these topics and to potentially serve as a briefing panelist or moderator. We also invite suggestions for additional aspects to include in the scope of the briefing.

We will begin reviewing nominations by August 7 COB. Please submit nominations via the online form or email Science Services Manager, John Thompson, PhD, at john.thompson@ccst.us.

Toward a Resilient California
AI Tools for Addressing Conservation and Biodiversity

Biodiversity is declining globally at unprecedented rates due to a variety of cascading and compounding anthropogenic impacts including climate change, habitat destruction, pollution, and overexploitation. To address this crisis, governments around the world are committing to the 30 by 30 initiative with the goal of conserving 30% of the world’s lands and waters by 2030 to protect and restore biodiversity. High quality biodiversity data on the distribution of animals and ecosystems is vital to successfully meeting these goals but is often costly or time consuming to collect. Advances in the fields of artificial intelligence and machine learning are a promising avenue for supporting researchers and practitioners who are building these datasets.

This Expert Briefing will explore examples of novel research projects that are utilizing artificial intelligence tools within the conservation and biodiversity space to augment ongoing efforts to collect robust datasets. The discussion will explore: lessons learned and challenges of incorporating AI into conservation work, opportunities for using these tools to inform management decisions, and pathways for leveraging civic science to support the development of this valuable conservation work.

CCST seeks nominations of individuals (including self-nominations) with relevant expertise who can give background on these topics and potentially serve as a briefing panelist or moderator.

Submit a Nomination

California Council on Science & Technology (CCST)

California Council on Science and Technology (CCST)

CCST is a nonpartisan, nonprofit organization that provides objective advice from California's best scientists and research institutions to the state's policymakers.

Original URL: https://ujr4x4qab.cc.rs6.net/tn.jsp?f=001J16Df5nLtZmn1D4hAsjjcB7UwMSphDS_z9dwhbOYsS0vEXLtBQ_-kA9Z7fJX0J_ekxctb-xmnAIse5v_RHIRE3GdiYo7_T7k1q2-buZuX9bgox3qt5Prpju_XfWiDUVXBbxz2gM9Ea8=&c=hkfyFmGjw4dmF-gBK_9DSndOH0iQfy8lu8v7C1qGXwYJNxCx-PefFA==&ch=dsremsumAM7-EcckEJSzgDOvtLX7TVpZlwWABwjsmtqFUfOhR1Ff8g==

Google Docs

Expert Referral Form

Thank you for your interest in nominating experts related to our call: Climate Impacts and Human Migration. Please use the form below to do so.

Original URL: https://ujr4x4qab.cc.rs6.net/tn.jsp?f=001J16Df5nLtZmn1D4hAsjjcB7UwMSphDS_z9dwhbOYsS0vEXLtBQ_-kA_7adpbiZEUdRnQAzKFg9eaK1aYFEfV0aV-3JKXhTNkRpqN4bO3pRa-wM__16OXqlQQZvNBaopp37hZbmfyn5EnoCSI8aRf9wORT4cPtuOJQ8OongkXL6rW5hM458uak1Rm3VanocnVCMUTYhVP8ttkeeo3ZCPpXLNq6CsA0pBJjoG8kpbGfV7X29mTAHhTUwgdPVmmba-u3YZGyhPs8Yor-I_cSfBVUobkFjuL_y_q&c=hkfyFmGjw4dmF-gBK_9DSndOH0iQfy8lu8v7C1qGXwYJNxCx-PefFA==&ch=dsremsumAM7-EcckEJSzgDOvtLX7TVpZlwWABwjsmtqFUfOhR1Ff8g==

Devis Tuia (devis.tuia@epfl.ch)

2024-08-15 11:26:35

The session in ON! Please join us

Sara Beery (https://aiforconservation.slack.com/team/ULWGNMZCK)

Interested in a PhD or Masters at the intersection of AI and Ecology? This research area is growing rapidly, and it can be hard to figure out which research groups where are doing what! Come join our infosession to hear from PIs worldwide about their research and goals!! (and if you're a PI and want to join, ping me!)

❤️ Oisin Mac Aodha, Ted Schmitt, Thor Veen, Stephanie O'Donnell, Omiros Pantazis, Shir Bar, Michael Bunsen, Subhransu Maji, Shawn Johnson

🎉 Michael Bunsen

🐝 Michael Bunsen

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2024-08-15 13:03:25

Fantastic event!! thanks to Sara and all organizers!

❤️ Sara Beery

Asa DeHaan (adehaan@agci.org)

2024-08-17 04:53:08

Where will I be able to find the recording?

👀 Antón Álvarez, Alan Stenhouse

Sara Beery (sara.beery@gmail.com)

2024-08-23 16:41:26

AI and Ecology Research Lab introductions now LIVE on YouTube!!!

Check it out : https://youtube.com/@AIforConservation/playlists …

And if you're a PI who works in AI+Ecology and want to be included, send me a 3-min intro video and we will add it!!

YouTube

AI for Conservation

Share your videos with friends, family, and the world

Original URL: https://youtube.com/@AIforConservation/playlists

YouTube

AI for Conservation

Share your videos with friends, family, and the world

Original URL: https://www.youtube.com/@AIforConservation/playlists

💚 Millie Chapman, Braden Charles DeMattei, Ana Maria Quintero, Rose Wenxin Zhao, Shir Bar, Jose Ruiz-Munoz, Alessandra Vidal Meza, Elizabeth Campolongo, Shravan Ambudkar, Esther Rolf, Aarshi Jain, Alan Stenhouse, Vanesa Reyes, Risa Shinoda, Gözde Cilingir, Viktor Domazetoski, Robin Zbinden, Violet Turri, Antón Álvarez, Takumi Sato, Sepand Dyanatkar

❤️ Brian Geuther, Aarshi Jain, Robin Zbinden

Alan Stenhouse (alan.stenhouse@csiro.au)

2024-08-23 23:05:36

That was great, thanks to all presenters and to @Sara Beery for organising! Caught everything on the Youtube channel just now.

❤️ Sara Beery

Peter Van Dijck (petervandijck@gmail.com)

2024-08-25 09:33:05

👋 Hi everyone! We are organizing a free, in-person event on AI for Climate Action during Climate Week NYC in September. If you're in town, you can RSVP now, the page is up here: https://events.work.co/aiforclimateaction Shares in channels where people might be interested are appreciated!

🙌:skin_tone_4: Chris Llorca

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2024-08-26 07:13:56

Hey guys!
I just wanted to let you know that we have published the final program for the 4th International Workshop on Camera Traps, AI, and Ecology! 📅
https://camtrap2024.fh-ooe.at/program/

Almost at the same time we also received the hundredth registration🎉 However, registration is still possible so, join us from September 5-6 at the Hagenberg Campus - FH Upper Austria or online for a deep dive into cutting-edge tech and ecological research. 🌍💡

Our hybrid agenda includes:

Four incredible keynotes by top experts such as Robin Sandfort (capreolus e.U.), Cliodhna Quigley (Universität Wien | University of Vienna), Stefano Mintchev (ETH Zürich), Claudia Probst and Georg Roman Schneider (Fachhochschule Oberösterreich)

Three tutorial sessions by Danielle McKenney (SWAROVSKI OPTIK), Andreas Leitner and the Smartmulticopter team, as well as Piotr Tynecki (TRAPPER)

Four paper sessions including ten incredible submissions in the realm of camera trapping, insect monitoring, drone based wildlife monitoring as well as analysis of forests

Social Event and dinner as part of this year's Ars Electronica Festival (only available for registered on site participants)

This is a fantastic opportunity for collaboration and innovation in wildlife monitoring, ecology research, and more. 🦌🐝🤖 Let’s shape the future of conservation together!

Join us and be part of the conversation!

Camera traps, AI, and Ecology 2024

Program

Programm of the workshop

Original URL: https://camtrap2024.fh-ooe.at/program/

📹 Marion Richardot, Timm Haucke, Valentin Gabeff

🙌 Marion Richardot, Timm Haucke, Carly Batist, Piotr Tynecki

Kalindi Fonda (kalindi.fonda@gmail.com)

2024-09-05 05:11:29

This is happening now (online too) but in case anyone is here would love to say hi 👋 🌱

❤️ Christoph Praschl, Antón Álvarez

Dan Morris (agentmorris@gmail.com)

2024-09-09 16:07:27

I didn't manage to call in to this event... folks who were there physically or virtually, share something neat you learned/heard/saw/said/met/detected/classified?

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2024-09-10 05:02:54

Oh no sad to hear that @Dan Morris. I hope due to time issues and not because of technical problems 🥲 Since I had the honor to host the event, I don‘t think I should say my opinion too loud (self-praise and so on 😂 ), but I really think we had some pretty awesome presentations within the workshop 🙂 FYI: I will upload all the recorded presentations the next days, so if somebody missed the workshop, you can rewatch at least the presentations on our homepage http://camtrap2024.fh-ooe.at (unfortunately withe one exception due to a missing agreement on recordings from the company).

camtrap2024.fh-ooe.at

conservationxlabs/miewid-msv2 · Hugging Face

Homepage of the Camera Traps, AI, and Ecology Workshop 2024

Original URL: http://camtrap2024.fh-ooe.at

Dan Morris (agentmorris@gmail.com)

2024-09-10 10:15:06

Clarifying: yes, just time zones and my own inertia, not technical issues. I'll keep an eye out for the recordings, thanks!

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2024-09-11 04:44:26

Still sad but I am calmed down that it was not because of technical issues 😄

Little follow up, I have quickly used the time to upload some impressions of the workshop as well as the presentations and papers:
https://camtrap2024.fh-ooe.at/gallery/
https://camtrap2024.fh-ooe.at/program/

The actual proceedings are currently in the making, but this will still take some time 🙂

Camera traps, AI, and Ecology 2024

Gallery

Gallery of the workshop

Original URL: https://camtrap2024.fh-ooe.at/gallery/

Camera traps, AI, and Ecology 2024

Program

Programm of the workshop

Original URL: https://camtrap2024.fh-ooe.at/program/

Kalindi Fonda (kalindi.fonda@gmail.com)

2024-09-12 06:01:41

I always say the reason I am into tech is the scalability and the leverage it provides, and I usually think about it from the point of view of how by creating tools they become accessible at scale...

There was a point made by @Robin Sandfort that really resonated, he was talking about how AudioMoth being cheap compared with highend devices, means that one can put many more out instead of one single highend one. And sure there is a bit of a loss in the quality of the recording, but one might have never gotten a recording in the first place with only one location being monitored.

Then we continued the conversation and Robin shared a story about how one area where he had installed some AudioMoths has been flooded, and that some devices survived, so he now has the "sound of a flood": the sounds of water rushing in, the frogs being closer and closer, and then the gluglu and then the whole thing in reverse.

The chance that one high end device would break in a flood is a certain amount, but the chance that all the AudioMoths would break in that flood is much lower.

So the point that was elusive before but has now been added to my "I <3 tech because scalability" is redundancy ,
And probably there are many ways to think about it, also in terms of knowledge, or types of projects...

🙂 Robin Sandfort

Kalindi Fonda (kalindi.fonda@gmail.com)

2024-09-12 06:07:53

Plus the organisation was incredible. The whole conference was gentle and caring, the talks were interesting, and covered various topics.

Then at the end of the first day we had a bus for us that took us to town, to an interactive art exhibition (which provided for external stimulus to further conversations), and then a dinner under some trees, where there were lots of chats and smiles.

I'll take a look at my notes I know I wrote down a few quotes of expressions I liked, and maybe even some more on the content. 🌟

Kalindi Fonda (kalindi.fonda@gmail.com)

2024-09-12 06:10:13

As a personal aside, I've been trying to rope my vet technician friend into a more active role in relationship to nature/conservation. She had studied forestry, and is now working at as a vet technician, but I know she has "the call of the forest". I was sending her pictures and insights, and I think something is starting to shift within her 🌱

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2024-09-12 06:22:28

Really glad to hear that @Kalindi Fonda! It was a pleasure having you at the workshop and meeting you 🙂

🥳 Kalindi Fonda, Dan Morris

👍 Robin Sandfort

Peter van Lunteren (peter@addaxdatascience.com)

2024-08-26 10:43:57

Does anyone know what the best way is to individually recognise Orangutans from images? My gut tells me to zoom in on the face, but I can't really find any literature about that.

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2024-08-26 11:14:30

Light Internet search suggests faces (primarily) with some challenges around certain age classes. If there is a dataset of individuals, you might try our multispecies model, which includes faces for some species:

https://huggingface.co/conservationxlabs/miewid-msv2

huggingface.co

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Original URL: https://huggingface.co/conservationxlabs/miewid-msv2

🙌 Peter van Lunteren

Devis Tuia (devis.tuia@epfl.ch)

2024-08-26 16:20:43

I think @Otto Brookes has been working on face recognition for gorillas. Maybe a good person to contact!

✅ Otto Brookes

🙌 Peter van Lunteren

Antón Álvarez (aalvarez@wwf.es)

2024-08-28 08:41:08

I got some "good" result using MegaDescriptor with bear faces, could be nice to compare Miewid and MD.

huggingface.co

BVRA/MegaDescriptor-T-224 · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Original URL: https://huggingface.co/BVRA/MegaDescriptor-T-224

🙌 Peter van Lunteren

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2024-08-28 10:10:59

Is the bear face dataset open?

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2024-08-28 10:12:04

https://huggingface.co/conservationxlabs/miewid-msv2

huggingface.co

conservationxlabs/miewid-msv2 · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Original URL: https://huggingface.co/conservationxlabs/miewid-msv2

Antón Álvarez (aalvarez@wwf.es)

2024-08-28 10:55:47

Nope, it was from the BearID proyect, I have access via the AI for Bears Fruitpunch Challenge, @Ed Miller could say you more about the permision of the dataset.

Otto Brookes (otto.brookes@bristol.ac.uk)

2024-08-29 04:55:41

If there's only a single Orangutan in each image (like the one shown), I would just train a model on the whole image! If you have multiple apes in each image then you'll likely need to localise them (either their face or full body). Happy to discuss further if it's helpful!

🙌 Peter van Lunteren

Peter van Lunteren (peter@addaxdatascience.com)

2024-08-30 02:44:41

Thanks for the info everybody! I'll play around with the models and if needed will get in touch with you @Otto Brookes 😁

👍 Otto Brookes

Serge Wich (sergewich@gmail.com)

2024-09-07 03:50:58

Hi Peter, together with Dan Schofield from Oxford university a group of orangutan researchers is working on this with some large datasets from wild orangutans. It follows the work Dan did on chimpanzees.

🙌 Peter van Lunteren

Serge Wich (sergewich@gmail.com)

2024-09-07 03:51:34

Just get in touch and I can give you more details.

Peter van Lunteren (peter@addaxdatascience.com)

2024-09-08 12:55:11

@Serge Wich , that sounds great! I’ll discuss this with the other project members and reach out to you via email.

Serge Wich (sergewich@gmail.com)

2024-11-29 08:09:09

Just to add. There has been good progress on the face detector. The next step is to do the actual Face ID models. A team of orangutan researchers has put together a large set of data from several sites and species to do this with the Oxford team. I will let you know how it goes. It needs large amounts of images from the same individuals. Another team is doing this for gorillas which is working well. We aim to publish this over the coming months.

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2024-11-29 18:15:39

https://huggingface.co/conservationxlabs/miewid-msv3

Has a good number of face presentations built into general re-ID

huggingface.co

conservationxlabs/miewid-msv3 · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Original URL: https://huggingface.co/conservationxlabs/miewid-msv3

👍 Ed Miller

Vanesa Reyes (vanesa.reyes@wildlabs.net)

2024-08-26 16:38:31

Hey there! At WILDLABS we have launched a bioacoustics horizon scan, and since AI has become a fundamental part of the field, we'd love to hear from you what innovations do you predict will revolutionise bioacoustics in the next two decades? These will be considered for inclusion in the horizon scan prioritisation process, involving diverse experts from across the globe. You can submit your ideas using this Google form.

👍 Jose Ruiz-Munoz, Nanticha Ocharoenchai (Lyn)

🙌 Carly Batist, Robin Sandfort

Rita Pucci (rita.pucci85@gmail.com)

2024-08-27 05:17:19

Hey everyone,
is there someone that published at "https://ietresearch.onlinelibrary.wiley.com/hub/journal/17519640/homepage/call-for-papers/si-2023-000769"?
I Submitted a paper 8 months ago but I didn't receive any feedback. 😞

😢 Burooj Ghani, Robin Zbinden, Avi Sundaresan

Devis Tuia (devis.tuia@epfl.ch)

2024-08-27 05:42:18

we have a coupl eof the editors in this Slack, maybe DM them with your inquiry 😉

Rita Pucci (rita.pucci85@gmail.com)

2024-08-27 05:42:32

Thanks

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2024-08-29 12:10:08

Maybe @Majid Mirmehdi or @Otto Brookes can tell you more 🙂

Majid Mirmehdi (m.mirmehdi@bristol.ac.uk)

2024-08-29 13:42:05

Dear @Rita Pucci,

I sincerely apologise for the delays with your paper. This is highly unusual and our turnaround times are normally well below 3 months.

The Editor handling your article has received 35 “decline”s from nominated reviewers which in itself is highly unusual. There is one review available and we are trying to push another referee whose review we are waiting for. I hope we can be in touch with a decision soon.

Best wishes,
Majid

Rita Pucci (rita.pucci85@gmail.com)

2024-08-29 14:15:45

Ok thank you for your interest

Dan Morris (agentmorris@gmail.com)

2024-08-27 20:11:23

This is a very camera-trap-specific answer, but...

If there are folks in the audience who aren't Python-savvy and may be intimidated by Colab, consider demo'ing EcoAssist, or at least highlighting that you don't need to deal with any Python to use AI.

But I'm not discouraging you from also showing stuff in Colab. FWIW if you haven't come across it, we have a pretty straightforward MegaDetector Colab notebook here.

Addax Data Science

EcoAssist - Simplifying camera trap image analysis with AI

An AI platform to analyse images with machine learning models for automatic detection, so that ecologists can save time and focus on conservation efforts.

Est. reading time

7 minutes

Original URL: https://addaxdatascience.com/ecoassist/

😁 Peter van Lunteren

👍 Judith Dekkers

Atul Ingle (ingle@uwalumni.com)

2024-08-28 02:30:48

gratitude-thank-you that's a great suggestion! I agree that diving right into a colab notebook might turn some folks away; so I'll start by demo'ing EcoAssist and timelapse.

And megadetector_colab.ipynb is indeed the notebook I was planning to demo 🙂

I created a barebones classifier demo using some scripts in your megadetector repo but maybe there's a megadetector_classifier.ipynb demo somewhere that I'm missing?

Dan Morris (agentmorris@gmail.com)

2024-08-28 08:15:09

Training a custom classifier is IMO a little advanced for an introductory session, i.e. it may overstate the degree to which you need to train a custom classifier (which is not typically worth it... IMO in 95% of cases you will save more time overall by using either no classifier or a classifier that's "close enough" than by doing all the work required to train a custom classifier). So instead, consider using the (now very impressive!) model zoo that's built into EcoAssist to demonstrate the growing variety of classifiers that are already out there, and the way you might use the results from a species classifier in Timelapse.

To your actual question, there isn't a notebook for training a classifier within the MegaDetector repo. To each their own, but for my two cents, classifier training in a notebook is a tough putt, it's such a fundamentally asynchronous and infrastructure-dependent experience. So if you want to demonstrate workflows for training classifiers, consider demo'ing one of:

MEWC (exactly what you want, but not easy to demo in a notebook): https://github.com/zaandahl/mewc

Zamba Cloud (for video rather than stills, but still a really slick workflow for no-code training of custom classifiers): https://www.zambacloud.com/

All that said, if you do end up putting together a notebook to demonstrate classifier training, please share here!

There is also this notebook (a couple years old) that fine-tunes MegaDetector to add species categories; the notebook is nice, although this is no longer the approach I would recommend:

https://www.kaggle.com/code/evmans/train-megadetector-tutorial

Also, unrelated to training, if it's a pretty broad session, consider reminding folks that they don't have to do everything on their laptops in 2024, i.e. for folks who might be up for a cloud-based solution for AI, image review, and data management, consider demo'ing, e.g., Wildlife Insights.

Atul Ingle (ingle@uwalumni.com)

2024-08-28 17:16:59

Really appreciate all these suggestions! I wasn't planning to make a colab notebook for training a classifier, just inference with a pretrained model (e.g. your megaclassifier-efficientnet-b3 model file). I guess that's pretty straightforward.

Dan Morris (agentmorris@gmail.com)

2024-08-28 18:10:01

Got it, all clear, ignore my unsolicited advice then. 🙂

There is not exactly a notebook for running MegaClassifier, but... there is a notebook I use for doing All The Things involved in a complete detection+classification batch, and "All The Things" includes running MegaClassifier:

https://github.com/agentmorris/MegaDetector/blob/main/notebooks/manage_local_batch.ipynb

I actually use the .py version, but it's the same as the .ipynb version (the latter is auto-generated from the former):

https://github.com/agentmorris/MegaDetector/blob/main/notebooks/manage_local_batch.py

This notebook doesn't run MegaClassifier directly from Python; rather, it writes a shell script (.bat or .sh, depending on the OS) that does the MegaClassifier stuff. See the cell called "Run MegaClassifier (actually, write out a script that runs MegaClassifier)".

Piotr Tynecki (piotr@tynecki.pl)

2024-08-28 18:27:16

@Atul Ingle feel free to include TrapperAI species model in your edu program after animal recognition. It is easy to load and use in notebook/colab, if the list of supported spieces is covering your context.

👍 Atul Ingle

Peter van Lunteren (peter@addaxdatascience.com)

2024-08-30 02:51:49

@Atul Ingle If you were planning on using https://github.com/zaandahl/mewc to train classifiers, note that there is an option of loading the resulting models into EcoAssist for inference.

https://github.com/PetervanLunteren/EcoAssist/blob/main/markdown/MEWC_integration.md

zaandahl/mewc

Stars

Last updated

8 days ago

<https://github.com/PetervanLunteren/EcoAssist/blob/main/markdown/MEWC_integration.md | MEWC_integration.md>

👍 Atul Ingle

Laura Madrid (lcmmadrid5@gmail.com)

2024-09-03 13:06:54

Hi everyone 💖
My name is Laura and I just recently graduated with a Bachelor's in Computer Science from the University of Toronto. I really like computer vision and I am planning to apply for a masters program related to Computer vision + (sustainability or healthcare or HCI). I have been working on Coursera's AI for good specialization which led me to prof Sara Beery's site and this slack! I want to get involved in this space because this intersection of fields seems really interesting. I would love to chat with any of you for advice and how to get involved. Hope that everyone is having a nice day :)

👋 Timm Haucke, Malte Pedersen, Atul Ingle, Chris Lange, Tor Henrik Ulsted, Ben Weinstein, Gustavo Perez, Dan Morris, Enis Berk Çoban, Sowbaranika, Jose Ruiz-Munoz, Sebastien Ouellet, Jason Holmberg (Wild Me), Andy Viet Huynh, Omiros Pantazis, Izzy Zhu, Sara Si-Moussi, Aakash Gupta, Robin Zbinden, Alexander Merdian-Tarko, Jonathan Roberts, Aarshi Jain, Quentin Geissmann, Elena Sierra, Tim Zhou, Julia Chae, Benjamin Tremoulheac, Elizabeth Campolongo, Angela Zhu

👋:skin_tone_3: Jess Tam, Alan Stenhouse

Benjamin Hoffman (benjaminsshoffman@gmail.com)

2024-09-05 14:57:50

Is anyone aware of a paper that uses the precision & recall (or other metric) of a detection model measured on a test set, in order to inform how error bars are estimated in downstream analyses? I am assuming this would be some hierarchical Bayesian thing, along the lines of what’s done in https://zslpublications.onlinelibrary.wiley.com/doi/pdfdirect/10.1002/rse2.171, equation (4), but modified to include what’s known about model error rates?

Ben Weinstein (benweinstein2010@gmail.com)

2024-09-05 15:02:14

We play in this area in attached. Its far from the kinda of hierarchical model that we want, but starts to gain some insight. This is kind of thing we were discussing @Toryn Schafer , @Ben Augustine, trying to couple non-calibrated scores with downstream confidence intervals. @Eric Orenstein has a paper on plankton that orbits this space as well. TLDR, nothing that i'm aware does precisely what you are looking for. Also summoning both @Casey Youngflesh and @Heather Lynch, since both are close colleagues and are in this channel, and you pointed at their paper.

👍 Casey Youngflesh

Benjamin Hoffman (benjaminsshoffman@gmail.com)

2024-09-05 15:08:47

fantastic, thank you!

Toryn Schafer (tschafer@tamu.edu)

2024-09-05 15:13:12

I've used multiple imputation in the past to propagate a first stage prediction uncertainty to a second stage inferential model: https://doi.org/10.1007/s13253-020-00399-y

Dan Morris (agentmorris@gmail.com)

2024-09-05 15:13:46

This paper is not about a detection model, but otherwise it gets at what you're asking about (the relationship between ML accuracy and downstream metrics):

Whytock RC, Świeżewski J, Zwerts JA, Bara-Słupski T, Koumba Pambo AF, Rogala M, Bahaa-el-din L, Boekee K, Brittain S, Cardoso AW, Henschel P. Robust ecological analysis of camera trap data labelled by a machine learning model. Methods in Ecology and Evolution. 2021 Jun;12(6):1080-92.

Benjamin Hoffman (benjaminsshoffman@gmail.com)

2024-09-05 15:16:57

thanks, these are all really great references!

Casey Youngflesh (caseyyoungflesh@gmail.com)

2024-09-05 16:18:32

I don't have any paper suggestions but here's my (extended) take. I'd probably add another level the traditional mark-recapture framework. You've got your observation model, your process model, and you could add the 'classification model'. I could see adding something to eq. 4, though I think the classification process is different from the observation process. Also, I think this matters for the creation of that vector z. Curious what folks think about the below, where t is time step:

y = ML classified observed state (given as data)
m = latent true observed state (estimated)
z = latent true occupancy state (estimated)
p = probability you observe, given occupancy (estimated)
psi = probability occupancy (estimated)
h = probability that classification is correct (given as data)
If yt = 0, ht = True positives / True positives + False negatives
If yt = 1, ht = True negatives / True negatives + False positives

yt ~ Bern(mt ** h_t)
mt ~ Bern(zt ** p)
z_t ~ Bern(psi)

The one issue here is that you need that vector of z's (latent true occupancy over the the series) which you get from the entire time series of true observed states (filling 1s between first and last observation, assuming a period of closure). That's typically fed to the model before hand (as data). And you don't know the entire time series of true observed states (m's) until your model has run through the time series -- typically your data are your true observed states. Maybe an approach similar to that taken by @Toryn Schafer above could work there for the z vector?

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-09-05 17:55:55

We did something slightly tangential. We used a bayesian characterisation of a detector, but we weren't so much interested in precision/recall but more in the location accuracy. Assume you have a true positive from a multibox detector, most papers only look into minimum jaccard overlap to count it as a true positive (0.5, 0.95, ...) or maybe a weighted distribution to compare models, but when you want to use the localization downstream, you need the localization error distribution of the detector.
We did that in https://doi.org/10.1109/LRA.2018.2850224 - the result is with close approximation a normal distribution - when measured over the evaluation set across different scale ranges:

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-09-05 17:56:30

the false positives fall into the tail edge of the distribution

Eric Price (eric.price@ifr.uni-stuttgart.de)

2024-09-05 17:57:26

that way we can use the projected position with a bayesian tracker and have a good estimate of the "sensor noise" when the DNN is the sensor

Matt Weldy (matthewjweldy@gmail.com)

2024-09-05 18:22:38

@Casey Youngflesh that is roughly the likelihood of a multiscale occupancy model where the availability component is recast.

Different context, but @Tessa Rhinehart did this with bioacoustic predictions [https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.13905] where continuous scores of reviewed files were modeled as a normal 2-part mixture of the logits.

👍 Casey Youngflesh

Ben Weinstein (benweinstein2010@gmail.com)

2024-09-05 19:34:56

I smell a review piece here if any graduate students are lurking. @Casey Youngflesh my problem is thinking about those latent states when you have more than just presence/absence. Ignoring for a moment that confidence scores are non-normal and only look like probabilities (already a fatal flaw), and that not all classes are confused equally (also fatal), the multinomial probability isn't just whether the classification is correct (h in your psuedo code), because it also adds to another class. So for example in our tree work, if we want to count all the trees in a location and the classification model incorrect predicts it to be an oak tree, but its a maple, it both effects the undercount of oaks, which is covered in your model, but also the overcounts of the maples. You can easily image any mcmc process just endlessly trading off when trying to estimate that true latent state (m), no identifiability, especially as you get to more than 2 or 3 species.

👍 Casey Youngflesh

Matt Weldy (matthewjweldy@gmail.com)

2024-09-05 19:37:51

Important to distinguish sigmoid or soft max type heads, where that tradeoff is different

👍 Ben Weinstein

Benjamin Hoffman (benjaminsshoffman@gmail.com)

2024-09-06 00:14:33

thanks for all this, definitely a lot of good papers I’ll have to go read. This type of question has certainly haunted me before, and I would be interested in a review paper if one ever appeared @Ben Weinstein 🙂

Casey Youngflesh (caseyyoungflesh@gmail.com)

2024-09-06 15:44:22

@Ben Weinstein Totally agreed about confidence vs. probability of correctness. But if you did have a proper probability of class in a vector [p sp1, p sp2, p sp3], that should account for the fact that species aren’t confused equally, right? Like, if the model knows they aren’t confused equally that should be baked into the probabilities (I see how it might not know though). I also agree that numerous classes are bound to screw things up. But I’m not sure I follow why you’d necessarily expect to have a non-identifiability problem.

Maybe I don’t understand the problem, but for trees in a given patch, you have a vector of probs for each tree. You draw a value for z (latent true species identity of a given tree i) from this vector of probabilities for tree i (let’s just assume these are actual probabilities, one for each species possibility). Maybe you do this 10k times. So for each tree i, you have 10k realizations of true species ID. Then at each of the 10k iterations, you sum the number of species of each tree. If you’re just interested in counts, you have 10k realizations of counts for each species. This is just being backed out of the (assumed actual) probs, not with a Bayes model.

If you wanted to model the composition of trees as a function of stuff, you could use a Bayes model with a Dirichlet, yeah? With each of the proportions having an error model (you would need a sum to zero constraint though)?

👍 Ben Weinstein

Brian Geuther (Brian.Geuther@gmail.com)

2024-09-06 16:39:05

Don't know if this is particularly helpful (most of this discussion is above my head), but the statistician in my group has used this in the past (conformal prediction for estimating confidence intervals): https://arxiv.org/pdf/2107.07511
At least in my understanding, if you get CIs on the model prediction (rather than probabilities), you should be able to propagate that uncertainty forward into the downstream analyses.

Also appears first author is in this slack channel.

Toryn Schafer (tschafer@tamu.edu)

2024-09-08 12:37:44

Just catching up with the discussion! For the model sketch from @Casey Youngflesh, I tried having a friend go down that route for a multi-stage model with classification errors. The model software wasn't happy with the specification so we ended up just defining each probability assuming independence of the classification and observation processes, we had:
P(y = s| latent = s) = P(detected and classified as s| latent = s ) = P(detected | latent = s)P(classified as s)

Toryn Schafer (tschafer@tamu.edu)

2024-09-08 12:38:47

This is probably also prone to some of challenges due to the simplifying assumptions pointed out above

Keiller Nogueira (keillernogueira@gmail.com)

2024-09-06 06:17:20

********************************************************************************************************************************

2nd Data-Centric Land Cover Classification Challenge

part of the Workshop on Machine Vision for Earth Observation and Environment Monitoring (MVEO)
in conjunction with the British Machine Vision Conference (BMVC) 2024

Glasgow, Scotland, UK

https://mveo.github.io/challenge.html
https://www.kaggle.com/competitions/data-centric-land-cover-classification-challenge/overview

********************************************************************************************************************************

Following the success of the previous challenge held in 2023, the Data-Centric Land Cover Classification Challenge is back. This time, participants will have to develop an AI-based ranking system that can rank the samples based on their levels of label noise.

To this end, a semantic segmentation dataset composed of 5,000 256x256 images and their corresponding (noisy) labels will be provided.

Success is measured by comparing the submitted ranking with an undisclosed one and calculating the Kendall Tau score.

********************************************************************************************************************************

The final results of this challenge will be presented during the Workshop. The authors of the top-ranked methods will be invited to present their approaches at the Workshop in Glasgow/UK, on 28 November 2024. These authors will also be invited to co-author a journal paper which will summarize the outcome of this challenge.

********************************************************************************************************************************

IMPORTANT DATES

Challenge Deadline: Sunday, 10 November 2024
Workshop: Thursday, 28 November 2024

********************************************************************************************************************************

ORGANIZERS

Keiller Nogueira, University of Liverpool, UK
Ronny Hänsch, German Aerospace Center (DLR), Germany
June Moh Goo, University College London (UCL), UK
Zichao Zeng, University College London (UCL), UK
Pallavi Jain, Inria, France
Zhipeng Liu, University of Exeter, UK

mveo.github.io

Workshop on Machine Vision for Earth Observation and Environment Monitoring

Original URL: https://mveo.github.io/challenge.html

kaggle.com

Data-Centric Land Cover Classification Challenge

Develop a ranking strategy that ranks each sample based on their level of label noise

Original URL: https://www.kaggle.com/competitions/data-centric-land-cover-classification-challenge/overview

👀 Valerie, Elizabeth Campolongo

Céline Angonin (C.Angonin@tilburguniversity.edu)

2024-09-09 11:01:45

[Bioacoustics dataset list]
Hello!

During the last months, I worked with gathering publicly available bioacoustics datasets into one list. I noticed that such a resource didn't exist, so I thought that it would be useful for the community. You can consult it here (https://bioacoustic-ai.github.io/bioacoustics-datasets/), and this is the link of the corresponding github repository (https://github.com/bioacoustic-ai/bioacoustics-datasets). I am and will be continuously adding information and new datasets to this list.

You're very welcome to contribute! You can either add new datasets, or open github issues with a request or a suggestion. If you prefer, you can also DM me on Slack. 😁 Please consider starring the github repo to help people discover this resource! ⭐

🙌 Carly Batist, Sara Beery, Burooj Ghani, Brian Geuther, Toryn Schafer, Ilyass Moummad, Rupa Kurinchi-Vendhan, Shir Bar, Subhransu Maji, Kishore Panaganti, Vincent Lostanlen, Ben Williams, charlotte, Aarshi Jain, Juan Sebastián Cañas, Sergei Nozdrenkov, Elizabeth Fawcett, Meredith Palmer

❤️ Inês Nolasco, Aakash Gupta, Vanesa Reyes, Sara Beery, Brian Geuther, Arthur Caillau, Maddie Cusimano, Laura Madrid, Chris Lange, Vincent Lostanlen, Lukas Picek, Aarshi Jain, Georgia Atkinson, Juan Sebastián Cañas, Carl Boettiger, Talia Speaker, Elizabeth Campolongo

😎 Jon Van Oast, Vincent Lostanlen, Aarshi Jain, Juan Sebastián Cañas

⭐ Maddie Cusimano, Vincent Lostanlen, Aarshi Jain, Juan Sebastián Cañas, Alexander Merdian-Tarko

🙌:skin_tone_3: Alan Stenhouse

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-09-09 11:04:16

Awesome!!

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-09-09 11:04:23

Had you seen this list before? It has a bunch more as well https://bioacousticsdatasets.weebly.com/

bioacousticsdatasets.weebly.com

BIOACOUSTICS DATASETS

A list of bioacoustics datasets. All datasets listed below are easily accessible and can be downloaded in one go. See below for a list of Bioacoustics Online Repositories that offer a more...

Original URL: https://bioacousticsdatasets.weebly.com/

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-09-09 11:06:02

It would be good to specify what you mean by ‘dataset’ too - species-specific calls, ML training datasets, annotated or un-annotated, etc.? Like are you wanting Xeno-Canto, Macaulay library, etc. there too?

Céline Angonin (C.Angonin@tilburguniversity.edu)

2024-09-09 11:12:25

Yes, I did see it! I included their datasets in my list. 😊

Like are you wanting Xeno-Canto, Macaulay library, etc. there too?

Good point, I should make it clearer in the repo. I'm not considering databases or repositories like Xeno-Canto whose content changes over time, I am considering ML-ready datasets. Both labeled and unlabeled data are fine. I'll update the README tomorrow!

👍 Carly Batist

🙌 Carly Batist

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-09-09 11:14:59

In that case, I would also recommend cross-checking your list with LILA BC too!

👍 Céline Angonin

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-09-09 11:15:01

there are a number of ML-ready annotated datasets there as well - https://lila.science/otherdatasets#bioacoustics

LILA BC

LILA BC - List of other conservation-related data sets

Written by

lilawp

Est. reading time

38 minutes

Original URL: https://lila.science/otherdatasets#bioacoustics

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-09-09 11:16:09

Might also be a good idea to post about this list on WILDLABS too!

wildlabs.net

WILDLABS Virtual Meetup: Bioacoustics Data Analysis and AI | WILDLABS

Original URL: https://wildlabs.net/

👍 Céline Angonin

💯 Vanesa Reyes

Sara Beery (sara.beery@gmail.com)

2024-09-09 11:22:43

Awesome!!!!

☺️ Vincent Lostanlen

Céline Angonin (C.Angonin@tilburguniversity.edu)

2024-09-09 11:33:47

In that case, I would also recommend cross-checking your list with LILA BC too!

Yup, already done that! I used their list to add some datasets 😊

Holger Klinck (hk829@cornell.edu)

2024-09-09 11:35:21

There are a bunch more datasets out there: https://zenodo.org/search?q=Klinck&f=resource_type%3Adataset&l=list&p=1&s=10&sort=bestmatch Scroll through the list...

Céline Angonin (C.Angonin@tilburguniversity.edu)

2024-09-09 11:37:45

I will go through it, thanks!

Sam Lapp (sam.lapp@pitt.edu)

2024-09-10 16:41:04

this is great, I’ve added it to the Bioacoustics resource page

here’s one more list by @Tessa Rhinehart to check
https://docs.google.com/spreadsheets/d/1KrmCB0vvSK7V3znJfycO-eOMZJKP2F-Ih6neRYPz1Xc/edit?gid=0#gid=0

🙌 Carly Batist, Céline Angonin

Céline Angonin (C.Angonin@tilburguniversity.edu)

2024-09-11 03:00:36

Thank you very much! I have already gone through Tessa's list when I curated the datasets list. I think I will mention in the README all the resources I have gone through to compile my list. 🤔

👍 Carly Batist, Elizabeth Campolongo

Ben Williams (ben.williams.20@ucl.ac.uk)

2024-09-11 06:57:53

Just submitted a pull request for our coral reef dataset 🪸

Céline Angonin (C.Angonin@tilburguniversity.edu)

2024-09-11 07:04:05

Thanks! 😍 By the way, if you found the procedure to add a dataset or anything else unclear, don't hesitate to tell me!

Lukas Picek (lukaspicek@gmail.com)

2024-09-11 08:01:01

Great work! Just a small comment, it is hard to read if you have a “dark-mode” ON in your browser.

Céline Angonin (C.Angonin@tilburguniversity.edu)

2024-09-11 08:03:52

That's not hard, that's impossible 😂 I didn't think about that, thank you for raising the issue, I'll look at mitigating it!

Dan Morris (agentmorris@gmail.com)

2024-09-11 17:40:40

Mmmm, lists of datasets are like my favorite thing (just behind baby animals, guitars, and football). I added a link to your page from the list of acoustic datasets on LILA's "other datasets" page (the same one linked to earlier on this thread):

https://lila.science/otherdatasets#bioacoustics

🙌 Carly Batist, Lukas Picek, Sara Beery

❤️ Céline Angonin, Tessa Rhinehart, Elizabeth Campolongo

Céline Angonin (C.Angonin@tilburguniversity.edu)

2024-09-12 03:58:45

Thank you! 🤩

Hunter P (hunter@hunterpitelka.com)

2024-09-11 23:38:17

I'm hiring a Software Engineer!

Skylight is an AI-powered platform designed to combat illegal fishing by detecting vessels in satellite imagery and providing actionable insights to governments and organizations. As part of the team, you’ll work on cutting-edge technology that helps protect our oceans and supports global sustainability efforts. We're a team of 5 software engineers that partner with some of the leading AI researchers to apply AI to real world conservation problems and deliver actionable products to real users. Today Skylight is used in about 80 countries around the world! Check out the job posting here and feel free to DM me if you have any questions or referrals!

🐟 Sara Beery, Alexander Merdian-Tarko, Alan Stenhouse

👍 Tim Gardner, Angela Zhu

Atriya Sen (atriya@atriyasen.com)

2024-09-16 19:07:33

Hiring a fully-funded Graduate Research Assistant at Oklahoma State University

I am an Assistant Professor in the Computer Science department at Oklahoma State University, looking to hire a PhD or MS student Graduate Research Assistant (full tuition waiver and standard stipend) on an NSF-funded project aiming to apply AI techniques to conservation biology applications, with a focus on intelligently resolving taxonomic uncertainty.

Here is the award: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2426835

I would love to hire someone who has a background in biology (and ideally an interest in systematics and/or conservation biology), but who is also interested in working with cutting-edge AI techniques and has some experience with programming and perhaps basic AI.

The prospective GRA should email me directly at atriya.sen@okstate.edu, with a CV and short statement of interest. The student would ideally start in Spring 2025, but Fall 2025 applicants will be considered.

😎 Jason Holmberg (Wild Me), Andy Viet Huynh

🎉 Hammed Akande

Atriya Sen (atriya@atriyasen.com)

2024-09-19 15:04:34

I am also open to hiring a post-doc for this position in lieu of a graduate student.

🙌 Hammed Akande

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2024-09-17 05:46:32

Hi all, If a PhD student here with computer vision background is looking for a 3-month internship in a project related to CV and Robotics for Conservation, and is based in the EU (preferably) please DM me asap. I will be happy to recruit someone from here before posting the ad widely. For more info on the project in which the internship will be embedded, please check: https://www.aamirahmad.de/projects/wildcap/

flightrobotics

WildCap

WildCap Autonomous Non-Invasive Monitoring of Animal Behavior and Motion

Original URL: https://www.aamirahmad.de/projects/wildcap/

🎉 Jon Van Oast, Sara Beery, Alex Rood, Nora Gourmelon, Marion Richardot

Dan Stowell (dan.stowell@naturalis.nl)

2024-09-18 06:45:18

Datasets for bioacoustics! So many of them: https://bioacoustic-ai.github.io/bioacoustics-datasets/ As part of the Bioacoustic AI project, @Céline Angonin has produced a new searchable listing of audio data sources for animal sounds. Check it out and find a new dataset (or contribute your own).

🐦 Nicolas Arrieta Larraza, Sara Beery, Shir Bar, Elizabeth Campolongo, Carl Boettiger

👏 Rita Pucci, Alex Rood, Talia Speaker, Enis Berk Çoban, Taku OP Iwaki

👍 Carly Batist

❤️ Jon Van Oast, Enis Berk Çoban, Morgan Ziegenhorn (she/they)

👏:skin_tone_5: Prabath Gunawardane

🎯 John Martinsson

Céline Angonin (C.Angonin@tilburguniversity.edu)

2024-09-18 06:53:35

Thanks Dan, I've already advertised it above 😁 (https://aiforconservation.slack.com/archives/CLWGQ4BJ6/p1725894105104259)

Céline Angonin (https://aiforconservation.slack.com/team/U06T7HE1HMG)

[Bioacoustics dataset list] Hello! During the last months, I worked with gathering publicly available bioacoustics datasets into one list. I noticed that such a resource didn't exist, so I thought that it would be useful for the community. You can consult it here (<a href="https://bioacoustic-ai.github.io/bioacoustics-datasets/">https://bioacoustic-ai.github.io/bioacoustics-datasets/</a>), and this is the link of the corresponding github repository (<a href="https://github.com/bioacoustic-ai/bioacoustics-datasets">https://github.com/bioacoustic-ai/bioacoustics-datasets</a>). I am and will be continuously adding information and new datasets to this list. You're very welcome to contribute! You can either add new datasets, or open github issues with a request or a suggestion. If you prefer, you can also DM me on Slack. 😁 Please consider starring the github repo to help people discover this resource! ⭐

Original URL: https://aiforconservation.slack.com/archives/CLWGQ4BJ6/p1725894105104259

🫢 Dan Stowell

Dan Stowell (dan.stowell@naturalis.nl)

2024-09-18 07:02:07

Great! (and sorry for double-posting) Thanks everyone for your comments on Celine's post

Alex Rood (alex.rood@wildlabs.net)

2024-09-18 09:12:55

Related event on bioacoustics data analysis and AI happening tomorrow! https://wildlabs.net/event/wildlabs-virtual-meetup-bioacoustics-data-analysis-and-ai

wildlabs.net

Join us for our third session in Season 5, where we will explore the latest advancements and emerging developments in data analysis and machine learning models for acoustic data. On Thursday, September 19 we will hear short talks from leading experts Justin Kitzes, Stefan Kahl, Kate Jones and Sarab Sethi, followed by a lively open discussion and community exchange.

Original URL: https://wildlabs.net/event/wildlabs-virtual-meetup-bioacoustics-data-analysis-and-ai

🙌 Nicolas Arrieta Larraza, Céline Angonin, Vanesa Reyes, Talia Speaker, Sara Beery, Yseult Hb, Antón Álvarez, Kit Lewers

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-09-20 18:22:32

Is anyone going to be at COP16 in Calí next month?! I would love to see some other AI for Conservation folks there 🙂

Justin Kay (justinkay92@gmail.com)

2024-09-20 18:26:21

@Sara Beery @Neha Hulkund and I will be there for part of it 🙂

🙌 Carly Batist, Neha Hulkund

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-09-20 18:31:04

Awesome! Which days will you be there? Are you doing any talks/panels?

We’re (WildMon) part of conservation tech event on the 24th - https://www.cbd.int/side-events/6023

Sara Beery (sara.beery@gmail.com)

2024-09-20 18:31:35

26-30th

👍 Carly Batist, Jon Van Oast

Sara Beery (sara.beery@gmail.com)

2024-09-20 18:32:52

I won't make that one in person unfortunately, but will follow up with a few other events we're taking part in!

🙌 Carly Batist

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-09-20 18:33:42

That would be awesome thanks! Trying to get a list of nature/conservation-tech-focused events together 🙂

👍 Sara Beery

❤️ Jon Van Oast

😎 Jon Van Oast

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2024-09-23 10:41:05

@Carly Batist is the event 6023 going to be broadcasted or can people join remotely?

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-09-23 10:44:56

I believe it’s in-person only unfortunately

👍 Aamir Ahmad

Jon Van Oast (jon@wildme.org)

2024-09-23 14:43:49

interested in any list of conservation tech events compiled and/or people going to be there. i will not personally be there, but likely/hopefully some of my colleagues at CXL will be. gladly pass along info to them.

👍 Carly Batist

🌱 Kalindi Fonda

Sebastien Ouellet (sebouel@gmail.com)

2024-10-22 14:00:51

This isn't related to AI (yet) but I'd be curious to see how solutions from other frameworks (like from the event on the 24th) could be standardized so this Pact could benefit: https://www.linkedin.com/posts/berlin-urban-nature-pact_berlinurbannaturepact-cop16-cities-activity-7254401158918586368-jEpm/

Jennifer Turliuk (jenn.turliuk@gmail.com)

2024-09-26 08:51:14

🌍 Ready to Tackle Climate Change? Join the MIT Energy & Climate Hackathon! 🚀

TLDR: Interested in Renewable Energy, Transportation, Future Mobility, Circularity, Climate Change, and tackling real-world problems - all while getting free swag, food, and the chance to win up to $1,000 per person?
Apply by October 18, 2024, and participate from November 15-17, 2024, to help create a cleaner future.

Past sponsors include Google, McKinsey Sustainability, Crusoe, Schneider Electric, Fifth Wall, and more.

Here’s what we’re diving into:
🌿 Renewable Energy: Envision a world powered entirely by wind, solar, and green hydrogen—clean energy driving everything we do!
🚀 Future Mobility: Picture electric vehicles, high-speed trains, and even flying taxis making your commute fast, fun, and emissions-free!
♻️ Circularity: Let’s rethink how we use and reuse materials to minimize waste and build a sustainable future.

Join a team of up to 4, collaborate with global competitors, and solve real-world energy challenges proposed by our sponsors. Plus, network with top industry leaders and win up to $1,000 per person!

To participate, fill out our form: https://tinyurl.com/mitec-hack-2024
To learn more about the MITEC Hackathon, check out our website: https://www.mitenergyhack.org/

Please reach out if you have any questions! And email me at jturliuk@mit.edu if you’re interested in your company being a challenge sponsor/partner.

Google Docs

MIT Energy and Climate Hackathon Application

This November 15th-17th, students from around the world will come to MIT for three days to hack. Focus areas include renewable energy, transportation and mobility, and circularity. Spots for the 2024 Energy and Climate Hack are limited, so fill out your application as soon as you can. We strongly encourage students from all backgrounds, majors, and levels of study (undergraduate, Master, PhD, MBA) to apply. Applications close on October 18th.

Original URL: https://tinyurl.com/mitec-hack-2024

🌱 Kalindi Fonda, Jennifer, Alan Stenhouse

Jennifer (jzhuge@alumni.cmu.edu)

2024-10-09 16:28:28

Hey! are recent working grads allowed to participate?

Aditya Jain (aditya.jain@mila.quebec)

2024-09-27 18:42:59

Is there any CV4E meetup happening at ECCV?

👀 Julia Chae, Robin Zbinden, Mia Chiquier, Malte Pedersen, Andrew Temple

Sara Beery (sara.beery@gmail.com)

2024-09-29 10:26:36

Organize something!!

Devis Tuia (devis.tuia@epfl.ch)

2024-09-30 11:41:46

tonight I am taken, but if you want to do smth tomorrow or wednesday, let me know

Sara Beery (sara.beery@gmail.com)

2024-09-30 11:44:13

Maybe we could meet up at the reception tomorrow?

👍 Devis Tuia

Aditya Jain (aditya.jain@mila.quebec)

2024-09-30 11:55:30

Sounds good!

Devis Tuia (devis.tuia@epfl.ch)

2024-09-30 11:57:08

Where could we meet? It will be busy…

Devis Tuia (devis.tuia@epfl.ch)

2024-10-01 11:47:44

What about 1830 at the Google booth?

❤️ Sara Beery

Katja Ovchinnikova (e.ovchinnikova@gmail.com)

2024-09-30 14:52:50

Hi everyone! I’m Katja, a data scientist with extensive experience in academia, specializing in AI/ML. Witnessing the urgency of the climate crisis, I made a decision to transition to the climate solutions industry and completed climate-focused courses with Terra.do and Airminers. As part of this transition, I’ve been working as a data science consultant as well as a hands-on freelancer for early-stage climate and environmental startups. This experience has been truly rewarding and has only deepened my desire for working on climate solutions.

I’m especially passionate about biodiversity, particularly regarding the ocean, but also beyond. My past and current projects include collaborations with fisheries and conservation organizations on tasks such as automatic scallop recognition, manta ray and whale identification, coral species recognition, and identifying species in sonar data.

Here’s how I can contribute as a data scientist:

Consult on developing strategies for data-driven products

Advise on data analysis workflows and infrastructure

Prototype and evaluate AI/ML models

Design experiments for data collection and analysis

Write scientific publications and blog posts

Lead workshops on data science in the climate space, as well as broader climate science and solutions topics

If you’re looking for a data science consultant, freelancer, or contractor for your project, I’d love to chat and explore how we can work together! I’m working remotely, self-employed in France. For more details, you can check out my website: www.ovchinnikova.me.

👋 Dan Morris, Sara Beery, Timm Haucke, Nicholas, Howard L Frederick, Alexander Merdian-Tarko, Lingchao Mao

🌱 Kalindi Fonda

👋:skin_tone_3: Alan Stenhouse

Alex Filazzola (alex.filazzola@outlook.com)

2024-10-02 12:14:12

🦋 Hi All! 👋

We are hosting a workshop on building reproducible workflows for species distribution models (SDMs) using SyncroSim. SyncroSim is a free software for non-commercial use (e.g., academics, non-profits, government). It provides an easy-to-use interface that works with open-source packages, whether through a graphical user interface, command line, R, or Python. SyncroSim simplifies the creation of workflows for SDMs, including models like MaxEnt and random forest. In this workshop, participants will learn the basics of running species distribution models and how to create reproducible workflows for scientific modeling, from start to finish. This workshop is part of the annual Society for Open, Reliable, and Transparent Ecology and Evolutionary biology (SORTEE) conference. Attendance is free for members of SORTEE and membership costs $10-$40 depending on career stage. There are also free waivers for members part of other open science organizations.

📅 Date: Tuesday, October 15, 2024
📍 Location: Online
🕰️ Time: 12:30 pm PT; 3:30 pm ET
✅ Sign up here: https://events.humanitix.com/sortee-conference-2024

SyncroSim | Deliver geospatial forecasting models direct to decision makers

SyncroSim | Deliver geospatial models to decision makers

SyncroSim is an open science tool for bundling map-based simulation code and data into geospatial models in a way others can actually use.

Original URL: https://syncrosim.com/

sortee.org

SORTEE

SORTEE blog

Original URL: https://www.sortee.org/

events.humanitix.com

SORTEE Conference 2024 | Humanitix

SORTEE Conference 2024

Location

Online

Date

Tuesday October 15th 2024

Original URL: https://events.humanitix.com/sortee-conference-2024

🤩 Sara Beery, Aakash Gupta, Robin Zbinden, Alan Stenhouse

Luke Meyers (lmeyers@SEATTLEU.EDU)

2024-10-02 20:02:17

Anyone work with (or know about) high performance computing facilities within the Smithsonian? If so would love to get in touch!

Dan Morris (agentmorris@gmail.com)

2024-10-06 10:15:48

New dataset on LILA, containing ~119k images from downward-facing small animal cameras (see the adorable chipmunk at the end of this post):

https://lila.science/datasets/ohio-small-animals/

Thanks to Greg Lipps and Sowbaranika Balasubramaniam from Ohio State for this dataset.

Hopefully this is the first of several datasets from downward-facing cameras, which are not covered in existing public datasets and are not handled well by existing models, including MD.

🎉 Timm Haucke, Sara Beery, Aude Vuilli, Justin Kay, Atul Ingle, murilo, Sankalpa Ghose, Shir Bar, Shawn Johnson, Risa Shinoda, Ștefan Istrate, Marion Richardot, Malte Pedersen, Viktor Domazetoski, Jenna Kline, Jon Van Oast, Roberta Hunt, Tiziana Gelmi Candusso, Sowbaranika, Eric Cunningham, Jennifer, Alan Stenhouse

😲 Mitchell Rogers, Marion Richardot, Arky

🐿️ Chris Lange, Jess Tam, Toryn Schafer, Alexander Merdian-Tarko, Elizabeth Campolongo

🐭 Jenna Kline

Tom August (tomaug@ceh.ac.uk)

2024-10-08 04:27:24

I'm looking for people attending COP16 in Cali, Colombia who would be interested in joining a panel on "What is the role of AI in addressing the biodiversity crisis?". The panel will be hosted by the British embassy, and take place for one hour between 22 October from 14:00 to 18:00. If you think you could contribute please drop me a message

🙌 Antón Álvarez, Justin Kay, Carly Batist, Jon Van Oast

👍 Carly Batist, Jon Van Oast

🙌:skin_tone_3: Alan Stenhouse

Drea Burbank (drea@savimbo.com)

2024-10-08 16:15:52

Hey guys, we need some data scientists to fix the 80% stat. https://news.mongabay.com/2024/09/do-indigenous-peoples-really-conserve-80-of-the-worlds-biodiversity. We’re getting together a scientific workgroup for this that will meet weekly for three months following COP and try do crank out a paper as a coalition the way the Open Science Collaboration did. Sign up here. Meetings run bi-weekly starting Nov 5th.

Mongabay Environmental News

Do Indigenous peoples really conserve 80% of the world’s biodiversity?

A new commentary piece in the journal Nature has caused an uproar in the conservation world and has prompted scientists to call for rigorous study. According to the authors, the much-cited claim that 80% of the world’s remaining biodiversity is protected by Indigenous peoples is not only a “baseless statistic” that is not supported by […]

Written by

Latoya Abulu

Est. reading time

14 minutes

Original URL: https://news.mongabay.com/2024/09/do-indigenous-peoples-really-conserve-80-of-the-worlds-biodiversity

Airtable

Airtable | Everyone's app platform

Airtable is a low-code platform for building collaborative apps. Customize your workflow, collaborate, and achieve ambitious outcomes. Get started for free.

Original URL: https://airtable.com/appAu6Y0slntKiUDV/pagvLbi0w9PqzMdFt/form

🙌 Carly Batist, Alessandra Vidal Meza, Sara Beery, murilo, Thijs van der Plas, Devi Ayyagari

❤️ Jon Van Oast

🙌:skin_tone_5: Prabath Gunawardane

Jon Van Oast (jon@wildme.org)

2024-10-08 17:23:29

is there a shareable/public version of this scientific workgroup call? i would like to share it with folks not on slack. thanks!

Pietro Perona (perona@caltech.edu)

2024-10-10 02:19:04

Has anyone read the 2024 WWF report "Living Planet"? Do people understand why the uncertainty in their measurements is increasing with time instead of going down? What's going wrong?

Aakash Gupta (aakash@thinkevolveconsulting.com)

2024-10-10 04:41:24

When making predictions about future trends, especially in ecology, the confidence in those projections diminishes because of the unpredictability of various environmental factors (climate change, human impacts, conservation efforts). Thus, future uncertainty grows as a function of both complexity and incomplete data.

🙏 Pietro Perona

Pietro Perona (perona@caltech.edu)

2024-10-10 05:18:05

Thank you @Aakash Gupta. The top plot does not extrapolate though. The y axis is on a log scale and thus, if the confidence interval is constant in time, the confidence interval will appear to become larger as the index declines. Still, I am surprised that the confidence interval does not shrink in time given all the ongoing work to obtain accurate measurements.

👀 Clemens Mosig

Holger Klinck (hk829@cornell.edu)

2024-10-10 06:35:23

This may also provide some insights: https://communities.springernature.com/posts/the-living-planet-index-is-not-a-reliable-measure-of-population-changes

Research Communities by Springer Nature

The Living Planet Index is not a reliable measure of population changes

A widely used index characterizing the average change in population sizes suffers from several mathematical and statistical issues, leading to a bias towards an apparent decrease even for balanced populations. The claim that vertebrate populations declined by 70% since 1970 is thus unsubstantiated.

Original URL: https://communities.springernature.com/posts/the-living-planet-index-is-not-a-reliable-measure-of-population-changes

👀 Steve Haddock

🙏 Pietro Perona

Teodoro Topa (teodorotopa@gmail.com)

2024-10-10 09:38:39

@Pietro Perona I think it’s because it is indexed to 1970 levels. The error bar size seems like a function of time rather than quality of biodiversity estimates. Since nobody has exact numbers on how many animals there were in 1970 or how many there are today, as we get further from 1970 and more time passes, we can be less and less sure of exactly how far we have fallen the starting point. Basically, in 1971, very little time had passed, so regardless of the quality of biodiversity estimates, the confidence range was small because there just was not enough time between 1970 and 1971 for estimations to get so far away from reflecting what was likely the reality.

🙏 Pietro Perona

👍 Omiros Pantazis

Heather Lynch (heather.lynch@stonybrook.edu)

2024-10-31 11:22:27

We address some of these issues in our recent-ish paper:

Ben Williams (ben.williams.20@ucl.ac.uk)

2024-10-11 11:55:01

Hi all, we're submitting a funding bid and want to put down a sensible ballpark figure for some software engineering work. Does anyone have an idea, or know a good place to look, as to how much it would cost to package up an existing ML model and processing pipeline into a nice user friendly interface with a few small bells and whistles?

In more detail, we would be passing through audio data, running an ensemble classifier on this, then having humans reviews the outputs usually just to select the true positives, then applying these labels to all the audio data. Some summary outputs, e.g number labelled number of matches etc would be nice. We would also likely want to be able to update the model checkpoint with new versions every now and then (trained outside the software) to handle model drift. It would be for in house use or use by close collaborators so doesn't need to look all fancy, it would also need to able to run locally on windows machines.

Nate Harada (nharada1@gmail.com)

2024-10-11 12:23:54

Hey Ben! I can share a few couple questions I’d prob ask if estimating to help nail it down more:

How complex is the ensemble? Do you need GPUs on the windows machines? What kind of software has to be on the machines to make it work (i.e. CUDA, etc)

Is the existing model easily compatible with windows? What format is the model in?

What format and language is the supporting code for the model? How much of it is there?

How technical are the people using the software? Does the update process need to be pretty foolproof or can people download and modify the software themselves?

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2024-10-11 17:51:56

I don't have a $$$ number, because this probably depends on a lot of factors like where you're located and what you're willing to pay (e.g. are you proposing that a postdoc/grad student does this, or would you contract it out?) Generally when I've submitted academic grants, we've based the amount on length of appointment x salary for position, so a year of postdoc time in the UK might be ~80-100k GBP when you include overheads and other lining of the university's pockets.

Conservatively I'd say it's at least a few months' work. For someone with experience in all the components, it's probably possible to get an MVP out in a matter of weeks. For example if you didn't need a standalone executable, you could use a ready-to-go UI like streamlit/gradio and run it straight from Python/Docker. What will take longer is things like end-user/technical documentation, getting installers working reliably (does it have to be a standalone thing?), etc.

You could find a consultant who could bang this out in much less time, but you'll probably pay for that expertise (think several hundred a day at least).

Realistically, you probably want to have someone around long enough to maintain and update things for at least 6 months while you test it.

🙏 Ben Williams

Ben Williams (ben.williams.20@ucl.ac.uk)

2024-10-14 04:56:13

Thanks both! Creation of the model is part of the grant so I can only give guesses as to answers, but here's my best intuition currently:

We are aiming to make it lightweight so it can run with no GPU and some patience

It will likely be written in standard libraries like TF, PyTorch. We'll be designing to make it compatible with windows from the off, as this is what are users in the field have.

Assume not technical at all! In my limited experience of this, I'd imagine something like we upload a new checkpoint file/s somewhere and they just replace the old one with this.

Let me know if that helps narrow down a range @Nate Harada 🙏

Super useful on the postdoc salary point, fully agree getting something stable is going to be where a lot of the hard yards are

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2024-10-14 14:19:51

I think that's sensible. Many models can be run fine on CPU if latency isn't an issue. I would suggest decoupling your training library from the "production" model. You can distribute your model in a standard open format like ONNX (e.g. onnx-runtime). You could also consider writing a native application, for example using C++/Qt. That tends to be a lot cleaner than trying to package up a mess of Python dependencies and it will likely run faster. The downside is that finding developers with that kind of experience (in academia), tends to be harder.

If you're also expecting the hire to train the model as well, then budget for at least a year (do you have datasets, etc ready to go?)

Universities tend to have baseline costs for hiring and they're often (much) higher than you expect - e.g. a postdoc in London might get paid 40k, but the University will charge double or more. We've been burned by this in the past (e.g. the HR overhead was so high that we had to scale back the project), so I would highly recommend speaking to someone in HR/grants and getting estimates for staff at different levels of seniority.

Ben Williams (ben.williams.20@ucl.ac.uk)

2024-10-18 12:51:47

Thanks @Josh Veitch-Michaelis! Will certainly be separating the training and deployment aspects, the model will already be developed and we would just be asking the hire to package it up into some kind of desktop app type thing. Good to know on ONNX and C++ etc, thats super helpful. Yeah my PI is experienced in all the HR stuff for grants so we're aware we basically need to double the salary cost for the overheads, we just needed a ballpark figure on this. So we're now budgeting for a 6month post doc and can use these funds to either look for one or for an external contractor, thanks so much for all the tips

👍 Josh Veitch-Michaelis

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2024-10-18 13:46:10

Great, best of luck. Feel free to get in touch if you have any other questions!

Jon Van Oast (jon@wildme.org)

2024-10-11 17:50:07

maybe of interest to some here: https://spatialagent.org/BiodiversityHackathon/banner.html

Brittany Aguilar (baguilar@schmidtsciences.org)

2024-10-12 16:19:06

Hi all, A friend of mine works for WWF and they shared an upcoming AI x Conservation virtual event/conversation they are hosting with some of their experts that you may be interested in attending. It is a virtual event. I think you just need to email Christina for an invite...

Your passion for technology and conservation is inspiring, and because of that I’m excited to invite you to a special virtual presentation showcasing how WWF is deploying AI across our conservation efforts, led by WWF experts Abby Hehmeyer and Lu Gao. We’re gathering a few like-minds to dive into the specifics of two AI-driven projects that are revolutionizing wildlife monitoring and combating wildlife trafficking, all to protect endangered species and preserve biodiversity.

Your expertise would be particularly useful to our team, especially as we discuss WWF’s broader technology vision.

Date: Tuesday, October 15, 2024
Time: 11:00 AM PST / 1:00 PM CST / 2:00 PM EST
Location: ZOOM (RSVP for details)

Please RSVP for more details by simply responding to this email or by calling/texting me at 202-495-4576. Abby, Lu, myself, and others are hoping to see you on the call and share our collective enthusiasm for the use of technology in conservation!

Contact:
Christina Armstrong, CFRE
christina.armstrong@wwfus.org

Sara Beery (sara.beery@gmail.com)

2024-10-14 11:44:46

Flagging this in case people hadn't seen it!

https://www.planet.com/pulse/planets-project-centinela-monitoring-vulnerable-biodiversity-hotspots-for-conservation-action/

planet.com

Planet’s Project Centinela: Monitoring Vulnerable Biodiver...

Planet | Planet’s Project Centinela: Monitoring Vulnerable Biodiversity Hotspots for Conservation Action

Original URL: https://www.planet.com/pulse/planets-project-centinela-monitoring-vulnerable-biodiversity-hotspots-for-conservation-action/

🙌 Anna Ding

Jes Lefcourt (jeslefcourt@gmail.com)

2024-10-14 18:03:35

Sara, are you connected to the team at Planet working on this?

Sara Beery (sara.beery@gmail.com)

2024-10-14 18:13:51

Not directly on this project! But I flagged because it seemed like it might provide opportunities for biodiversity projects to access data.

Jes Lefcourt (jeslefcourt@gmail.com)

2024-10-14 18:42:28

Absolutely! It's great!

Riley Knoedler (mknoedler@west-inc.com)

2024-10-16 15:55:00

I posted this on WILDLABs but I'm cross-posting here: it occurred to me and my colleagues recently that the self-driving vehicle industry must be working on their own animal classification models, because it matters if you are about to collide with a turkey or a moose or a pet dog. Is anyone in this community involved in that sphere? Does anyone know of opportunities to access these data or collaborate on model development?

🤔 Elizabeth Campolongo

Elizabeth Campolongo (e.campolongo479@gmail.com)

2024-10-16 16:04:15

That's a good point. Though, I'd expect them to be more focused on size considerations. Curious if anyone has any insight and would be interested in the data too!

Riley Knoedler (mknoedler@west-inc.com)

2024-10-16 16:26:04

Well, I wonder if there's any appetite for being able to distinguish between sensitive and non-sensitive species as well, at least from PR perspective

Elizabeth Campolongo (e.campolongo479@gmail.com)

2024-10-16 16:45:46

I would certainly hope so 😭 or maybe they don’t want to hit pets or a skunk 🤷‍♀️

Levi Cai (levi@levicai.com)

2024-10-16 17:27:46

Seems vaguely like (a) yes they do make their own detectors/classifiers for animals but (b) probably extraordinarily painful to collaborate/get access to anything of the kind for an external user of any kind.

Wenxin Yang (wenxinyang@ucsb.edu)

2024-10-16 18:23:36

I’m not working in this direction but remember Volvo was trying to develop a system to detect animals and the system was confused by kangaroos (news article).

the Guardian

Volvo admits its self-driving cars are confused by kangaroos

Swedish company’s animal detection system can identify and avoid deer, elk and caribou, but is yet to work against the marsupials’ movements

Original URL: https://www.theguardian.com/technology/2017/jul/01/volvo-admits-its-self-driving-cars-are-confused-by-kangaroos

michele volpi (michele.volpi@sdsc.ethz.ch)

2024-10-17 10:43:33

I have a friend who worked in the industry, for lidar data they mostly consider large animals (which crash the car), cats and dogs (potential lawsuits) and birds (flying in front of sensors could lead to problems). Some animal classes were considered but just for evaluation, not for training. Accessing data depends on the company but is often a no-go

👍 Chris Lange, Elizabeth Campolongo

Riley Knoedler (mknoedler@west-inc.com)

2024-10-17 13:11:04

Thanks for all the replies, very helpful insights!

Tom August (tomaug@ceh.ac.uk)

2024-10-17 11:33:44

For anyone attending COP16 you can now register for our event, which includes a panel "What is the Role of AI in Addressing the Biodiversity Crisis?", and free drinks 🍸

Eventbrite

The UK's Role in Biodiversity Monitoring and Modelling

Gathering of top experts in biodiversity monitoring and modelling to discuss science's role in delivering the Kunming-Montreal Framework.

Where

4-84 Avenida 4 Oeste, Cali, Valle del Cauca 760045, Colombia

When

Tue 22 Oct 2024 at 14:00

Original URL: https://www.eventbrite.co.uk/e/the-uks-role-in-biodiversity-monitoring-and-modelling-tickets-1049475037087?aff=oddtdtcreator

👏 Morgan Ziegenhorn (she/they)

🙌 Thijs van der Plas, Sara Beery, Arky, Ted Schmitt, Carly Batist

♥️ Morgan Langley

Tom August (tomaug@ceh.ac.uk)

2024-10-17 11:35:04

I should add that while the whole event it tagged as "The UK's Role" - the panel will be globally focussed!

Arky (hitmanarky@gmail.com)

2024-10-18 01:34:59

Cerulean platform for monitoring oil spills is now in public beta.

https://cerulean.skytruth.org/

🙌 Jonathan Roberts, Sonny Burniston, Antón Álvarez

Filippo Varini (fppvrn@gmail.com)

2024-10-19 09:00:47

Hello everyone, has anyone tried out META’s SAM2 for ecological application?
I am exploring its potential for segmenting and tracking objects in underwater videos and would love to discuss pros and cons of the approach!

Specifically, I find the following challenges:

Can SAM2, track automatically videos, without prompt, once fine-tuned?

it seems like you can automatically generate masks but it is not clear whether you can automatically track as well.

Can SAM2 detect and track new objects appearing in the video?

Does it perform well with many objects (i.e. a school of fish)?

Would love to discuss the potential of this new tech with the community 🙂

👍 Sonny Burniston, Mitchell Rogers, Nathan Fox, Roberta Hunt

👍:skin_tone_3: Alan Stenhouse

Filippo Varini (fppvrn@gmail.com)

2024-10-19 09:06:52

Would love to share code and insights.
So far I have been playing the following notebooks:

video predictor

automatic_mask_generator

ultralytics implementation

Brian Geuther (Brian.Geuther@gmail.com)

2024-10-21 15:14:03

I haven't tried SAM2 yet, but I did try out TAM, which was based on SAM1 (https://arxiv.org/abs/2304.11968)
A couple notes I had from testing out TAM:

We could only fit ~600 frames at once (a restriction that may exist with SAM2)

SAM supervision works fairly well when things that look identical are separated, but fail by merging them together when they're touching

IR/monochrome imaging isn't well represented in their foundational training data and as such have basic issues for things like shadow influencing borders

Despite those drawbacks, we could pretty reliably get okay tracking with 1 click per object for about 10-15s (of the 20s clip)

Filippo Varini (fppvrn@gmail.com)

2024-10-22 00:57:51

Thanks Brian, that sounds interesting!

Georgia Atkinson (g.atkinson@ncl.ac.uk)

2024-10-22 05:46:34

Hi Filippo, I'm also looking into applying SAM2 for underwater applications! I've been trying to get around having to do point/bounding box prompts so I've been having a look at Grounded SAM2 to do text prompting instead which seems to be giving ok results thus far (I only started playing around with it last week). I'm a bit unclear on whether after some fine tuning I'd be able to automatically generate masks for the specific classes I'm interested in or not however.

Filippo Varini (fppvrn@gmail.com)

2024-10-22 08:19:35

Thanks Georgia! I will have a look at Grounding Dino and let’s keep in contact about updates!

Donna Auguste (donna@augusteresearchgroup.com)

2024-11-21 14:40:43

I would like to learn more about this use case. I'll begin collecting underwater images in 2025 from a local lagoon.

Aakash Gupta (aakash@thinkevolveconsulting.com)

2024-10-29 00:09:26

New Dataset: A comprehensive annotated image dataset for real-time fish detection in pond settings

The dataset contains 10,607 annotated fish instances across 586 images. The dataset features several computer vision challenges, including occlusion, turbid water conditions with turbidity levels ranging from 40 to 80 NTU, high fish density per frame, and varying lighting conditions with illumination values between 100 and 500 lux. These factors significantly impact image quality and detection performance, making the dataset ideal for developing robust computer vision models capable of handling complex real-world scenarios.

https://www.sciencedirect.com/science/article/pii/S2352340924009697

🙌 Elizabeth Campolongo, Jason Holmberg (Wild Me), Braden Charles DeMattei, Morgan Ziegenhorn (she/they), Carly Batist, murilo

🐟 Jason Holmberg (Wild Me), Dan Morris, Matthias Zuerl, Leonie Hodel, Shir Bar, murilo, Alan Stenhouse

🐠 Jason Holmberg (Wild Me), Urs, murilo

Dan Stowell (dan.stowell@naturalis.nl)

2024-10-30 08:46:49

PhD to recruit! A collaboration between Naturalis and Tilburg University on fossil images, in a project called "HAICu" - Please circulate this short link to anyone who might be interested https://tiu.nu/22517

👍 Oisin Mac Aodha, Sara Beery, murilo, Shir Bar, Jason Holmberg (Wild Me)

Elke Windschitl (Elke.Windschitl@islandconservation.org)

2024-10-30 16:53:39

Hi all, I am working on a project with Island Conservation where we have georeferenced drone imagery of burrow holes. There are usually many burrow holes per image (dozens). Previously these holes have been counted by hand with ImageJ, but I am exploring alternative solutions with machine learning. I would love to connect with someone who might have experience with this or a similar application. Please comment or message me if you would be willing to discuss. Thanks so much!

❤️ Sara Beery, David Russell, Timm Haucke, Andy Viet Huynh, Seven Song, Anna Willoughby

Ben Weinstein (benweinstein2010@gmail.com)

2024-10-30 16:54:40

can you drop in a screenshot?

Elke Windschitl (Elke.Windschitl@islandconservation.org)

2024-10-30 16:57:32

Here’s an original image plus zoomed in on one section of burrows

Ben Weinstein (benweinstein2010@gmail.com)

2024-10-30 16:58:11

and you want to locate them, like with a box or point on the entry.

Ben Weinstein (benweinstein2010@gmail.com)

2024-10-30 17:00:48

okay, so depending on how exactly image J was run, you may be able to recover the position of previous annotations over the image. If not, in like an hour on http://labelme.csail.mit.edu/Release3.0/, you probably could label 25 images, and run then through an object detection tool like ours. https://deepforest.readthedocs.io/en/latest/user_guide/11_training.html you can make an discussion on the github repo where you get stuck.

David Russell (davidrussell327@gmail.com)

2024-10-30 17:04:14

Once you have the manual annotations and/or ML predictions, do you have to worry about double counting burrows across multiple overlapping images? If so, some of the work we've done such as this tool might become useful. You could use this as an additional post-processing step after following the steps Ben described to generate ML predictions on each image.

open-forest-observatory/geograypher

Enabling Geospatial Predictions from Individual Drone Images

Website

<https://open-forest-observatory.github.io/geograypher/>

Stars

❤️ Timm Haucke

Elke Windschitl (Elke.Windschitl@islandconservation.org)

2024-10-31 11:27:33

Thank you so much I will check out these tools! Correct in that we don’t want to double count holes. We were thinking of breaking apart the stitched together orthomosaic.

👍 David Russell

David Russell (davidrussell327@gmail.com)

2024-10-31 11:33:30

Cool, this workflow also relies on running structure from motion as a first step, you just generate predictions on the raw images (rather than ortho) and then try to suppress duplicate detections geometrically. The motivation is you may have higher quality detections on the raw images than the orthos, since they don't have stitching artifacts and you get multiple perspectives.

Elke Windschitl (Elke.Windschitl@islandconservation.org)

2024-10-31 12:00:11

Oh that is very cool

Riley Knoedler (mknoedler@west-inc.com)

2024-11-01 12:54:18

Hi Elke, I've been working on a similar project to detect prairie dog burrows from drone imagery. We trained a model in YOLOv8, I think we are using a similar approach to what David has referenced to merge detections together from different overlapping images. Happy to discuss more with you!

❤️ Elke Windschitl

Lasha Otarashvili (otarashvililasha@gmail.com)

2024-10-31 11:17:32

New model is out for wild animal re-id, covering 64 species from mixed data sources.

The is Wild Me's strongest generation of re-id model at use on Wildbook platforms, covering cetaceans, mammals, carnivores and more. The model shows state-of-the-art generalization ability for search queries on Wildbook platforms. For this release, the model is stripped down and made easy to run and experiment with.

Model weights and List of species included
Ready-to-use scripts

Hopefuly this helps not only ones that have or have not worked with MiewID previously. The knowledge transfer lowers the barrier and opens up the possibility to adapt deep learning models to datasets previously too small to train.

huggingface.co

conservationxlabs/miewid-msv3 · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Original URL: https://huggingface.co/conservationxlabs/miewid-msv3

😎 Jason Holmberg (Wild Me), Timm Haucke, Jon Van Oast, Matthias Zuerl, Dante Wasmuht, Sara Beery

🎉 Jason Holmberg (Wild Me), Lukas Picek, Vojta Čermák, Justin Kay, Timm Haucke, Rohan Sawahn, Meredith Palmer, Avi Sundaresan, Nicolas Arrieta Larraza, Dante Wasmuht, Christoph Praschl, Carly Batist, Jennifer, Alexander Merdian-Tarko, Antón Álvarez

👍 Kostas Papafitsoros, Timm Haucke, Dante Wasmuht, Piotr Tynecki, Seven Song, Ian Ingram

❤️ Sara Beery, Ian Ingram

Lasha Otarashvili (otarashvililasha@gmail.com)

2024-10-31 11:27:13

Thread Reply:

Lasha Otarashvili (otarashvililasha@gmail.com)

2024-10-31 11:27:46

From-scratch vs leveraging model weights to adapt to new datasets.

Lasha Otarashvili (otarashvililasha@gmail.com)

2024-10-31 11:28:50

A paper with detailed zero-shot and fine-tuning studies will be available next week. Model also transfers well to 'similar' species in a 0-shot manner, for example, the dataset includes lots of dorsal fin samples, so the model just generally does well at them, regardless of species. Moreover, the 64 species can easily be fine-tuned for a new domains to make use of knowledge transfer for a general re-id task. The positive effect varies on per-species basis from limited to dramatic. The knowledge transfer could especially be handy for those working with just small amounts of labelled data.

Matthias Zuerl (matthias.zuerl@fau.de)

2024-11-04 17:28:10

Awesome work! Can you please also post the mentioned publication here as well? I am very interested in reading it!

👍 Savita Owens

Matthias Zuerl (matthias.zuerl@fau.de)

2024-11-14 17:25:42

Hey there! Is the publication already available? I am particularly interested in the plot you show and would love to read further information on that! 🙂

➕ Sepand Dyanatkar

Kostas Papafitsoros (k.papafitsoros@qmul.ac.uk)

2024-12-06 05:17:47

Hi @Lasha Otarashvili I also wonder whether the paper is online yet?

☝️ Matthias Zuerl, Sepand Dyanatkar

Lasha Otarashvili (otarashvililasha@gmail.com)

2024-12-07 04:59:22

Hi all! Sorry for stalling the submission. We just submitted and are waiting for the Arxiv review - hopefully, the paper should be available early next week!

👍 Sepand Dyanatkar

Matthias Zuerl (matthias.zuerl@fau.de)

2024-12-07 11:52:25

Awesome! 🙂

Lasha Otarashvili (otarashvililasha@gmail.com)

2024-12-11 11:19:12

Hi! The paper is now available: https://arxiv.org/pdf/2412.05602

It describes the previous version (MSv2) of the same (MSv3). The next version is already in the works. The difference between the models really is the number of species included. Happy to bounce ideas.

Ilyass Moummad (mr.ilyassmoummad@gmail.com)

2024-11-04 17:53:30

Check out our robust and efficient feature extractor (Pytorch) of bird sounds, with only ~20M parameters, trained on the Xeno-Canto dataset, covering over 10,000 bird species, ideal for bioacoustic projects. Explore it here: https://huggingface.co/ilyassmoummad/ProtoCLR

huggingface.co

ilyassmoummad/ProtoCLR · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Original URL: https://huggingface.co/ilyassmoummad/ProtoCLR

👍 Masato Hagiwara, Meredith Palmer, Jose Ruiz-Munoz, murilo, Thor Veen, Juan Sebastián Cañas, Burooj Ghani, Don Cosseboom

😎 Jason Holmberg (Wild Me), Don Cosseboom, Clément Sage

🐦 Shir Bar, Juan Sebastián Cañas, Don Cosseboom, Nicolas Arrieta Larraza, Aude Vuilli, Elizabeth Campolongo

❤️ Sara Beery

Ilyass Moummad (mr.ilyassmoummad@gmail.com)

2024-11-05 04:23:22

Classification accuracy of ProtoCLR compared to various models on one-shot and five-shot bird sound classification tasks

Aria Ma (aria@luneaera.com)

2024-11-05 13:51:57

Hello all, my team is putting together focus groups for a conservation community platform to empower a network of knowledge sharing. Please let me know if you're working in a conservation nonprofit and would like to join! Happy to send more information via DMs (-:

Brittany Aguilar (baguilar@schmidtsciences.org)

2024-11-05 14:29:32

am I wrong in thinking that's what this slack channel is for? or https://wildlabs.net/ ? I am a funder who is involved with various conservation efforts and I try to support pre-existing communities rather than create new ones... Could you give a little more information on your goals for this group?

Aria Ma (aria@luneaera.com)

2024-11-05 15:12:27

I totally understand where you're coming from! This is a group that is all facets of conservation, not just AI or technology. This means natural solutions from ocean conservation to mangrove restoration. The goal is to connect people from all facets of conservation - not just AI nor technology.

Eugene Galaxy (eugene@animaldetect.com)

2024-11-06 04:39:29

Hey everyone! 👋

I am Eugene and together with my two teammates Hugo & Elina we are building a software tool called Animal Detect to process (filter, label, sort, classify, output) wildlife data as quickly as possible, not necessarily relying on existing AI classifiers. The idea is to make it simple and accessible to people working with wildlife of all technical backgrounds, and do it FAST.

Little about me... My background is in MSc. in Robotics Engineering with focus on Computer vision and AI from Aalborg University, Denmark. I find my resort and peace being alone with nature. I love animals, got PADI diver certification, did climbing for a while. Not surprising I found myself in the intersection of wildlife and technology. 😃

We are looking for feedback, input, and testers for Animal Detect. The first "testable" version is coming in 2 months and we would love to hear your opinions on it: positive, negative, improvements, opinions, anything that comes to mind.
I will drop a couple of links here.

https://www.animaldetect.com - Our homepage with waitlist. If you want to test the platform - join it! 🙂

My X profile - https://x.com/eugene_galaxy, LinkedIn - https://www.linkedin.com/in/eugenegalaxy/
Demo video on Youtube (very early showcase of embeddings technology) - https://www.youtube.com/watch?v=mkT1_jpmewg

DM me if you are interested or know someone who is.
Thank you all and have a nice day! :)

Animal Detect

Manage your wildlife data, remove empty images, and accurately categorize animal species in minutes.

Original URL: https://www.animaldetect.com

❤️ Nicolas Arrieta Larraza, Sara Beery, Malte Pedersen, Jennifer, Antón Álvarez, Elizabeth Campolongo

👀 Helena Russello, Gaspard Dussert, Alexander Merdian-Tarko

Jennifer (jzhuge@alumni.cmu.edu)

2024-11-06 15:34:28

@Velizar

Jennifer (jzhuge@alumni.cmu.edu)

2024-11-06 15:40:05

Hey cool tool! When I go to the link for a website i get 'There's been a glitch' because it takes me to slack.com/yourwebsite, your actual website works!

Eugene Galaxy (eugene@animaldetect.com)

2024-11-07 01:20:57

Oh you are right, thank you for the notice! 💡

It was an extra "/" in the link, whoops. 😇

Jodi Rowley (jodi.rowley@austmus.gov.au)

2024-11-07 22:52:41

Hi everyone! I have a #job opportunity at the Australian Museum in Sydney, Australia. I’m looking for someone to support scientific research using machine learning to identify frog species in audio from the FrogID project (frogid.net.au). The role is full time for one year. You have until 25 Nov to apply. Unfortunately at this stage, you must be an Australian citizen, resident or have current visa that allows them to work in Australia. Sorry!
https://iworkfor.nsw.gov.au/job/scientific-officer-495522

🐸 Brittany Aguilar, Jennifer, Benjamin Hoffman, Akash Nagaraj, Josh Veitch-Michaelis, Nathan Fox, Jess Tam, Brian Geuther

Zhongqi Miao (zhongqi.miao@berkeley.edu)

2024-11-08 22:26:05

Hello everyone, we have just released MegaDetectorV6 and Pytorch-Wildlife V1.1.0. (https://github.com/microsoft/CameraTraps/releases/tag/pw_v1.1.0). In addition, in this release, we also have a third party overhead animal detection model (HerdNet), enabled directly loading customed classification model weights in Pytorch-Wildlife and our gradio app, and further simplified our codebase. We have addressed some of our previous licenses issues and are having a rolling release schedule of many different versions of MegaDetectorV6 to meet with various use cases and provide people with maximum flexibilities. Feel free to test it out and ask us any questions! Thank you very much!

🙌 Kaitlyn Gaynor, Timm Haucke, Piotr Tynecki, Yseult Hb, Justin Kay, Meredith Palmer, Mitch Fennell, Jose Ruiz-Munoz, Aakash Gupta, Polina Stucke, Dante Wasmuht, Malte Pedersen, Ilyass Moummad, Sara Beery, Tiziana Gelmi Candusso, Fagner Cunha, Sepand Dyanatkar, Savita Owens, Jason Holmberg (Wild Me), Arthur Caillau, Alexander Merdian-Tarko

❤️ Avi Sundaresan, Timm Haucke, Justin Kay, Aakash Gupta, Carly Batist, Jennifer, Ilyass Moummad, Sara Beery, Fagner Cunha, Libby White, Jason Holmberg (Wild Me), Arthur Caillau

🙌:skin_tone_2: Cara Appel

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-11-11 11:57:04

This is amazing! As a user, thank you so much for working on this!! You make a big difference for everyone using camera traps 🙌

❤️ Zhongqi Miao

Jennifer (jzhuge@alumni.cmu.edu)

2024-11-14 14:02:00

Amazing work thank you!
I have a dataset that I would like to use the AI4GAmazonRainforest model on, but my data has one more class. What is the best way to incorporate the new class? Just train the model from scratch again adding my dataset to the Pytorch wildlife training dataset? Is the finetuning module able to add classes so that I can finetune with my new data (should I add a subset of the Pytorch wildlife training data to it)?

Zhongqi Miao (zhongqi.miao@berkeley.edu)

2024-11-15 13:42:48

Hello Jennifer, I would suggest either finetuning the model using our fine-tuning module or treat the additonal class as "other" class in the current 35 categories. Please let us know hoe it works. Thanks

Sanjukta Moorthy (sanjukta@thesmc.group)

2024-11-12 08:16:30

Hello everyone, lovely to be part of this community! I am a researcher and contracted by IIED/REDAA to conduct a scoping study on the use and application of AI/tech to support locally-led restoration and conservation projects.

🌲 SURVEY on the use of AI and Tech to support locally-led restoration and conservation

We'd love your expertise and recommendations, especially with those who work in Asia and Africa. We've prepared a survey here which should take you just 15 minutes. https://5ttxcc4bnbv.typeform.com/to/kenOlU1D

We would also like to invite anyone who'd like to share their experiences with AI and technology to support projects to implement community-led, -based, or -driven projects, centring Indigenous expertise, valuing traditional knowledge, and supporting to resource locally-led approaches to the work.

🌿 INTERVIEW

If you would like to be interviewed, either instead of or alongside your survey response, please book a good time to speak here.

I will then follow up with more details (including our informed consent policy, the interview questions if that would help you prepare, logistical notes, etc). Please don't hesitate to reach out with any questions, and thank you so much! Look forward to hearing from you!

5ttxcc4bnbv.typeform.com

AI and Technology Scoping Study [REDAA Community and Public]

Turn data collection into an experience with Typeform. Create beautiful online forms, surveys, quizzes, and so much more. Try it for FREE.

Original URL: https://5ttxcc4bnbv.typeform.com/to/kenOlU1D

doodle.com

AI and Restoration REDAA Interview - Sanjukta Moorthy

Please book a time for a 1-hour interview to support a scoping study around the potential uses of AI and technology to support locally-led restoration in Africa and Asia. You will receive further details, including an informed consent form and the interview questions in advance to help you prepare if you requested it, via email.

Original URL: https://doodle.com/bp/sanjuktamoorthy/ai-and-restoration-redaa-interview

Timothy Keitt (tkeitt@utexas.edu)

2024-11-12 14:58:12

Hi Folks. Here is an opportunity that might be of interest to some in this group.

https://utexas.infoready4.com/#competitionDetail/1956602

I will also post a faculty cluster hire advert. in the near future.

💯 Tor Henrik Ulsted, Sara Beery

Timothy Keitt (tkeitt@utexas.edu)

2024-11-13 15:26:41

Assistant/Associate/Full Professor in Advances in Artificial Intelligence and Data Science for Environmental Systems

https://apply.interfolio.com/158908

This is part of a university-wide initiative to hire in AI and data sciences. This particular cluster is jointly between geology and integrative biology and is anticipated to fill three lines. Hires will be homed in one of the departments. Split appointments will only be considered for senior hires.

❤️ Sara Beery

😎 Jon Van Oast

Rebecca Kleinberger, PhD (rebecca.kleinberger@gmail.com)

2024-11-13 22:15:01

Hi everyone, I am new to this community and looking forward to connecting with many of you! I wanted to advertise a PhD Opportunity in Animal-Computer Interaction 🐕 🐈 🐓 🦜 🦍
—-
Join the INTERACT Animal Lab at Northeastern University (Boston, MA, USA) for a fully funded PhD position in Animal-Computer Interaction starting Fall 2025.

About Us: We leverage technology to improve animals' experiences, creating novel ethical enrichment systems to increase animals' control over their environments and enhance interspecies connectedness. Our lab’s recent projects include sonic enrichment for zoo animals🎼📢🐒, video-calling for parrots 🦜📺🦜, and ergonomic game design for pets 📱👈🐈. Visit our website for more info (http://interactanimallab.com)

Impact: Our work has been featured in the New York Times, the Washington Post, and over 1000 media outlets. We collaborate with various institutions, our local zoo and aquarium, husbandry professionals, and industry leaders to enhance animal life quality in zoos, homes, farms, and beyond.

The ideal candidate should be a creative technologist, well-versed in prototyping and with graduate level expertise in at least one area including engineering, software development, and/or animal studies.

Application deadline: December 1, 2024.

The program offers up to five years of funding, which includes a stipend, full tuition coverage for approved courses and a student health insurance plan. For more information and to apply, please visit:
https://camd.northeastern.edu/programs/interdisciplinary-design-and-media-phd/

Interested? Send your CV, portfolio, projects and any other relevant information to mailto:applications@interactanimallab.com

For questions or additional details, feel free to reach out r.kleinberger@northeastern.edu

interactanimallab.com

INTERACT Animal Lab

Research on animal technology, parrots video calling and touchscreens, games for animals, AI and animal, technology enrichment at the zoo, pet technology

Original URL: http://interactanimallab.com

🦜 Mitchell Rogers, Brian Geuther, Jason Holmberg (Wild Me)

Jake Burton (jake.burton@fauna-flora.org)

2024-11-15 06:28:49

Hi everyone! 🦧

WILDLABS is excited to announce we are once again teaming up with Dan Morris from Google's AI for Nature and Society program to run AI for Conservation Office Hours, where we bring together conservationists and data scientists from Google and beyond to help solve your AI problems.

If you are stuck on an AI or ML challenge in your conservation work, apply now for the chance to get 1:1 tailored advice. No matter what stage you’re at with AI or your previous expertise level, if you think this opportunity could be productive for you and your conservation project, apply now!

Read more details and how to apply: https://wildlabs.net/article/apply-now-ai-conservation-office-hours-2025

If you have any questions about any aspect of this opportunity or your project's eligibility, leave a comment on our Q&A thread or email me directly at jake.burton@wildlabs.net.

Deadline for applications is Monday 9 December 2024.

🙌 Nicolas Arrieta Larraza, Timm Haucke, Vanesa Reyes, Bernie Boscoe, Dan Morris, Chris Lange, Jason Holmberg (Wild Me), Sara Beery, Elizabeth Campolongo, Arthur Caillau, Mitch Fennell, Ragib Nihal, Sulagna Saha (Rasha), Carly Batist, Alex Rood, Malika Nisal Ratnayake, Alexander Robillard

🔥 Jason Holmberg (Wild Me), Sara Beery, Mitchell Rogers, Alexander Merdian-Tarko, Arthur Caillau, Hugo, Alexander Robillard

❤️ Jon Van Oast, Navodita Mathur, Talia Speaker, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Robin Sandfort, Alexander Robillard

Ian McFadden (ian.mcfadden@wsl.ch)

2024-11-15 09:23:16

📣 Job alert! 📣

Do you want to decode and map Earth's species interactions with AI while earning a PhD in London?

The project is funded by the CSC and open to citizens and permanent residents of China

Application: bit.ly/4i35cj5

🌍 Oisin Mac Aodha, Yuerou Tang

Isadora Nun (isadoranun@gmail.com)

2024-11-20 10:59:06

Hi everyone,
@Peter van Lunteren kindly invited me to join this group, and I'm thrilled to be here! My name is Isadora Nun, and I'm from Chile. While my background is in biotech engineering and AI/data science, rather than formal ecology and conservation studies, conservation has always been a deep passion of mine. I'm eager to learn from all of you and contribute my skills to any AI projects focused on conservation and/or ecology. I'm happy to help in any way I can!

Best,
Isadora

linkedin.com

Isadora Nun - NotCo | LinkedIn

Harvard MSE Data Scientist & Product Manager. 7+ years experience building and deploying impactful ML and AI solutions across diverse industries (AI-driven food tech, biotech, astronomy, revenue management). Expertise in machine learning, data analysis, and product strategy. Proven ability to deliver results and exceed expectations. · Experience: NotCo · Education: Harvard University · Location: San Francisco · 500+ connections on LinkedIn. View Isadora Nun’s profile on LinkedIn, a professional community of 1 billion members.

Original URL: http://linkedin.com/in/isadora-nun

👋 Tor Henrik Ulsted, Sara Si-Moussi, Jason Holmberg (Wild Me), Dan Morris, murilo, Chris Lange, Jon Van Oast, Elizabeth Campolongo, Shir Bar, Viktor Domazetoski, Eugene Galaxy, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Diego Kiedanski, Arthur Caillau, Mamoon, Alexander Merdian-Tarko, Alexander Robillard

😁 Peter van Lunteren, Alexander Robillard

👋:skin_tone_3: Alan Stenhouse

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2024-11-21 18:14:41

Columbia Ecology, Evolution and Environmental Biology (E3B) and Statistics are organizing a "young investigators" symposium on AI and Global Change on Feb 18, 2025!

Travel and lodging would be fully covered for all participants. Applications due Jan 1, 2025.

See attached — please do pass on!

(FYI, I have nothing to do with the organization of this event, but very thrilled to see it happening 😊)

🎉 Justin Kay, Jon Van Oast, Andrew Schulz, Seven Song, Eugene Galaxy, Eric Greenlee, Amara McCune, Sara Beery

💚 Millie Chapman

Eugene Galaxy (eugene@animaldetect.com)

2024-11-22 02:34:32

Would love to attend as a visitor but alas, we are in Europe. 😁

One day, one day...

❤️ Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

Abigail Oppong (abigoppong@gmail.com)

2024-11-22 04:11:56

🚀Announcing the launch of the Frugal AI Challenge! 🚀

🖥️ AI has the potential to address our most pressing environmental challenges... but AI models are becoming larger and more energy-intensive. The environmental impacts of these models are enormous, often overshadowing their potential environmental benefits. But it doesn't have to be this way! 🌍

🌱 Hugging Face, Data For Good, and Ministères Territoires Écologie Logement are launching a challenge to promote frugal AI models for addressing climate tasks.

🧮 Participants will be asked to solve one (or more) of the following three tasks: 📄 textual detection of climate misinformation, 🔎 classification of images of forest fires, and 🔊 sound detection of illegal deforestation.

🔋 Using real datasets, models will be evaluated on their performance and efficiency, taking into account real constraints and energy consumption. Each task will be ranked independently.

🏆 The winners of the challenge will be announced at the AI Action Summit 2025 in Paris, where heads of state, policymakers and AI experts will gather to discuss the future of AI.

🌍 Are you interested in the challenge? Want to join an international community on frugal AI? Register here: https://frugalaichallenge.org/

📢 The challenge will officially start in January 2025. In the meantime, we will share additional information on the tasks, datasets and evaluation metrics in the coming months. An in-person event is being organized on Tuesday, December 3, 2024 in Paris - stay tuned for more details!

2025 Frugal AI Challenge

Frugal AI Challenge

The goal of the Frugal AI Challenge is to encourage both academic and industry actors to keep efficiency in mind when deploying AI models. By tracking both energy consumption and performance for different Aasks, we can incentivize frugality in AI deployment while addressing real-world challenges.

Original URL: https://frugalaichallenge.org/

👍 Michael Bunsen, Owen Martin, Maddie Cusimano, Johannes Dollinger, Mitchell Rogers, Ilyass Moummad, murilo, Sara Beery, Jennifer

🐝 Michael Bunsen, Owen Martin, murilo

👍:skin_tone_3: Alan Stenhouse

Aakash Gupta (aakash@thinkevolveconsulting.com)

2024-11-25 02:26:38

believe the sign-up link is not working

Aakash Gupta (aakash@thinkevolveconsulting.com)

2024-11-25 02:26:54

Thread Reply:

Abigail Oppong (abigoppong@gmail.com)

2024-11-25 05:47:21

https://framaforms.org/2025-frugal-ai-summit-1732038008 - It works from my end. Maybe you could try again

Abigail Oppong (abigoppong@gmail.com)

2024-11-25 05:48:12

Thread Reply:

Aakash Gupta (aakash@thinkevolveconsulting.com)

2024-11-25 06:09:18

I checked with a Edge and a chrome browser. unfortunately still not working. Maybe related to the https signature?

Tanya Berger-Wolf (berger-wolf.1@osu.edu)

2024-11-25 18:49:31

SECOND WORKSHOP ON IMAGEOMICS @ AAAI 2025
https://sites.google.com/view/imageomics-aaai-25/
Discovering Biological Knowledge from Images using AI

Held as part of AAAI 2025
March 3, 2025, 9 am to 5 pm
Philadelphia, PA, United States

CALL FOR PARTICIPATION AND PAPERS
Symposium Key Dates:

Paper Submission Deadline: December 6, 2024. 11:59 PM AOE

Acceptance/Rejection Decision: December 16, 2024

Early Registration Deadline: December 19, 2024. 11:59 PM AOE

Camera-Ready Deadline: January 5, 2025, 11:59 PM AOE

sites.google.com

Imageomics-AAAI-25

Overview Imageomics is an emerging field of science that uses images, ranging from microscopic images of single-cell species to videos of charismatic megafauna such as giraffes and zebras, as the source of automatically extracted biological information, specifically traits, in order to gain

Original URL: https://sites.google.com/view/imageomics-aaai-25/

sites.google.com

Imageomics-AAAI-25

Original URL: https://sites.google.com/view/imageomics-aaai-25/#h.dov3jtf7xuyp

🙌 murilo, Sara Beery, Nathan Fox, Roberta Hunt, Chris Lange

❤️ murilo, Alan Stenhouse

Ștefan Istrate (stefan.istrate@gmail.com)

2024-11-26 10:09:44

[ MESSAGE TEXT EMPTY ]

Ștefan Istrate (https://aiforconservation.slack.com/team/UMCALAPBP)

To crop or not to crop? We're excited to share our new paper on the benefits of cropping with MegaDetector prior to species classification in camera trap images. One funky detail: we do not discard low confidence detections before running the classifier; instead, we train the classifier on "blank" bounding boxes as well! Read more in the IET Computer Vision journal (<a href="https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cvi2.12318">https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cvi2.12318</a>), where we also share lots of details about our complex data curation and model training decisions. Kudos to a whole team of dedicated nature enthusiasts!

😎 Jason Holmberg (Wild Me)

🔥 Jason Holmberg (Wild Me), Elizabeth Campolongo, Mitch Fennell, Kalindi Fonda, Tanya Berger-Wolf

🎉 Dan Morris, Alexander Merdian-Tarko, Alan Stenhouse

Lukas Picek (lukaspicek@gmail.com)

2024-11-26 10:13:24

Hi @Ștefan Istrate, looks like you did more less what we did in this ECCV workshop paper. https://arxiv.org/pdf/2410.12769 Interestingly, we got a little bit different results :)

Sara Beery (sara.beery@gmail.com)

2024-11-26 10:29:40

Cool work @Lukas Picek! Definitely interesting that the takeaways are different, lets look at the differences between the two papers (data, training, models, etc) and learn what drives the difference :)

Ștefan Istrate (stefan.istrate@gmail.com)

2024-11-26 10:30:59

Hi @Lukas Picek! We noticed your paper and it's great to be working in the same space. We also published our LILA split (see the "Supporting information" section) if you're curious to try your method on a much larger dataset.

Steve Haddock (haddock@mbari.org)

2024-12-09 14:40:29

👀 @Danelle Cline

Filippo Varini (fppvrn@gmail.com)

2024-11-26 14:57:36

Hi community! A general question. Besides the conventional, how do you keep up to new trends and research in AI fields like computer vision?
Do you have specific blogs, websites, twitter/linkedin pages or any other resource that you use to keep up with such a rapidly evolving field?

Brian Geuther (Brian.Geuther@gmail.com)

2024-11-26 15:06:07

This slack workspace 😎
But more seriously, I watch https://paperswithcode.com/
It's bias towards the big and popular benchmarks and tends to be a lot more sparse on biology/conservation.

paperswithcode.com

Papers with Code - The latest in Machine Learning

Papers With Code highlights trending Machine Learning research and the code to implement it.

Original URL: https://paperswithcode.com/

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2024-11-26 15:28:06

This is more of a technique than a specific location to look at. I try and do a periodic (exhaustive) literature review for the project I'm working on, maybe 2-3 times a year. I keep this in a document, written like the related work section of a paper.

The advantage of that is you have a good idea of how busy the space is and whether it's worth committing 6 months on a task that someone else is likely to do first - or has already done! (which is a win, because it saves me work). Also if you do this at the very start of the project, subsequent reviews are a lot faster because you're not looking through years of history.

I've often worked on industry/academic projects with a lifespan of 12-18 months with fixed research+code deliverables. In that arrangement, you don't have time to re-configure everything to take advantage of a new paper unless it's really groundbreaking. I prefer to structure things so that it's easy to swap in a new model, or train on a new dataset, later. That might mean committing to using a standard annotation format, or only models that HuggingFace provides, etc.

For conservation work, this Slack is about as good as it gets for CV/AI (and you can always ask). Otherwise I do skim through proceedings of the big conferences to see what's going on. You can also look at what models are added to HuggingFace or OpenMMLab to see what's new and trendy. For general tech news, HackerNews is still pretty good.

EDIT: I would add that if you're in grad school, bi-weekly paper clubs or similar arrangements can work well.

Within the astro community, a lot of people get daily arXiv digests and review it religiously, but I find that somewhat exhausting.

🙏 Filippo Varini, Thomas Radinger

Filippo Varini (fppvrn@gmail.com)

2024-11-26 15:28:31

Agreed - This community (and Wildlabs) IS such a great resource to keep up with AI in conservation.

Dan Morris (agentmorris@gmail.com)

2024-11-26 16:51:02

YMMV, but my two cents... outside of the very narrow domain of camera trap stuff, I don't try to keep up with AI in general, or more specifically, I don't do anything to keep up that I can't do while I'm also walking my dog.

But in a nod to the fact that I should have some idea what's happening in the industry, I do listen to episodes that sound interesting from a few AI podcasts, maybe a total of one episode a week across all of these podcasts (leaving plenty of time for podcasts about football):

https://podcasts.apple.com/us/podcast/the-twiml-ai-podcast-formerly-this-week-in-machine/id1116303051
https://podcasts.apple.com/us/podcast/last-week-in-ai/id1725552677
https://podcasts.apple.com/us/podcast/dataframed/id1336150688
https://podcasts.apple.com/us/podcast/data-skeptic/id890348705
https://podcasts.apple.com/us/podcast/latent-space-the-ai-engineer-podcast-practitioners/id1674008350
https://podcasts.apple.com/us/podcast/no-priors-artificial-intelligence-technology-startups/id1668002688

Sidebar: this is an amazing site for finding podcasts similar to podcasts you like:

https://rephonic.com/graph

Double-sidebar: this is a list of almost every podcast I listen to:

http://rockicon.net/wp/2018/02/02/podcasts/

Triple-sidebar: my LinkedIn feed is only people who post about conservation technology, except for this person, who I think does an amazing job reporting on the state of the art in LLMs without ever being boring:

https://www.linkedin.com/in/emollick/

I don't know him personally, but he makes me interested in LLMs. If you want to waste some time and be amazed by LLMs, read like his last 200 posts.

🙌:skin_tone_3: Alan Stenhouse

❤️ Bowen Zhang

Filippo Varini (fppvrn@gmail.com)

2024-11-27 04:36:39

🙏 🙏 thanks Dan!

Chris Lange (s2125675@ed.ac.uk)

2024-11-28 08:49:22

@Dan Morris you might like 'Criminal' if you're into true crime podcasts that mostly stay away from the gruesome stuff.

Much less focus on particular famous murders and more interviews with people involved in various aspects of crime and the justice system.

Filippo Varini (fppvrn@gmail.com)

2024-11-28 10:01:35

I like the serendipity of this thread !

Angela Zhu (azhu@umass.edu)

2024-12-05 20:50:44

I'm new to this space too and looking for resources!

It seems like "science-twitter" was pretty popular for people posting research material in a digestable manner (similar thing with the ai for conservation bluesky starterpack)

I'm currently applying to phd programs, so I've come across researchers in this space. I can give you a list. but I don't have a better way than following their work manually.

also @Chris Lange and others are doing cool work with species distribution models with iNaturalist and eBird!
-https://www.inaturalist.org/blog/84677-introducing-the-inaturalist-geomodel

iNaturalist

Introducing the iNaturalist Geomodel

Today, we’ve changed the Seen Nearby label on suggestions to Expected Nearby. The label comes from predictions made by the iNaturalist Geomodel that we’re introducing for the first time. What is the Geomodel? Most of you are familiar with the iNaturalist Computer Vision Model which takes an image as input and returns the most likely species based on visual similarity as an output. We train that model on a set of about 80 thousand species with enough data and update it monthly (we released version 2.7 today). The iNaturalist Geomodel takes a location as input and returns the most likely species at that location as output. Like the Computer Vision Model, it is a Deep Learning model trained on the same set of taxa and updated on the same monthly schedule. We developed and published the Geomodel in collaboration with the same Visipedia team that assisted with the iNaturalist Computer Vision Model. The map below shows Geomodel predictions for American Pika. The Geomodel is trained only o...

Original URL: https://www.inaturalist.org/blog/84677-introducing-the-inaturalist-geomodel

🙏 Chris Lange

Filippo Varini (fppvrn@gmail.com)

2024-12-06 05:09:44

Thank you Angela!

Ben Williams (ben.williams.20@ucl.ac.uk)

2025-01-01 14:44:15

I'm a month late to this, but I love the AI Explained channel on YouTube for keeping up to date with the frontier of AI more broadly. Its of course got a big LLM focus, but has been the best way for me to keep somewhat up to date on this a stuff that would otherwise have passed me by over the last two years
https://www.youtube.com/@aiexplained-official

YouTube

AI Explained

Covering the biggest news of the century - the arrival of smarter-than-human AI. The author of Simple Bench, exposing the remaining human-LLM reasoning gap. Join me at AI Insiders, with exclusive videos and a 1000+ network of AI enthusiasts and professionals: <https://www.patreon.com/AIExplained> Business Enquiries: <a href="mailto:aiexplained@outlook.com">aiexplained@outlook.com</a> Are you a high-flying consultant, CPA or finance professional? North America/UK-based, love keeping up-to-date AI developments, w/ a spare few hours/week? Keen to work together on something cool, yet potentially quite rewarding? Do get in touch, email above.

Original URL: https://www.youtube.com/@aiexplained-official

Filippo Varini (fppvrn@gmail.com)

2025-01-04 10:52:07

Thank you Ben! Looks like a great channel 🙂

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2024-11-26 15:09:59

For Master students in Germany: If you are looking for a thesis topic, I am offering multiple currently. For example: https://www.linkedin.com/posts/ahmadaamir_biodiversity-conservation-masterthesis-acti[…]131621548033-gij3?utm_source=share&utm_medium=member_desktop

linkedin.com

Aamir Ahmad on LinkedIn: #biodiversity #conservation #masterthesis #robotics #drones

🎓 Master Thesis Topic: Autonomous and Adaptive Drone Control based on Perceived Animal Behavior 🎓 Interested in doing your Master thesis with us at Flight…

Original URL: https://www.linkedin.com/posts/ahmadaamir_biodiversity-conservation-masterthesis-activity-7267266131621548033-gij3?utm_source=share&utm_medium=member_desktop

:zebra_face: Blair Costelloe, Timm Haucke, Filippo Varini

Talia Speaker (talia.speaker@wildlabs.net)

2024-11-27 11:37:02

Hi everyone, two conservation tech opportunities of potential interest from our crew at WILDLABS for funding and contributing to sector-wide research :persondoingcartwheel: Ping me or @Adrien Pajot for any questions!

💵 Apply for funding: The WILDABS Awards are back! This year, there are two grant categories of $50,000 and $10,000 to support up to 15 conservation technology projects. Deadline is February 13, 2025: https://wildlabs.net/funding-opportunity/wildlabs-awards-2025-apply-now

🌎 Help shape the sector: Have your say in the State of Conservation Technology 5-year trends report by taking the 2024 survey today! This research captures shared needs and priorities over time so that we can advance our tools together. If you're using or developing technologies for conservation, take 10 minutes to share your thoughts before Dec 31: https://wildlabs.net/article/take-2024-wildlabs-state-conservation-technology-survey

😎 Jason Holmberg (Wild Me), Arthur Caillau, Alex Rood

🎉 Jason Holmberg (Wild Me), Jon Van Oast, Carly Batist, Arthur Caillau, Filippo Varini, Eric Greenlee, Abigail Oppong, Eugene Galaxy, Alan Stenhouse

👀 Robin Sandfort

👏:skin_tone_5: Donna Auguste

Sunandan (sunchak@iu.edu)

2024-12-03 01:45:25

Announcement for:
ACM Conference on Computing and Sustainable Societies (COMPASS) - July 2025 - Toronto

ACM COMPASS 2025, hosted at the University of Toronto, invites interdisciplinary research on sustainability with a focus on how computing and moves toward sustainability are embedded in and shaped by specific local contexts—computing in place. Rooted in the broader discourse of sustainability but also open to questioning its prevailing narratives, COMPASS welcomes work that critically examines how technology intersects with social, environmental, and economic systems. This year's theme encourages a deeper, situated perspective on how technological practices emerge as solutions, but also contributors to the complexities they claim to address.

COMPASS draws from a wide array of disciplines: computer and information sciences, social sciences, critical geography, environmental studies, economics, climate studies, and engineering. We aim to foster a discussion that interrogates not only the potentials but also the limitations of technology-led interventions in sustainability. This includes research that explores power dynamics, inequality, and extraction—how technology might reinforce or disrupt existing systems. We encourage work from activists, worker advocates, and indigenous scholars, alongside those studying topics such as labour movements, climate adaptation, and the ethics of sustainability practices. COMPASS values research that brings forth the often-unseen political and material consequences of digital infrastructure in shaping futures.

COMPASS explicitly promotes multi-disciplinary and cross-disciplinary research and practice that address key challenges for sustainable societies including, but not limited to, equity, health, education, poverty, accessibility, conservation, climate change, energy, infrastructure and economic growth. We also welcome research on the ethics of technology, especially from a critical perspective, that explores limitations and concerns with technology-led solutions for sustainable societies. COMPASS also has a particular commitment to approaches and research that address the challenges faced by under-represented and marginalized communities.

The theme of the conference will be Computing in Place. We also invite submissions that speak to that theme and/or contribute to one or more of the following areas:

Systems and IoT for Sustainable Societies

HCI, Design and Critical Perspectives

AI, ML and Data Science for Sustainable Societies

Development, Economics and Policy

Environment, Sustainability and Climate Change

Technology, Media, and Social Practice

Link to CFP: https://compass.acm.org/call-for-papers/
Link to Call for Late Breaking Work: https://compass.acm.org/call-for-late-breaking-work/

❤️ Eric Greenlee

Jack Alston (jack.alston@bbc.co.uk)

2024-12-03 04:44:21

Hi there, anyone know of any good open source camera trap video datasets I could use to test a detection pipeline I'm developing?

michele volpi (michele.volpi@sdsc.ethz.ch)

2024-12-03 04:51:24

I’ve been using on and off some data from there: https://lila.science/datasets

Dan Morris (agentmorris@gmail.com)

2024-12-03 09:38:29

Public camera trap video data is pretty sparse... among datasets on LILA, Desert Lion Conservation Camera Traps includes ~200 videos:

https://lila.science/datasets/desert-lion-conservation-camera-traps/

...Lindenthal Camera Traps includes 775 videos:

https://lila.science/datasets/lindenthal-camera-traps/

...and this dataset includes gobs and gobs of thermal videos, but I don't think that's what you mean:

https://lila.science/datasets/new-zealand-wildlife-thermal-imaging/

Outside of LILA, the only public camera trap video dataset I'm aware of is:

https://timm.haucke.xyz/datasets/plittersdorf/

...which contains 221 videos.

✅ Otto Brookes

Valentin Gabeff (valentin.gabeff@protonmail.ch)

2024-12-03 12:25:21

PanAf from @Otto Brookes maybe? https://obrookes.github.io/panaf.github.io/

Otherwise there could be LoTE but I never worked with it https://lote-animal.github.io

This one too on wild pandas (see github for link to data) https://www.sciencedirect.com/science/article/pii/S157495412400339X?via%3Dihub - https://github.com/ijliu/PandaFormer

And this one has everything you need to test detection and tracking pipelines from fixed cameras, but it is not camera trap based: https://meerkat-dataset.github.io

🐼 Dan Morris

❤️ Otto Brookes

Otto Brookes (otto.brookes@bristol.ac.uk)

2024-12-10 07:03:13

@Jack Alston the PanAf dataset that @Valentin Gabeff mentioned has ~500 videos of great apes with frame-by-frame annotations for great apes (mainly chimpanzees), if that's of any use to you!

Ruben Remelgado (ruben.remelgado@gmail.com)

2024-12-05 08:32:15

We are looking for a Postdoctoral Researcher within the newly established group of Agro-Ecological Modeling, led by Professor Anna Cord. The lab is based at the University of Bonn, in Germany.

This is a full time-time position, with a minimum duration of 3 years (+3, upon positive evaluation). The selected candidate will develop their own research agenda within the framework of our research group, and will teach and mentor students in BSc and MSc programs.

At our lab, you are expected to use novel technologies for biodiversity monitoring and modeling, particularly passive acoustics. You will use these technologies to study habitat selection, occupancy, and species interactions in agroecosystems. To achieve this, you will have access to several long-term research farms, and other sites and locations affiliated to, and managed by, the University of Bonn. Furthermore, you will engage with other scientific activities in the lab, including complementary monitoring initiatives (e.g., using camera traps).

Applications are welcome until the 12th of January 2025. For more details, please click here.

Universität Bonn

Postdoctoral Researcher at the Professorship Agro-Ecological Modeling

Vollzeit oder Teilzeit full time or part time, befristet, temporary, EG 13, Kennziffer: 3.2/2024/86, Reference number: 3.2/2024/86

Original URL: https://www.uni-bonn.de/en/university/working-at-the-university/job-opportunities/vacancies-1/postdoctoral-researcher-at-the-professorship-agro-ecological-modeling

❤️ Timm Haucke, Sara Beery, Aamir Ahmad

Justin Kay (justinkay92@gmail.com)

2024-12-05 09:37:53

fyi for anyone working with YOLO object detectors the recent pypi versions seem to be compromised and will launch a crypto miner - careful when installing with pip! https://github.com/ultralytics/ultralytics/issues/18027

#18027 Discrepancy between what's in GitHub and what's been published to PyPI for v8.3.41

Bug Code in the published wheel 8.3.41 is not what's in GitHub and appears to invoke mining. Users of ultralytics who install 8.3.41 will unknowingly execute an xmrig miner. Examining the file utils/download.py, the contents in the published wheel are not representative of what's in GitHub: ``` $ wget -q <a href="https://files.pythonhosted.org/packages/d0/99/13d92174aa6a470d348a95e31164769f2cdf77838ea3c3e3fd476285777d/ultralytics-8.3.41-py3-none-any.whl">https://files.pythonhosted.org/packages/d0/99/13d92174aa6a470d348a95e31164769f2cdf77838ea3c3e3fd476285777d/ultralytics-8.3.41-py3-none-any.whl</a> $ unzip -q ultralytics-8.3.41-py3-none-any.whl $ cat ultralytics/utils/downloads.py | sed -n "278,301p" <code> <a href="https://private-user-images.githubusercontent.com/65414824/392681218-a67c53c4-9276-4bf8-bb67-165be06fbf8e.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.arXctcwoHnL1ioy1XooGw7gxjJbgtl_VYIWGPVfw28k">https://private-user-images.githubusercontent.com/65414824/392681218-a67c53c4-9276-4bf8-bb67-165be06fbf8e.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3MzM0MDk3NzQsIm5iZiI6MTczMzQwOTQ3NCwicGF0aCI6Ii82NTQxNDgyNC8zOTI2ODEyMTgtYTY3YzUzYzQtOTI3Ni00YmY4LWJiNjctMTY1YmUwNmZiZjhlLnBuZz9YLUFtei1BbGdvcml0aG09QVdTNC1ITUFDLVNIQTI1NiZYLUFtei1DcmVkZW50aWFsPUFLSUFWQ09EWUxTQTUzUFFLNFpBJTJGMjAyNDEyMDUlMkZ1cy1lYXN0LTElMkZzMyUyRmF3czRfcmVxdWVzdCZYLUFtei1EYXRlPTIwMjQxMjA1VDE0Mzc1NFomWC1BbXotRXhwaXJlcz0zMDAmWC1BbXotU2lnbmF0dXJlPTgwNzgwMDAxNjlhODExYzVkYThlMWZhZTEzMGVhMDBlYzEwYTUzNmZmNWYyMDM3YjEzOTBhMDZlYzFmYzVlNjEmWC1BbXotU2lnbmVkSGVhZGVycz1ob3N0In0.arXctcwoHnL1ioy1XooGw7gxjJbgtl_VYIWGPVfw28k</a> For the 8.3.41 tag, the content looks quite different: <a href="https://github.com/ultralytics/ultralytics/blob/v8.3.41/ultralytics/utils/downloads.py#L278">https://github.com/ultralytics/ultralytics/blob/v8.3.41/ultralytics/utils/downloads.py#L278</a> **Environment** Ultralytics 8.3.41 **Minimal Reproducible Example** </code> $ wget -q <a href="https://files.pythonhosted.org/packages/d0/99/13d92174aa6a470d348a95e31164769f2cdf77838ea3c3e3fd476285777d/ultralytics-8.3.41-py3-none-any.whl">https://files.pythonhosted.org/packages/d0/99/13d92174aa6a470d348a95e31164769f2cdf77838ea3c3e3fd476285777d/ultralytics-8.3.41-py3-none-any.whl</a> $ unzip -q ultralytics-8.3.41-py3-none-any.whl $ cat ultralytics/utils/downloads.py | sed -n "278,301p" ```

Labels

bug, dependencies, fixed

Comments

🤥 Malte Pedersen, Timm Haucke, Lukas Picek

😬 Cameron Trotter, Timm Haucke, Lukas Picek, Sara Beery, Yseult Hb, Arthur Caillau, Catherine Villeneuve, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

❗ Brian Geuther, Dante Wasmuht, Tarun, Jon Van Oast, Dan Morris, Cara Appel, Elizabeth Campolongo, Nico Lang, Jess Tam

🤨 Connor Levenson

😧 Chris Lange, Taiki Sakai - NOAA Affiliate, Mitchell Rogers, Valentin Gabeff, Jennifer, Sam Lapp

:gratitude_thank_you: Nils Blum-Oeste

Dante Wasmuht (dante@conservationxlabs.org)

2024-12-05 10:17:18

We might have been a victim at CXL. Thanks for flagging!!

❤️ Jon Van Oast, Justin Kay, Sara Beery

Lukas Picek (lukaspicek@gmail.com)

2024-12-05 10:21:23

Wow, that is huge! A long story short for the whole community in here: (i) All is good now, so you don’t have to care about the future installations, (ii) all versions before v8.3.41 are safe, (iii) check your server if you install it within the last 3 days.

👍 Justin Kay, Malte Pedersen, Sara Beery, Yseult Hb

Sara Beery (sara.beery@gmail.com)

2024-12-06 16:20:52

We've just officially released our INQUIRE benchmark ahead of presenting the work at NeurIPS next week!!

Read the thread on bluesky:
https://bsky.app/profile/sarameghanbeery.bsky.social/post/3lcnzhmzmcc2x

Or twitter: https://x.com/sarameghanbeery/status/1865125980769071140

Work led by @Eddie Vendrow and @Omiros Pantazis and supervised by me, @gvanhorn and @Oisin Mac Aodha, collaborating with @Kate Jones Gabe Brostow @Alex Shepard and input from iNat, Macauley Library, and the Visipedia community

bluesky

Sara Beery (https://bsky.app/profile/sarameghanbeery.bsky.social)

🎯 How can we empower scientific discovery in millions of nature photos? Introducing INQUIRE: A benchmark testing if AI vision-language models can help scientists find biodiversity patterns- from disease symptoms to rare behaviors- hidden in vast image collections. Thread👇🧵

Original URL: https://bsky.app/profile/sarameghanbeery.bsky.social/post/3lcnzhmzmcc2x

X (formerly Twitter)

A multi-state occupancy model to non-invasively monitor visible signs of wildlife health with camera traps that accounts for image quality - PubMed

Amazing work led by my student @EdwardVendrow and @panom1 from @ucl on enabling scientists to interactively probe massive-scale data on @inaturalist via text If you'll be at #NeurIPS2024 next week come learn more at our poster! Dec. 13th, 11am-2pm, East Exhibit Hall A-C #4510

Original URL: https://x.com/sarameghanbeery/status/1865125980769071140

📣 Oisin Mac Aodha, murilo, Rupa Kurinchi-Vendhan, Nico Lang, Valentin Gabeff, Louis Freeland-Haynes, Sepand Dyanatkar

🙌 murilo, Shir Bar, Rupa Kurinchi-Vendhan, Viktor Domazetoski, Elizabeth Campolongo, Nico Lang, Robin Sandfort, Eric Greenlee, Aakash Gupta, Robin Zbinden, Roberta Hunt, Angela Zhu, Mamoon, Andrew Schulz

🎉 Jon Van Oast, Avi Sundaresan, David Russell, Meredith Palmer, Nico Lang, murilo, Tessa Rhinehart, charlotte, Alan Stenhouse

🙌:skin_tone_2: Cara Appel

Ronan Wallace (rwallace@macalester.edu)

2024-12-07 01:13:15

Hi everyone! My name is Ronan Wallace, and I am carrying out a Fulbright research grant in Nepal using computer vision and ethnography for environmental conservation and heritage preservation. I use photogrammetry and remote sensing in the Himalayas to create 3D reconstructions of vulnerable heritage sites and climate disaster impacts.
Recently, I have been researching Neural Radiance Fields (NeRFs) as a method for 3D and 4D modeling. I’ve been thinking a lot about how this might be used in the conservation space. I wanted to introduce myself briefly, and see if anyone here has used NeRFs for their research. If you (or someone you know) have experience with NeRFs, I would love to connect and hear about your work. 😊

👋:skin_tone_3: Jess Tam

👋 Owen Martin, mustafa angı, Alessandra Vidal Meza, Sara Beery, Daniel Yang, Marion Richardot

😎 Jon Van Oast, Marion Richardot, Aakash Gupta

🙌 Maximiliane Jousse

Jess Tam (j.tam@unsw.edu.au)

2024-12-07 07:20:01

Hi Ronan! I'm working on camera traps species ID and recently joined a lab that does a lot of work with NeRF & 3D systhesis. Hence I'm also very interested in this stuff! Happy to connect and chat more about some possible projects, tho I don't actually know much about view synthesis (yet?)

Sara Beery (sara.beery@gmail.com)

2024-12-08 17:23:36

@Levi Cai has a labmate that works on this!

Levi Cai (levi@levicai.com)

2024-12-08 17:24:18

@Daniel Yang 🙂

❤️ Sara Beery, Daniel Yang

Daniel Yang (daniel.yang17@gmail.com)

2024-12-09 15:40:38

Hey Ronan! We’ve been using NeRFs / Gaussian Splatting to model some underwater coral reef data! Most naively, it’s just a fancier way to visualize scenes and related data in 3D/4D. From a conservation point of view, I think most of the questions remain the same between traditional photogrammetry and NeRFs / GS (e.g. biomass estimation, change detection, motion modeling)

Some of the newer approaches are quite fast / computationally efficient though which opens up potential applications in resource limited environments though (e.g. robots, drones, AR/VR, etc.).

Happy to chat more!

🪸 Sara Beery, Jess Tam, Alan Stenhouse

Robin Sandfort (sandfort@wildbiologie.org)

2024-12-14 12:55:06

Hi Ronan, we use Gaussian Splatting quite a lot. Great advance in this field!

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 09:32:16

Hi everyone, quick question, probably far fetched, but are there any detection models that can give the proportion of an animal body visible in an image?

Diego Kiedanski (dkiedanski@tryolabs.com)

2024-12-10 09:34:28

size of the animal in the image / size of the animal IRL (❓, probably not)
size of the animal in the image / total image area (✅

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 09:37:59

More like proportion of body visible, i.e. a model that detects that part of the animal's body is missing in the picture and it tells you e.g. 80% of animal visible in image (i.e. 20% of the body is not in the picture).

Oisin Mac Aodha (macaodha@caltech.edu)

2024-12-10 09:39:12

Perhaps methods like 3D Fauna or SAOR could be adapted to do this?

https://arxiv.org/abs/2401.02400
https://arxiv.org/abs/2303.13514

@Mehmet Aygun what do you think?

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 09:40:42

sounds like it! because it would have to recognize, body parts or at least body sections.

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 09:42:04

This is for coyotes but it is literally a dog body shape, and there are plenty of german shepperd pictures to use to train it, if the dataset you guys use doesnt have a dog shaped animal

Dante Wasmuht (dante@conservationxlabs.org)

2024-12-10 09:42:52

A little less sophisticated than Oisin's suggestions but running a simple pose detector might do an ok job as well?

👍 Oisin Mac Aodha, Sara Beery, Tiziana Gelmi Candusso

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 09:43:34

We are using the information on a model for animal skin disease, the proportion of body visible contributes to a confidence parameter on how much the observer is able to identify mange occurrence. https://pubmed.ncbi.nlm.nih.gov/33942308/

PubMed

Camera traps are an increasingly popular tool to monitor wildlife distributions. However, traditional analytical approaches to camera trap data are difficult to apply to visible wildlife characteristics in single images, such as infection status. Several parasites produce visible signs of infection …

Original URL: https://pubmed.ncbi.nlm.nih.gov/33942308/

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 09:44:55

A pose detector, that could work too

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 09:45:27

Thread Reply:

Blair Costelloe (blaircostelloe@gmail.com)

2024-12-10 09:49:03

I would try a pose detector - body keypoints that aren't visible should have low confidence scores, and you could then tell what proportion of keypoints are not seen (and which ones, which may be helpful if some body parts are more informative for recognizing mange than others)

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 09:55:41

I can see the pose detector recognizing the four legs, the tail and the head, that's all it needs. because the missing body part pictures are generally because the animal is entering or leaving the frame

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 10:02:43

someone went all in with dog poses here, so cool https://learnopencv.com/animal-pose-estimation/

LearnOpenCV – Learn OpenCV, PyTorch, Keras, Tensorflow with code, & tutorials

Animal Pose Estimation: Fine-tuning YOLOv8 Pose Models

Animal pose estimation can be performed by fine-tuning pre-trained YOLOv8 pose models for analyzing animal postures, and performing specific keypoint analysis.

Original URL: https://learnopencv.com/animal-pose-estimation/

Oisin Mac Aodha (macaodha@caltech.edu)

2024-12-10 10:03:39

There is also this:
https://barc.is.tue.mpg.de/

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 10:05:54

This is so cool Oisin! I remember Silvia Zuffi's presentation on 3D animals, this would be very useful

Dante Wasmuht (dante@conservationxlabs.org)

2024-12-10 10:06:04

haven't used it in a while but DeepLabCut used to be the go-to tool for animal pose estimation (not 3D models etc.) (at least in the Neuroscience domain, but I'm biased there): https://arxiv.org/pdf/2203.07436v4

Dante Wasmuht (dante@conservationxlabs.org)

2024-12-10 10:06:32

https://www.mackenziemathislab.org/deeplabcut

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 10:11:40

yes, Dante, something like that then one just gets which dots IDs are in the frame and which arent, and you get a proportion of body visible.

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 10:13:38

Thread Reply:

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 10:18:17

This one is nice too. https://sites.google.com/view/wldo/home

sites.google.com

Who Left the Dogs Out?

Who Left the Dogs Out? 3D Animal Reconstruction with Expectation Maximization in the Loop ECCV 2020

Original URL: https://sites.google.com/view/wldo/home

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 10:23:51

ok it seems it is very doable, but there is no one-liner code ready to use for this, like I was wishing, lol. I will tag the 200 images I have left manually, but will leave the idea here in case someone gets into it. I have around 10 thousand coyote pictures with proportion of body visible tagged if needed.

😍 Dante Wasmuht

Dante Wasmuht (dante@conservationxlabs.org)

2024-12-10 10:24:58

yes please!

🎉 Tiziana Gelmi Candusso

Oisin Mac Aodha (macaodha@caltech.edu)

2024-12-10 10:27:07

https://huggingface.co/spaces/Kyle-Liz/3DFauna_demo

❤️ Tiziana Gelmi Candusso

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 10:32:01

I see four legs, a head and a tail in that reconstruction

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 10:33:59

Thread Reply:

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 10:34:05

works nicely

Oisin Mac Aodha (macaodha@caltech.edu)

2024-12-10 10:34:07

You would have to do an extra step to compute how much of the 3D model is visible in the image. Should be possible, but requires some extra 3D geometry know how. Their full code base might do this.

🙌 Tiziana Gelmi Candusso

Dante Wasmuht (dante@conservationxlabs.org)

2024-12-10 10:34:37

crazy how good those have become 🤯

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 10:37:59

Is this the github repository for that 3d fauna model https://github.com/3DAnimals/3DAnimals?

3DAnimals/3DAnimals

A machine learning framework for reconstructing articulated 3D animals from images

Stars

Language

Python

👍 Oisin Mac Aodha

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 10:41:50

awesome! this is definitely doable. Thank you so much for getting into this quest with me and all the resources shared!

Tiziana Gelmi Candusso (tiziana.gelmi@gmail.com)

2024-12-10 10:42:45

I will keep reading about this, and see if I manage to run anything on my end to get these values!

Dan Morris (agentmorris@gmail.com)

2024-12-12 12:17:57

New dataset on LILA from the Snapshot Safari program:

https://lila.science/datasets/snapshot-safari-2024-expansion/

This represents recent seasons of (almost) all of the Snapshot Safari projects (including Snapshot Serengeti), totaling over 4M images, making it one of the largest camera trap datasets on LILA. Thanks as always to Sarah Huebner for providing well-curated data that makes it easy to get everything ready for LILA!

Also thanks to the Snapshot Safari program for using a dedicated label for lion cubs, which makes it easy for me to pull a gratuitous lion cub picture for this post (attached).

Using that as an example to tie everything together, you can find that image at the following URLs, in the LILA buckets on GCP/AWS/Azure, respectively:

https://storage.googleapis.com/public-datasets-lila/snapshot-safari-2024-expansion/SER/SER_S12/K05/K05_R2/SER_S12_K05_R2_IMAG2362.JPG

http://us-west-2.opendata.source.coop.s3.amazonaws.com/agentmorris/lila-wildlife/snapshot-safari-20[…]ER_S12/K05/K05_R2/SER_S12_K05_R2_IMAG2362.JPG

https://lilawildlife.blob.core.windows.net/lila-wildlife/snapshot-safari-2024-expansion/SER/SER_S12/K05/K05_R2/SER_S12_K05_R2_IMAG2362.JPG

🎉 Timm Haucke, murilo, Dante Wasmuht, Jon Van Oast, Arthur Caillau, Kalindi Fonda, Mélisande Teng, Judith Dekkers, Angela Zhu, Jason Holmberg (Wild Me), Lasha Otarashvili, Mitch Fennell, Alan Stenhouse

🙌 murilo, Dante Wasmuht, Malte Pedersen, Angela Zhu, Jason Holmberg (Wild Me), Lasha Otarashvili, Alexander Merdian-Tarko

❤️ Matthias Zuerl, Sara Beery, Atul Ingle, Cara Appel, Jon Van Oast, Viktor Domazetoski, takatomi-k, Mitchell Rogers, Angela Zhu, Jennifer, Jason Holmberg (Wild Me), Lasha Otarashvili, Valentin Gabeff, Sepand Dyanatkar

🦁 Mitchell Rogers, Jason Holmberg (Wild Me), Robin Zbinden

✅ Valentin Gabeff

Jon Van Oast (jon@wildme.org)

2024-12-12 13:16:47

amazing! thanks everyone.

Dan Morris (agentmorris@gmail.com)

2024-12-13 22:41:06

I also posted MD results (including the post-repeat-detection-elimination results that are more suitable for classifier training) to the LILA MD results page:

https://lila.science/megadetector-results-for-camera-trap-datasets/

🙌:skin_tone_3: Alan Stenhouse

Josh von Nonn (jvonnonn@usgs.gov)

2024-12-15 15:10:55

Hi all, a brief intro - I'm a geographer at the Western Geographic Science Center/USGS, and my research involves very-high resolution drone data (RGB, MS, lidar) for habitat modeling, invasive species monitoring, and geomorphic change detection. I'm also co-chair on the DOI open source geospatial software community of practice and we are always looking for presenters. I'm just starting to dig into AI/ML to classify drone data using pytorch (torchgeo+torchvision) with Mask R-CNN. Looking forward to being a part of this server. Cheers! https://www.usgs.gov/staff-profiles/josh-von-nonn

usgs.gov

Josh von Nonn | U.S. Geological Survey

Josh von Nonn is a Geographer with the Western Geographic Science Center in Moffett Field, CA.

Original URL: https://www.usgs.gov/staff-profiles/josh-von-nonn

👋 Timm Haucke, Dan Morris, Elizabeth Campolongo, Isadora Nun, Jenna Kline

Raj (therajsmile@gmail.com)

2024-12-16 01:22:22

Hi, just wondering if I can get some suggestions on tracking multiple fish (around 5-10 fish at the moment, but may increase in future) from recorded videos. Does anyone know any open-source free applications/tool/software where this task can be performed with higher accuracy?
Thanks in advance.

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2024-12-16 01:31:43

I would try OpenCV, assuming you're able to label the individual fish in frame #1. There are a bunch of methods you can use and they all use the same interface, so you can swap out algorithms easily. Most will struggle if the object leaves the frame, but that's a hard problem to solve generally.

https://pyimagesearch.com/2018/07/30/opencv-object-tracking/

OpenCV's docs can be a bit... obscure sometimes. But everything should work in Python as well as C++. This is a good route before you try doing anything more involved. Ping here if you need any help though, it should only be a few lines of Python if you have the starting boxes for the fish.

Accuracy of tracking (how reliably the tracker can keep locked onto an object) depends on the algorithm and how good the video is (static camera will be easier than handheld, is it well lit, etc).

Raj (therajsmile@gmail.com)

2024-12-16 02:04:55

Thanks a lot Josh, it seems openCV’s trackers cannot handle high FPS, and, closely moving to each other.
I haven’t tried on my recent fish videos but earlier I tried. It didn’t work well and fish frequently left the marked boxes and also switched IDs when nearby often.
Without much of accuracy, I think the results are not good enough to compare in my videos. Thanks a lot once again.

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2024-12-16 02:08:17

Next in terms of effort would be a framework like MMTracking https://github.com/open-mmlab/mmtracking

I'd be surprised if the frame rate (on its own) would have much effect, that just determines how far an object moves between frames at a given speed. But closely spaced objects I can see might be an issue, especially if they cross paths a lot.

open-mmlab/mmtracking

OpenMMLab Video Perception Toolbox. It supports Video Object Detection (VID), Multiple Object Tracking (MOT), Single Object Tracking (SOT), Video Instance Segmentation (VIS) with a unified framework.

Website

<https://mmtracking.readthedocs.io/en/latest/>

Stars

3601

Raj (therajsmile@gmail.com)

2024-12-16 02:10:22

I would try this and see how it goes in my fish videos settings. Thanks a lot once again Josh.

murilo (murilogustineli@gatech.edu)

2024-12-16 02:45:15

Adding on what Josh said, I believe YOLO might be a viable option. YOLO from ultralytics supports high FPS and real-time tracking of objects. I’ve used YOLO on identifying marine animals, including different types of fish. You could also easily integrate YOLO with roboflow if you intend to annotate your dataset and fine-tune a custom model. https://docs.ultralytics.com/modes/track/

docs.ultralytics.com

Track

Discover efficient, flexible, and customizable multi-object tracking with Ultralytics YOLO. Learn to track real-time video streams with ease.

Original URL: https://docs.ultralytics.com/modes/track/

Aakash Gupta (aakash@thinkevolveconsulting.com)

2024-12-16 03:40:36

I would go with Yolov8 or later versions for the simplicity of starting out. Other options that we have used are FairMOT, deepsort and paddledetection,.

Arthur Caillau (arthur@caillau.me)

2024-12-16 04:17:21

Hey there!
I worked on such a project before and here is how we did it: https://www.earthtoolsmaker.org/projects/wild_salmon_migration_monitoring/
TL;DR: Yolo + bytetrack
You can also play with the model here: https://www.earthtoolsmaker.org/spaces/wild_salmon_migration_monitoring/

earthtoolsmaker.org

Wild Salmon Migration Monitoring | EarthToolsMaker

The project monitors wild salmon migration to ensure the number passing through meets state regulations, addressing threats from human activities like fisheries and dams.

Original URL: https://www.earthtoolsmaker.org/projects/wild_salmon_migration_monitoring/

Raj (therajsmile@gmail.com)

2024-12-16 04:47:26

Thank you so much Murilo @murilo for your suggestion. I will try to look at the documentation you shared. Thanks a ton.

👍 murilo

Raj (therajsmile@gmail.com)

2024-12-16 04:48:54

Thanks Aakash @Aakash Gupta for input, I never heard some of them other than YOLO, but will definitely try to look at if it’s fits with my specifications of analysis. Thanks for your suggestions.

Raj (therajsmile@gmail.com)

2024-12-16 04:50:57

Thanks for your sharing your experience Arthur @Arthur Caillau. It seems YOLO with some other model as you suggested would be a good choice to look at. I will go through the documentation you shared. Thanks a lot.

Aakash Gupta (aakash@thinkevolveconsulting.com)

2024-12-16 05:12:18

@Raj
Some of my notes on that topic. Hope its useful.

https://thinkevolve.notion.site/Object-tracking-and-Re-identification-099001252a5444d8afa12e7d58f1ab5f?pvs=4

Think Evolve Consulting on Notion

Object tracking and Re-identification | Notion

Original URL: https://thinkevolve.notion.site/Object-tracking-and-Re-identification-099001252a5444d8afa12e7d58f1ab5f?pvs=4

David Russell (davidrussell327@gmail.com)

2024-12-16 09:41:51

There's an open-source platform called VIAME that is designed for fishery applications. I believe it has some tracking capabilities.

Dan Morris (agentmorris@gmail.com)

2024-12-16 11:15:27

I'm interpreting the question as being about both detection (where are the fish in this image?) and tracking (which fish in this frame corresponds to which fish in the previous frame?)... for the detection piece, I try to keep track of publicly-available models for fish detection here:

https://github.com/agentmorris/agentmorrispublic/blob/main/fish-datasets.md#publicly-available-models-for-fish-detection

AFAIK the only one that includes an explicit tracking component is SharkTrack:

https://github.com/filippovarini/sharktrack

...which uses the Ultralytics tracking tools:

https://docs.ultralytics.com/modes/track/

...which should be straightforward to use with almost any object detector, although integration may be easier if your detection model is based on an Ultralytics detector (YOLOv5, YOLOv8, YOLOv9, YOLOv10, YOLOv11).

🙌 murilo

Raj (therajsmile@gmail.com)

2024-12-16 18:38:06

Thank you Aakash @Aakash Gupta for your notes, it’s great help.

Raj (therajsmile@gmail.com)

2024-12-16 18:39:19

Thanks David @David Russell for this suggestion, I never heard of before so I will explore it with my own datasets. Thanks a ton.

Raj (therajsmile@gmail.com)

2024-12-16 18:42:28

Thanks Dan @Dan Morris, and yes, detection followed by tracking but more of interest is tracking. I will explore it for sure and find if it’s fits with my own datasets. Thanks a ton.

Chaim (c.elchik123@gmail.com)

2024-12-19 09:48:34

I did something very similar for my master thesis this year, I used yolov8 but v11 would work even better. I have linked my thesis below if your interested to read it as it may be relevant. If you have any questions feel free to aks them. https://scripties.uba.uva.nl/search?id=record_54895

scripties.uba.uva.nl

Scripties - Bibliotheek - Universiteit van Amsterdam

Scripties van studenten van de Universiteit van Amsterdam.

Original URL: https://scripties.uba.uva.nl/search?id=record_54895

Raj (therajsmile@gmail.com)

2024-12-19 19:50:03

Thanks a lot Chaim @Chaim, I will surely read it through and take notes what could be best for my own datasets. Thanks a ton.

Hannah Yin (hannah.yin@rice.edu)

2025-01-15 12:29:33

'lil late to the party, but idtracker.ai is another generally accessible and well-documented open-source tool for controlled lab settings. I just do not know how it compares to other tools that use YOLO though.

Raj (therajsmile@gmail.com)

2025-01-15 18:38:31

Thanks a lot Hannah @Hannah Yin, for this suggestion. I am trying some tools suggested here, but the object ID is not so accurate and requires lot of manual correction. Though, trying to get to work with that fits with our settings. Thanks a ton.

Hannah Yin (hannah.yin@rice.edu)

2025-01-15 23:08:38

Great, glad it's another option for you. I've always found these tools to be finnicky at times and to benefit from trial and error--you're not alone on that front!

Raj (therajsmile@gmail.com)

2025-01-16 01:54:49

Thanks Hannah for your kind words. Yes, it seems trial and error method since there is no perfect tool for such tasks. Glad to hear that people are kind enough to help me:) Thanks.

Beni Beeri (beni@reearth.ai)

2024-12-17 03:34:21

Hello everyone!

👋 Elizabeth Campolongo, Ilyass Moummad

John Martinsson (john.martinsson@ri.se)

2024-12-17 06:19:32

Hey everyone!

I've just launched my webpage, johnmartinsson.org, where I share my research on machine listening for bioacoustics and ecosystem monitoring.

I am especially happy about the little clickable animal icons in the top. Visit the site and click on a few to make a soundscape! 🎶🐦🐸🐳🐺🎶

I work on developing machine learning methods for audio (machine listening), and I make sure that the development is motivated by the ability to apply these methods in bioacoustics for ecosystem monitoring. I'm aiming to update the publications with comments on their conservation context in the near future.

Hope you enjoy your holidays! 🎅

🙌 Ilyass Moummad, Viktor Domazetoski, Yseult Hb, Alexander Merdian-Tarko, Robin Zbinden, Dan Morris, Elizabeth Campolongo, Jenna Kline, Owen Martin, Isadora Nun, Matthias Zuerl, Taiki Sakai - NOAA Affiliate, Magali Frauendorf, Clare Price, Alexander Robillard

🦜 Alan Stenhouse, hannah miro, Alexander Robillard

Kalindi Fonda (kalindi.fonda@gmail.com)

2024-12-17 16:13:34

oh too cute 🐳

Benjamin Kellenberger (benjamin.kellenberger@yale.edu)

2024-12-17 12:06:02

Hi all,

I am (co-) advertising 5 PhD projects in AI for ecology & environment! 🤖🦋

Together with wonderful collaborators from UCL, Queen Mary University of London, and EPFL Switzerland, we are looking for talented PhD candidates interested in the intersection of AI/Machine/Deep Learning with Earth observation/Remote sensing, genomics, and ecology.
Come and work in fabulous London on AI for good! 😃

Positions are part of two UK doctoral training programmes, TREES and AI-INTERVENE. Application deadlines are coming up in January!

PhD 1: https://tinyurl.com/uav-wildlife
PhD 2: https://tinyurl.com/forest-GNN
PhD 3: https://tinyurl.com/genome-sweep
PhD 4: https://tinyurl.com/interaction-cv
PhD 5: https://tinyurl.com/interaction-map

Happy to answer questions – and wish you lovely festivities! 🎄😊

🔊 Oisin Mac Aodha, Robin Zbinden, Shubhi Sharma, Angela Zhu, Sara Beery, Jason Holmberg (Wild Me)

🎉 Isadora Nun, Taiki Sakai - NOAA Affiliate, Jon Van Oast, Cameron Trotter, Nina van Tiel, Dan Morris, Catherine, Justin Kay, murilo, Yseult Hb, Robin Zbinden, Shubhi Sharma, Angela Zhu, Sara Beery, Malte Pedersen, Viktor Domazetoski, Jess Tam, Jason Holmberg (Wild Me), Nico Lang, Meredith Palmer, Jonathan Sauder, Millie Chapman, Alan Stenhouse, Ben Williams

😎 Jon Van Oast, Shubhi Sharma, Angela Zhu, Sara Beery, Jason Holmberg (Wild Me)

🙌 murilo, Robin Zbinden, Shubhi Sharma, Jason Holmberg (Wild Me), Johannes Dollinger

👍 Alsh, Ben Weinstein, Valentin Gabeff, Mamoon, Marzi

Magali Frauendorf (magali.frauendorf@slu.se)

2024-12-18 06:03:11

📣 Job alert 🚨
Interested in camera traps, remote sensing and AI? 🌿 🌍 📡 🦌

Our Conservation Technology group is growing!
We are hiring a fully funded (2 years) postdoc working on the interface between camera traps, satellite remote sensing, and AI. The postdoc will mostly work in the Biodiversa+ funded project Big Picture in a work package focusing on linking camera-trap data with satellite remote sensing to study plant phenology, snow cover and different habitats, and linking these to wildlife data at a European scale.

👉 See link for more info and how to apply
📝 Apply now - deadline is January 19, 2025

SLU.SE

Jobs and vacancies at SLU. | slu.se

Jobs and vacancies at SLU. Read more about each job by clicking the job title. Please, follow the instructions closely when applying.

Original URL: https://www.slu.se/en/about-slu/work-at-slu/jobs-vacancies/?rmpage=job&rmjob=11842&rmlang=UK

👍 Piotr Tynecki, Gaspard Dussert, Robin Zbinden, Justin Kay, Sara Beery, Aleksis Pirinen, Timon

❤️ Cara Appel, Stephanie Roilo, Aleksis Pirinen, Maximiliane Jousse

👍:skin_tone_3: Alan Stenhouse

Dan Stowell (dan.stowell@naturalis.nl)

2024-12-18 13:27:56

Academic job! Come and work with us as "Assistant Professor, Data Science & AI" https://tiu.nu/22594 - friendly department, lots of stimulating colleagues with diverse research interests, Dutch work-life balance ... 🙂

🎉 Jon Van Oast, Alan Stenhouse, Elena Sierra

Ben Weinstein (benweinstein2010@gmail.com)

2024-12-18 13:39:07

Hi everyone, a reminder to introduce yourself as you join! I see lots of new people.

🎉 Jon Van Oast, Taiki Sakai - NOAA Affiliate, Alexander Robillard

Chris Cosma (chris.cosma@consbio.org)

2024-12-18 14:08:55

Hi everyone, I joined a while ago and have been learning a lot from your posts. I’m an ecologist at the Conservation Biology Institute, focusing mostly on insect conservation in urban and agricultural areas. I’m developing several projects to monitor insects and plant-pollinator interactions with AI-equipped camera traps. In a current project I am trying to roughly quantify nocturnal flying insect abundance from time-lapse camera trap images using machine learning to count “moth streaks” from the images. If anyone has any good suggestions for that, I’d love to hear them! Cheers

Taiki Sakai - NOAA Affiliate (taiki.sakai@noaa.gov)

2024-12-18 14:33:23

@Michael Bunsen am I remembering correctly that you do insects? Also maybe its time for another PNW meetup sometime soon...

🙌 Chris Cosma

Michael Bunsen (notbot@gmail.com)

2025-07-24 16:29:15

Hi @Taiki Sakai - NOAA Affiliate thanks for the mention here! Chris and I were able to connect this year! And I definitely am up for another PNW meetup any time.

Arthur Caillau (arthur@caillau.me)

2024-12-18 14:22:07

Hi everyone, I also joined some time ago and never introduced myself. I love the AI for Conservation community and I have learnt a lot from the different posts.

I am a Software/ML Engineer. If you have some interesting problems in the conservation space where you think AI can be leveraged, feel free to reach out!
I am also part of the Ai2/Earthranger crew, fighting poaching on a global scale 🐘 🦏 :giraffe_face:
Cheers!

👋 Alexander Merdian-Tarko

Zhongqi Miao (zhongqi.miao@berkeley.edu)

2024-12-18 14:52:03

Hello everyone, today we are happy to announce Project SPARROW (Solar-Powered Acoustic and Remote Recording Observation Watch), our self sustained solar-powered wildlife monitoring solution using satellite for camera trapping and bioacoustics! Our goal is to create a monitoring system that collects and processes the data through edge computing and send the results to the cloud using satellite. There are three major advantages: 1) it does not rely on GSM coverage because of the use of satellites; 2) every component in this system (including the edge computers, cameras, audio recorders, and satellites) is powered by solar panels, so we don’t need to change batteries; and 3) data collected are preprocessed with Pytorch-Wildlife using MegaDetector and other classification models and are directly send to the cloud. There will be no manual data collection needed! Everything we use for this solution is based on off-the-shelf devices with minimum hardware engineering, and we are starting a physical deployment in the Amazon Rainforest in January. We will release more technical details in the next couple of months! We will keep you posted and feel free to ask us any questions! Happy holidays!

linkedin.com

Juan M. Lavista Ferres on LinkedIn: #aiforgood #biodiversity #globalconservation

I am proud to introduce SPARROW (Solar-Powered Acoustic and Remote Recording Observation Watch), developed by my team at the AI for Good Lab in collaboration…

Original URL: https://www.linkedin.com/posts/jlavista_aiforgood-biodiversity-globalconservation-activity-7275201548425576448-49Qo/?utm_source=share&utm_medium=member_desktop

👀 Timm Haucke, Cara Appel, Eric Greenlee, Taiki Sakai - NOAA Affiliate, Meredith Palmer, Diego Kiedanski, Elizabeth Campolongo, Jose Ruiz-Munoz, Owen Martin, Jason Holmberg (Wild Me)

🙌 Carly Batist, Jennifer, Diego Kiedanski, Jose Ruiz-Munoz, Owen Martin, Adrien Fontvielle, Jenna Kline, Jason Holmberg (Wild Me), Avi Sundaresan

🎉 Eugene Galaxy, Jason Holmberg (Wild Me), Meredith Palmer

Juan Sebastián Cañas (jscanass@gmail.com)

2024-12-19 07:17:05

Hello everyone! Here are 26 new PhD projects in the UK focused on AI for Biodiversity. These projects are part of the new AI-INTERVENE program hosted by UCL and the University of Reading in partnership with the Natural History Museum, ZSL, Kew Gardens, and UK Centre for Ecology & Hydrology https://www.linkedin.com/feed/update/urn:li:activity:7275160249471270913/

linkedin.com

Kate Jones on LinkedIn: AI-INTERVENE

🎓 Join the AI-INTERVENE PhD Programme – Applications Now Open! 🌍🤖 Are you passionate about using artificial intelligence (AI) to help with biodiversity and…

Original URL: https://www.linkedin.com/feed/update/urn:li:activity:7275160249471270913/

🔊 Oisin Mac Aodha, Timm Haucke, Gustavo Perez, Abigail Oppong, Sara Beery

👏 Isadora Nun, Timm Haucke, Carly Batist, Dan Morris, Maddie Cusimano, Aude Vuilli, Mamoon, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

👏:skin_tone_3: Alan Stenhouse

Carly Batist (cbatist@gradcenter.cuny.edu)

2024-12-19 12:59:39

Website for the 2025 IUCN World Conservation Congress (Oct. 9-15 in Abu Dhabi) is up!

Exhibition booth reservations are open here (first come, first serve): https://iucncongress2025.org/exhibition

Quick glance at pricing - for a booth, IUCN members: $4100, non-members: $5400

Call for proposals open here (deadline-Feb .7): https://iucncongress2025.org/forum#call

Registration opens in March: https://iucncongress2025.org/participate/registration

iucncongress2025.org

Forum | iucncongress2025.org

The Forum is the largest knowledge platform on the science, practices and innovation of conservation and sustainable development. Through thematic sessions, the Forum addresses the economic, social, scientific and technical aspects of a wide range of issues related to nature conservation, from wildlife to oceans, protected areas, sustainable businesses, climate change or human rights.

Original URL: https://iucncongress2025.org/forum#call

Kate Rakelly (krakelly@gmail.com)

2024-12-19 14:01:18

Hi everyone! I am an AI researcher interested in opportunities to apply machine learning to conservation efforts - actively looking for work in this space! My background is in image recognition, reinforcement learning, and robotics. My website: https://katerakelly.github.io/

👍 Ștefan Istrate, Kalindi Fonda, Owen Martin, Ben Weinstein, Aamir Ahmad

👋 Isadora Nun, Owen Martin, Patrick Beukema

👍:skin_tone_3: Alan Stenhouse

Sankalpa Ghose (sg@anart.ai)

2024-12-19 16:37:27

hi everyone! love this slack group & just want to invite everyone to this conference:

🌟 AI for Animals 2025 Conference - Get Early Bird Tickets Now

📅 March 1-2, 2025
📍 UC Berkeley
🎓Call for speakers / sessions

aiforanimals.org

AI for Animals

Advancing the development of AI to create a better future for all Animals.

Original URL: https://www.aiforanimals.org

🙌 Owen Martin, murilo, Malte Pedersen, Jennifer, Jason Holmberg (Wild Me)

🎉 murilo, Sara Beery, Jason Holmberg (Wild Me)

🙌:skin_tone_5: Prabath Gunawardane

🙌:skin_tone_3: Alan Stenhouse

Brittany Aguilar (baguilar@schmidtsciences.org)

2024-12-20 13:12:42

Not sure if you have contacted the Earth Species Project but they would be interested Im sure

🌱 Sankalpa Ghose

Sankalpa Ghose (sg@anart.ai)

2024-12-21 05:46:45

thanks @Brittany Aguilar! yes we know Katie & some of their team, & Jane Lawton spoke at our last conference last summer at LSE. hoping to have their involvement

if you are ever interested to learn more about our efforts, including in biomedical sciences (from NAMs to science communication like Asimov Press), please do reach out. & hope you can attend conference if you are in the area then :)

Brooke Goodman (Brg226@pitt.edu)

2024-12-20 13:21:41

Hi all! The Schoodic Institute (nonprofit partner of Acadia National Park) is looking for a Science Research Fellow for 2025! The position is for folks 0-3 years out from their undergrad degree who have not yet completed any graduate school. For the last two years the position has focused on bird conservation in wetlands using bioacousitcs and AI! I was a fellow in 2023 and I'm happy to answer any questions about the position 🙂 feel free to share! https://schoodicinstitute.org/wp-content/uploads/2024/12/SciResearch_GeroFellow_2025-SI.pdf

❤️ Sara Beery, Pat Lyon

Piotr Tynecki (piotr@tynecki.pl)

2024-12-23 07:34:50

Good morning everyone. I am seeking community members who specialize in radar data, particularly in applications related to radar ornithology 🐦 for Wildlife Hazard Management (including long-range precision measurements and behavior pattern analysis) and marine research 🌊 (such as bird migration, detection and classification). Could you please share more information about who is working on these topics within the community? Thanks!

Piotr Tynecki (piotr@tynecki.pl)

2024-12-23 07:39:47

@Adriaan Dokter I would be happy to hear more from you about ^

❓ Do we have any dedicated channel or chat with GloBAM - A. Farnsworth, K. Horton, C. Nilsson and other rock stars in that domain?

✅ Vincent Lostanlen

Dan Sheldon (sheldon@cs.umass.edu)

2024-12-24 10:46:55

Hi Piotr, I think the relevant folks in this slack are myself, @Adriaan Dokter, @Benjamin Van Doren, @Subhransu Maji. We mostly work with data from the various national networks of dedicated weather radars, e.g., NEXRAD in the US or OPERA in Europe. Some people also use smaller dedicated radars for ornithological applications--they can give more precise measurements but cover a much smaller area. Do you know which you are interested in?

👍 Piotr Tynecki

Dan Morris (agentmorris@gmail.com)

2024-12-29 22:40:04

This is somewhere between a "new dataset announcement" and a "random hobby project", but... for not-entirely-silly reasons, I'm releasing ~20k images and ~5k videos from camera traps around my house:

https://lila.science/datasets/seattleish-camera-traps/

The motivations for releasing this data are:

Images of humans are almost entirely missing from public camera trap data, which makes even basic debugging and cursory validation of camera trap AI systems difficult. I am the only human who appears in this dataset (I removed everyone else manually), so this isn't going to help anyone train models with a human class (other than a hypothetical MegaDanDetector), but I hope it helps people at least debug.

Images from consumer-grade camera traps are almost entirely missing from public camera trap data, and these cameras - as you might expect - can be extra-tricky for AI systems, both because the image quality is lower and because they are placed in random places by non-professionals. If we want to know how systems work on this type of camera, we need some images.

As per recent threads here on this Slack, even as video becomes an increasingly important medium for camera trap research, video is almost entirely missing from public camera trap data.

I want to encourage other camera trap folks to contribute as well, so we can eventually break the logjam around images of humans by contributing images as a community where we have strong consent to release the images. As I mention in the dataset description, I'd like to eventually have a reason to change the name of this dataset from "Seattle(ish) Camera Traps" to "Camera Trappers in Camera Traps", so if you're interested in contributing, email me!

I want to win the game of making the Internet look at pictures of my dog (and, with surprising frequency, other random dogs).

😎 Jason Holmberg (Wild Me), Nate Harada, Kalindi Fonda, Timm Haucke, Ed Miller, Jon Van Oast, murilo, Alan Stenhouse

🔥 Jason Holmberg (Wild Me), Nate Harada, Ben Weinstein, Kalindi Fonda, Rose Wenxin Zhao, Timm Haucke, Mitchell Rogers, Aakash Gupta, Ritwik, Cameron Trotter, Robin Sandfort, Lukas Drees, Sara Beery, murilo, Mitch Fennell, Malte Pedersen, Clare Price, Angela Zhu, Tessa Rhinehart, Alexander Dungate, Will Armstrong

🙌 Nate Harada, Justin Kay, Kalindi Fonda, Timm Haucke, Peter van Lunteren, Jeegn Dani, murilo, Alexander Dungate, Sepand Dyanatkar

🐕 Nate Harada, Kalindi Fonda, Robin Zbinden, Vincent Lostanlen, Klára Janoušková, Meredith Palmer, Jennifer, Timm Haucke, Cameron Trotter, murilo, Tessa Rhinehart

⛄ Kalindi Fonda, Timm Haucke

🐶 Klára Janoušková, Timm Haucke, Eric Cunningham

Ben Weinstein (benweinstein2010@gmail.com)

2024-12-30 00:10:38

@Dan Morris, when was it -10 C at your house?

Dan Morris (agentmorris@gmail.com)

2024-12-30 10:16:28

I mention on the page that the absolute dates and times are totally meaningless, so one time that it definitely wasn't is 2010/08/16. But based on some other context, it was shortly after Dec 17, 2022 (the previous camera check date), and based on timeanddate.com, the coldest day shortly after that in Bellevue was Thursday Dec 22 (18F somewhere in Bellevue, so 14F somewhere else in Bellevue is plausible), so that's my guess.

I chose the thumbnail because dog, but it's a fun coincidence that it might be the coldest day in Bellevue over the entire time I've had camera traps out.

✅ Ben Weinstein

Matthias Zuerl (matthias.zuerl@fau.de)

2024-12-30 13:06:20

I propose a competition for the best MegaDanDetector 😁
Thank you for that release! 🙏

👍 Dan Morris, Jason Holmberg (Wild Me), Robin Sandfort

😂 Mitchell Rogers, Jason Holmberg (Wild Me), Ed Miller, Alan Stenhouse

💕 Jon Van Oast

Ariel Chamberlain (ariel9chamberlain@gmail.com)

2024-12-31 12:19:03

Hello everyone, and Happy (almost) New Year!
Does anyone have any recommendations for comprehensive and scientific plant databases, with details like sun/water needs, habitat ranges, pollution tolerance/remediation, ease of establishment, hardiness, good companion plants etc.?
I am specifically looking for a reference that includes many plants in India, but as a plant lover, I would love to know of any others too!

👀 Aakash Gupta

Ariel Chamberlain (ariel9chamberlain@gmail.com)

2024-12-31 12:24:51

for context, there are the attributes I am looking for:

Sun Needs (hours/day)

pH range

Height Range (meters)

Ok in flood? (yes/no)

Ok in drought? (yes/no)

Eaten by grazing animals (e.g. cows & goats)? (yes/no)

Food consumed by humans? (yes/no)

Ease of establishment in the first 1-2 years? (hard, medium, easy)

Root mass: amount of stabilization created (high, medium, low)

Invasive in India? (yes/no)

Native to India? (yes/no)

Pollution Tolerance? (yes/no)

Pollution Remediation? (yes/no)

Hardiness (high, medium, low)

If specific to only part of Uttar Pradesg describe where it grows best.

Companion plants that it grows especially well with

Jess Tam (j.tam@unsw.edu.au)

2025-01-01 02:52:33

Not sure about India, but Austraits might be a good start for Australian species: https://austraits.org/

:gratitude_thank_you: Ariel Chamberlain

Aakash Gupta (aakash@thinkevolveconsulting.com)

2025-01-09 05:59:14

@Ariel Chamberlain - have you found any relevant databases? I am working on a soil fertility problem. If that is of interest to you. We can connect offline or via DM

Sara Si-Moussi (sara.si-moussi@univ-grenoble-alpes.fr)

2025-01-11 06:16:49

Some of those traits can be found in the TRY database, indirectly. For the native/invasive there are some databases like EASIN that could be of help.

Hannah Yin (hannah.yin@rice.edu)

2025-01-15 12:02:43

Could be a stretch, but there's a botanical illustrator based in Bangalore who has drawn plants in the local area and who might have connections with botanists there.
https://www.nirupa-rao.com/profile

Andrew Lensen (andrew.lensen@vuw.ac.nz)

2025-01-05 20:35:30

Hi all! We have two PhD scholarships available to help in the re-identification of Kākā (a boisterous New Zealand parrot) in Wellington, New Zealand. One in Ecology and one in AI 🙂 Please spread the word!

https://www.wgtn.ac.nz/scholarships/current/phd-scholarship-urban-kaka-ecology-and-conservation
https://www.wgtn.ac.nz/scholarships/current/phd-scholarship-recognising-taonga-with-ai-facial-recognition-for-kaka-conservation-management

:flag_nz: Mitchell Rogers, Anastasios Angelopoulos, Dan Morris, Seven Song, Ben McEwen, Chris Lange, Timm Haucke, Brian Geuther, Sara Beery

🙌 Mitchell Rogers, Anastasios Angelopoulos, Oisin Mac Aodha, Malte Pedersen, Abigail Oppong, Timm Haucke

🐦 Mitchell Rogers, Anastasios Angelopoulos, Jess Tam, Rose Wenxin Zhao, Benjamin Hoffman, Yseult Hb

❤️ Anselm Bradford, Anastasios Angelopoulos, Tiffany Vlaar, Alan Stenhouse

👀 Valerie

Sara Beery (sara.beery@gmail.com)

2025-01-09 13:04:39

Hey AI for Conservation Community!

This is a Hail Mary, but I'm here in Pasadena with the CV4Ecology school doing our best to handle the situation with the fires. We are looking for any ideas and/or options for somewhere we could all relocate, at least until the situation dies down. There are 27 of us, and we could pick up and drive anywhere within 4-5 hours of LA. Does anyone have any ideas?

❤️ Oisin Mac Aodha, Antón Álvarez, Jon Van Oast, Shir Bar, Catherine, Malte Pedersen, murilo, Lauren Chronister, Pat Lyon, Declan, Justin Kay, Levi Cai, Rupa Kurinchi-Vendhan, Roberta Hunt, Gracie Ermi, Ruby Krasnow, Mitchell Rogers, Cara Appel, Toshio Matsuoka, Jess Tam, Burooj Ghani, Robin Zbinden, Olof Mogren, Chris Lange, Carly Batist, Ștefan Istrate, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Gustavo Perez, Angela Zhu, Kate Rakelly, Juan Sebastián Cañas, Johannes Dollinger, Aleksis Pirinen, Alan Stenhouse, Kishore Panaganti, Sepand Dyanatkar

🙏 Jennifer, Yuyan Chen, Burak Ekim, Lukas Picek, Catherine Villeneuve, Valentin Gabeff, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Juan Sebastián Cañas, Malika Nisal Ratnayake, Filippo Varini, Kishore Panaganti

Ben Weinstein (benweinstein2010@gmail.com)

2025-01-09 13:18:16

@Ruth Oliver is there a large conference center at UC Santa Barbara?

Justin Kitzes (justin.kitzes@pitt.edu)

2025-01-09 13:21:03

What are the housing needs @Sara Beery?

Sara Beery (sara.beery@gmail.com)

2025-01-09 13:28:24

There are 27 of us, but we are pretty scrappy so happy to pile in

Sara Beery (sara.beery@gmail.com)

2025-01-09 13:28:52

I think I could even possibly get some tents/sleeping bags from the Caltech Y

Ruth Oliver (rutholiver@bren.ucsb.edu)

2025-01-09 13:51:29

@Ben Weinstein we put up the bat signal to folks at NCEAS/UCSB. hoping they can help!

Matt Weldy (matthewjweldy@gmail.com)

2025-01-09 13:52:15

Sara, it is probably not a great solution... But, Airbnb announced an offer of temporary housing for those impacted by the wildfires.

Kaitlyn Gaynor (kaitlyn.gaynor@ubc.ca)

2025-01-09 14:12:09

I was gonna suggest NCEAS too, but looks like Ruth is on it! lmk if I can nudge too tho

❤️ Sara Beery

Sara Beery (sara.beery@gmail.com)

2025-01-09 15:22:30

We're heading up to UCSB for a few days!! Thank you for the help everyone

🙌 Kaitlyn Gaynor, Mitchell Rogers, Ruth Oliver, Ben Weinstein, Carly Batist, murilo, Tessa Rhinehart, Howard L Frederick, Aleksis Pirinen, Levi Cai

👍 Justin Kitzes

🙏 Lily Xu - OLD ACCOUNT, DO NOT MESSAGE

💚 Angela Zhu, Alan Stenhouse

🙌:skin_tone_3: Alan Stenhouse

Burooj Ghani (buroojghani@gmail.com)

2025-01-10 03:24:05

Please stay safe!

❤️ Sara Beery

Toshio Matsuoka (toshiomatsuoka9@gmail.com)

2025-01-09 13:45:12

Hi everyone! I’ve been a member of this group for a while and have really enjoyed learning about the amazing work being shared here. I’m currently working as a fisheries biologist, studying how the foodscape of freshwater drift-feeding fishes varies across space and time in the western United States and Canada. While my current research doesn’t directly involve AI, I’ve previously worked on projects using computer vision, and I’m eager to stay connected with advancements in this field. I find the intersection of AI and conservation incredibly exciting and inspiring!

🐠 Shir Bar

🐟 Mitchell Rogers, Dan Morris

Lauren Chronister (lmc150@pitt.edu)

2025-01-10 18:31:48

Hi all! I'm a graduate student in Justin Kitzes' lab at the University of Pittsburgh. I and some other folks in the lab are organizing a symposium for the 2025 American Ornithological Society meeting August 11-15 in St. Louis. We'd like to center this symposium around the topic of acoustic individual identification of birds (like mark-recapture but using acoustic features of the vocalizations to recognize individuals) and are looking to invite speakers from both automated methods and more traditional perspectives. I'd like to put a feeler out here to see if anyone might be interested or know someone who would be interested in speaking in this symposium?

❤️ Leonardo Viotti, Angela Zhu, Timm Haucke, Camille Desjonquères, Mélisande Teng, Juan Sebastián Cañas, Sara Beery, Tessa Rhinehart, Pat Lyon, Mark Cartwright, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, ash

Nathan Jacobs (jacobsn@wustl.edu)

2025-01-10 20:31:05

That sounds super fun! I'm based in St. Louis. Let me know if you are looking for local folks.

My PhD student Srikumar has done some bird / species classification / mapping work:

https://vishu26.github.io/taxabind/index.html

https://openaccess.thecvf.com/content/WACV2024/html/Sastry_BirdSAT_Cross-View_Contrasti[…]s_for_Bird_Species_Classification_and_WACV_2024_paper.html

https://arxiv.org/abs/2312.08334

vishu26.github.io

TaxaBind

TaxaBind: Multimodal Ecological Models

Original URL: https://vishu26.github.io/taxabind/index.html

arXiv.org

LD-SDM: Language-Driven Hierarchical Species Distribution Modeling

We focus on the problem of species distribution modeling using global-scale presence-only data. Most previous studies have mapped the range of a given species using geographical and environmental features alone. To capture a stronger implicit relationship between species, we encode the taxonomic hierarchy of species using a large language model. This enables range mapping for any taxonomic rank and unseen species without additional supervision. Further, we propose a novel proximity-aware evaluation metric that enables evaluating species distribution models using any pixel-level representation of ground-truth species range map. The proposed metric penalizes the predictions of a model based on its proximity to the ground truth. We describe the effectiveness of our model by systematically evaluating on the task of species range prediction, zero-shot prediction and geo-feature regression against the state-of-the-art. Results show our model outperforms the strong baselines when trained with a variety of multi-label learning losses.

Original URL: https://arxiv.org/abs/2312.08334

Carly Batist (cbatist@gradcenter.cuny.edu)

2025-01-12 20:23:17

@Elly Knight !

❤️ Elly Knight

Elly Knight (ecknight@ualberta.ca)

2025-01-13 07:38:48

Thanks @Carly Batist!

ash (a.h.blum@gmail.com)

2025-03-29 12:32:22

Hi! I'm working in bird vocal signature/individual identity and would love to participate if I'm not too late 🙂

Aleksis Pirinen (aleksis.pirinen@ri.se)

2025-01-13 05:05:56

Hello AI-for-conservation community!

Just wanted to inform you of a new Nordics-centred research network that may be relevant to many of you: Climate AI Nordics (https://climateainordics.com/). You are most welcome to join the network (https://climateainordics.com/join/), where 70+ researchers (mainly from the Nordics, but others are welcome) have already joined.

We're also arranging a physical 1-day workshop in Gothenburg, May 13th 2025 (https://climateainordics.com/events/2025-nordic-workshop). Welcome to join the event as well!

Cheers!

climateainordics.com

Climate AI Nordics

Original URL: https://climateainordics.com/

Climate AI Nordics

Join Climate AI Nordics

Original URL: https://climateainordics.com/join/

❤️ Olof Mogren, Filip Dorm, Oisin Mac Aodha, John Martinsson, Tiffany Vlaar, Nico Lang, Jon Van Oast, Catherine Breen, stefano puliti

Olof Mogren (olof.mogren@ri.se)

2025-01-13 05:14:16

Keynotes by Oisin Mac Aodha and Christian Igel. Full info on web page above!

🙌 Aleksis Pirinen, Oisin Mac Aodha, John Martinsson

Suzanne Stathatos (suzanne.stathatos@gmail.com)

2025-01-13 16:21:36

If anyone works in (or wants to work in) the intersection of computer science (AI, systems, networking.. they’re open to a lot) and deep-sea-related research, this sounds like a pretty nifty opportunity
https://www.linkedin.com/posts/twilighted_madeira-activity-7280920551332728834-Nnff?utm_source=share&utm_medium=member_desktop

linkedin.com

TWILIGHTED on LinkedIn: #madeira

Apply to join our Impossible Things Workshop! We have a list of impossible challenges related to deep-sea research, and we want your help to solve them!…

Original URL: https://www.linkedin.com/posts/twilighted_madeira-activity-7280920551332728834-Nnff?utm_source=share&utm_medium=member_desktop

🙌 murilo, Levi Cai

❤️ murilo, Jonathan Sauder, Ștefan Istrate, Filippo Varini, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Jennifer

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2025-01-14 19:11:53

Climate Change AI is excited to announce our next workshop on “Tackling Climate Change**** with Machine Learning” to be hosted at ICLR 2025 (one of the major ML conferences) in Singapore. There are several ways to participate:

Submit a paper or proposal on your work at the intersection of climate change and machine learning (submissions due Jan 31, 2025)

Attend the workshop in Singapore on Apr 27 or 28 (TBD)

We hope to provide limited travel assistance (by application) for some workshop attendees. We also have an interactive Q&A on our Community Platform where you have an opportunity to ask questions regarding the mentorship program and workshop submissions. You can also refer to previous informational webinars on our workshop series via the Climate Change AI YouTube channel.

For more details, see our workshop website at: https://www.climatechange.ai/events/iclr2025

**** Biodiversity loss & nature totally count! 💚🐘

❤️ Suzanne Stathatos, Mark Goldwater, David Russell, Jess Tam, Eric Greenlee, Samantha Biegel, Oisin Mac Aodha, Juan Sebastián Cañas, Robin Zbinden, Johannes Dollinger, Sarthak Arora, Viktor Domazetoski, Justin Kay, Elena Sierra, John Martinsson, Chris Llorca, Alan Stenhouse, Evan Eskew, Sepand Dyanatkar

Lily Xu - OLD ACCOUNT, DO NOT MESSAGE (lily_xu@g.harvard.edu)

2025-01-14 19:11:59

To stay up to date on Climate Change AI’s upcoming workshops and activities, please consider signing up for our newsletter: https://www.climatechange.ai/newsletter

ICLR 2025 CCAI Workshop Team:
Konstantin Klemmer (Microsoft Research)
Melissa Chapman (UC Santa Barbara & ETH Zurich)
Lily Xu (University of Oxford)
Ivan Poon (National University of Singapore)
Mélisande Teng (Université de Montréal & Mila)
Patrick Emami (NREL)
Yoshua Bengio (Université de Montréal, Mila, CIFAR & IVADO)

Climate Change AI

Climate Change AI - Newsletter

Climate Change AI newsletters

Original URL: https://www.climatechange.ai/newsletter

Magali Frauendorf (magali.frauendorf@slu.se)

2025-01-15 06:59:07

A few days left to apply! Please spread the word to everyone who might be interested...

Magali Frauendorf (https://aiforconservation.slack.com/team/U04V7LTTNSG)

📣 Job alert 🚨 Interested in camera traps, remote sensing and AI? 🌿 🌍 📡 🦌 Our Conservation Technology group is growing! We are hiring a fully funded (2 years) postdoc working on the interface between camera traps, satellite remote sensing, and AI. The postdoc will mostly work in the Biodiversa+ funded project Big Picture in a work package focusing on linking camera-trap data with satellite remote sensing to study plant phenology, snow cover and different habitats, and linking these to wildlife data at a European scale. 👉 See <https://www.slu.se/en/about-slu/work-at-slu/jobs-vacancies/?rmpage=job&rmjob=11842&rmlang=UK|link >for more info and how to apply 📝 Apply now - deadline is January 19, 2025

👍 Piotr Tynecki, Antón Álvarez

Lukas Picek (lukaspicek@gmail.com)

2025-01-15 09:09:59

🚨 We are excited to announce that the 12th Workshop on Fine-Grained Visual Categorization (FGVC) has been once again accepted to CVPR and will soon accept paper submissions.

📅 Key Deadlines:

Proceedings Track: March 5, 2025

Non-Archival Track: March 31, 2025

Nectar Track: April 30, 2025

📍 Learn more on our website: Submission Page.

We look forward to your contributions! 🚀

👏 Malte Pedersen, Oisin Mac Aodha, Justin Kay, Shir Bar, Gustavo Perez, Sara Beery, Subhransu Maji, murilo, Klára Janoušková

❤️ Nico Lang, murilo, Oisin Mac Aodha, Justin Kay, Ben Weinstein, Suzanne Stathatos, Thijs van der Plas, Gustavo Perez, Sara Beery, Jess Tam, Lily Xu - OLD ACCOUNT, DO NOT MESSAGE, Klára Janoušková, Avi Sundaresan, Abigail Oppong, Ilyass Moummad, Sepand Dyanatkar

🙌 murilo, Oisin Mac Aodha, Justin Kay, Kostas Papafitsoros, Gustavo Perez, Sara Beery, Klára Janoušková, Ilyass Moummad

Hannah Yin (hannah.yin@rice.edu)

2025-01-15 12:36:47

Hi! I'm in the middle of reading up on reinforcement learning and want to ask: are there commonly used or recommended computing resources/environments for small-scale projects in conservation that use RL? By small-scale, let's say completed within 6 months. If some cost modest $ that's fine--I'd rather know about it in the off-chance these resources can be funneled into a budget.

Ben Weinstein (benweinstein2010@gmail.com)

2025-01-15 12:55:13

no personal experience, but have heard others use https://sfcompute.com/

sfcompute.com

SF Compute | H100s with 3.2tb/s InfiniBand

The San Francisco Compute Company: Large, low-cost GPU clusters you can rent by the hour, for pre-training, inference, and more. Get H100s with 3.2tb/s InfiniBand, parallel storage, fast networking, and priority support.

Original URL: https://sfcompute.com/

Hannah Yin (hannah.yin@rice.edu)

2025-01-15 17:01:42

Thanks!

Kate Rakelly (krakelly@gmail.com)

2025-01-22 12:48:03

@Hannah Yin I’d be curious to hear how RL is being applied in conservation!

Hannah Yin (hannah.yin@rice.edu)

2025-01-22 23:55:23

@Kate Rakelly Happy to chat! This publication discusses RL along with other ML approaches and was fairly helpful for me in getting a big picture perspective. To me it really goes to show how much we don't know about RL applications in conservation 😉
I believe some of the authors on the pub are in this Slack group haha.

👀 Kate Rakelly

Navodita Mathur (navoditamathur1998@gmail.com)

2025-01-15 16:02:22

Hi, I joined a while ago and have really enjoyed the work being shared here. I am an AI researcher with experience in computer vision. My interests are in developing multimodal models for biodiversity monitoring and research, more specifically how integrating domain specific data could help in OOD and domain shifts. Looking forward to learning from and contributing to this community!

Ben Weinstein (benweinstein2010@gmail.com)

2025-01-15 17:47:31

We are writing some grants about this, can you tell me more about your work?

Ben Weinstein (benweinstein2010@gmail.com)

2025-01-15 17:49:06

from https://github.com/Navoditamathur, "iNaturalist Species Classification: Working on an LLM-powered species classification pipeline with a question-answer format in Streamlit to provide detailed information about identified species using agents." I'd be interested in seeing the source and goals aroudn this.

Navodita Mathur

A Tech Enthusiast with a focus on Computer Vision and Deep learning

URL

<https://navoditamathur.github.io/>

Repositories

👀 Alan Stenhouse

Navodita Mathur (navoditamathur1998@gmail.com)

2025-01-15 19:28:52

Thank you for your interest! I’ve trained the classification model and am still working on the question-answering component to refine the framework. I’ll update the repository soon and share the progress when it’s ready.

Navodita Mathur (navoditamathur1998@gmail.com)

2025-01-15 19:53:26

The goal was to create a tool that not only identifies species but also provides qa answers about them, such as their behavior and habitat. For now, I’m using Wikipedia as the source for this information to get the system up and running.

Devis Tuia (devis.tuia@epfl.ch)

2025-01-16 03:18:40

Earthvision 2025 will happen at CVPR!
Consider sending your wonderful research involving AI and remote sensing. Applications in ecology are super welcome!

see you there, deadline on March 3, 2025.

More info here: https://www.grss-ieee.org/events/earthvision-2025/

GRSS-IEEE

EARTHVISION 2025 - GRSS-IEEE

EARTHVISION 2025 June 11/12th, Nashville, TN, USA in conjuction with the Computer Vision and Pattern Recognition (CVPR) 2025 Conference Aims and Scope Important Dates People Sponsors Submission Program CVPR 2025 […]

Est. reading time

4 minutes

Original URL: https://www.grss-ieee.org/events/earthvision-2025/

🌍 Oisin Mac Aodha, Robin Zbinden, Mitchell Rogers, Jess Tam, Konstantin Klemmer, Owen Martin, Sara Beery, Risa Shinoda, Andy Viet Huynh, Alan Stenhouse, Thijs van der Plas, Aleksis Pirinen, Valerie, Sepand Dyanatkar

🙌 Valentin Gabeff, Andy Viet Huynh

Carly Batist (cbatist@gradcenter.cuny.edu)

2025-01-16 11:16:54

Good funding opp!
Google.org Accelerator: Generative AI open call

This global open call is looking for nonprofits, civic entities, academic institutions, and social enterprises to apply to participate in the Generative AI Accelerator. Selected organizations will receive a share of $30M alongside 6 months of structured support with access to dedicated pro bono assistance from Google employees, technical training, and Google Cloud credits to build socially impactful gen AI-powered solutions across Google.org’s three focus areas: Knowledge, Skills, and Learning; Scientific Advancement; and Resilient Communities.

https://impactchallenge.withgoogle.com/genaiaccelerator/about
Due Feb. 10

Google.org Accelerator: Generative AI open call

@Googleorg Accelerator: Generative AI open call

@Googleorg is launching a $30M global open call to help to help organizations use gen AI to accelerate their social impact. Learn more ➡️ <a href="http://g.co/ImpactChallenge/GenAIAccelerator">g.co/ImpactChallenge/GenAIAccelerator</a>

Original URL: https://impactchallenge.withgoogle.com/genaiaccelerator/about

👀 Antón Álvarez, Eric Greenlee, Ben Weinstein, Alan Stenhouse

✔️ Jon Van Oast, mimi, Sepand Dyanatkar

Ben Weinstein (benweinstein2010@gmail.com)

2025-01-16 18:44:41

I will be applying to this if anyone wants to bounce ideas and compare notes. I’ll post an abstract here in a couple days for critique.

Ben Weinstein (benweinstein2010@gmail.com)

2025-01-21 13:02:10

As promised, here is an abstract and figure for those who are interested and skeptics!

- The challenge this project addresses is: [50 words]
Fine-grained computer vision for airborne imagery requires expensive image annotation due changes in class appearance and variation in the acquisition environment. Data on classes is plentiful in ground-based data catalogs, but machine learning approaches cannot take in ground views and generalize across perspective to top-down views for airborne image prediction.

- The challenge is significant because: [50 words, include data]
Airborne observation is a critical tool for species conservation and natural resource management. Conservation teams mostly review images by hand. When using automation, the majority of effort is searching for examples of classes that are common in ground-based views. When humans review imagery, they reference these materials, models should too.

Proposed Solution
- The generative AI-powered solution we are proposing is: [50 words]
Using ground-based data and text prompts to generate realistic examples to pretrain model backbones. This will allow users to select a location on earth and receive a pre-trained backbone for those species that are known to occur and are visible in airborne imagery.

- The solution will effectively address the problem described above by: [50 words, include data/examples]
Vastly reducing the annotation effort and by providing synthesized detections of species. This strategy will provide the model with the same data given to annotators to aid species classification and offers a proactive approach to data discovery compared to manually searching for examples within target imagery.

The basic tenant here is that when I go birding, I know the species that are present, and that goes beyond just a species distribution model, I know what they look like, I know their behavior, I look at field guides and compare species, I compare photographs. Airborne models don't get this information, and can't inherently port images from egocentric to top-down views.

Ben Weinstein (benweinstein2010@gmail.com)

2025-01-21 13:05:38

@Kate Rakelly, @Benjamin Kellenberger @Kirk Larsen, @Hannah Kerner, @David Russell, @Sara Beery @Yi Jin Toh @Caleb Robinson and others interested in airborne/simulated data.

👀 Yi Jin Toh, Sara Beery, Esther Rolf, Nathan Fox, Garrett, Hannah Kerner

👍 Kirk Larsen

❤️ Hannah Kerner

Chaim (c.elchik123@gmail.com)

2025-01-21 17:00:45

@Ben Weinstein this is a very interesting proposal, I am working on detecting and tracking African wildlife from drone videos and have the same issue that annotating the videos takes up much time and resources. I would be very interested in contributing to this project.

👍 Ben Weinstein

Devis Tuia (devis.tuia@epfl.ch)

2025-01-22 02:49:47

Sounds interesting as an idea, Ben. Why not throwing in also some information about position, that should help ruling out species that are unlikely

Devis Tuia (devis.tuia@epfl.ch)

2025-01-22 02:50:21

And there is all the question of avoiding the model hallucinating traits of the birds

👍 Ben Weinstein

David Russell (davidrussell327@gmail.com)

2025-01-22 10:17:53

This is a cool concept. Any sense how prevalent multiple views of the same individual are? And how frequently birds are in flight vs. not? I'm thinking about some of the geometric de-double-counting ideas we'd talked about previously and whether that could be used to provide multiple images of the same individual at some point in this pipeline.

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2025-01-22 11:23:44

Thanks for sharing this great proposal idea @Ben Weinstein .. I think the pipeline sounds solid. For training detection and pose models for zebras and horses in airborne images, we have been doing partly similar things (training with synthetic data and then fine-tuning with human-assisted, semi-automatic annotations (that does tracking of annotations and speeds the human's effort in annotation). For synthetic data generation (and augmentation) we have been relying on available 3D assets in sim environment. I really like your generation idea, including using texts here. Perhaps clarifying how realism will be ensured can make it stronger.

Ben Weinstein (benweinstein2010@gmail.com)

2025-01-22 12:41:57

@Aamir Ahmad what sim tools have you been using? https://docs.blender.org/api/current/info_overview.html There are two distinct visions here, one possibility is to create 3D models of birds -> place them in simulation landscapes -> extract 2D images for training (@Elizabeth Bondi-Kelly). The other is to directly synthesize training data in the target modality using images + text. I'm open to either idea, 3D landscapes sound like a harder path, it was done with whales though. https://github.com/prgumd/SeaDroneSim2. @Devis Tuia, you are definitely right about hallucinations, prelim tests with just a single image leads gemini to get a bit creative. Here is a synthesized image of a great shearwater that is a bit of a frankenstein.

Kate Rakelly (krakelly@gmail.com)

2025-01-22 12:57:49

This could be relevant: https://arxiv.org/abs/2404.12333

arXiv.org

Customizing Text-to-Image Diffusion with Object Viewpoint Control

Model customization introduces new concepts to existing text-to-image models, enabling the generation of these new concepts/objects in novel contexts. However, such methods lack accurate camera view control with respect to the new object, and users must resort to prompt engineering (e.g., adding ``top-view'') to achieve coarse view control. In this work, we introduce a new task -- enabling explicit control of the object viewpoint in the customization of text-to-image diffusion models. This allows us to modify the custom object's properties and generate it in various background scenes via text prompts, all while incorporating the object viewpoint as an additional control. This new task presents significant challenges, as one must harmoniously merge a 3D representation from the multi-view images with the 2D pre-trained model. To bridge this gap, we propose to condition the diffusion process on the 3D object features rendered from the target viewpoint. During training, we fine-tune the 3D feature prediction modules to reconstruct the object's appearance and geometry, while reducing overfitting to the input multi-view images. Our method outperforms existing image editing and model customization baselines in preserving the custom object's identity while following the target object viewpoint and the text prompt.

Original URL: https://arxiv.org/abs/2404.12333

Kate Rakelly (krakelly@gmail.com)

2025-01-22 13:15:27

I wonder how far you could push this - could you learn a representation of an animal species based on a bunch of text about it (wikipedia, encyclopedia), videos of its movement, maps of its range etc. maybe you could use that to generate data in locations/habitats/situations where collection is currently sparse.

🎉 Ben Weinstein

Kate Rakelly (krakelly@gmail.com)

2025-01-22 13:15:59

Also, generating data seems particularly useful for improving rare species detection? Maybe worth mentioning that?

👍 Ben Weinstein

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2025-01-22 13:29:09

@Ben Weinstein We have been taking the first route.. 3D assets, placements, flying drones within that environment, rendering images from those viewpoints, etc. Our pipeline has ISAAC SIM, OMNIVERSE and is ROS-based. See these: https://grade.is.tue.mpg.de/ https://zebrapose.is.tue.mpg.de/

👍 Ben Weinstein

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2025-01-22 13:55:45

Birds and fish are tough to simulate but there might be some useful assets out there, a quick search revealed this: https://www.youtube.com/watch?v=sR-Xdt51YwI

YouTube

Regoliste (https://www.youtube.com/@Regoliste)

Create Cinematic Bird Animations in Blender - Tutorial

Original URL: https://www.youtube.com/watch?v=sR-Xdt51YwI

Elizabeth Bondi-Kelly (ecbk@umich.edu)

2025-01-25 06:24:30

@Ben Weinstein in case you wanted to go the "create 3D models of birds -> place them in simulation landscapes -> extract 2D images for training" route, we used AirSim in Unreal Engine (https://microsoft.github.io/AirSim/). we created an Unreal Engine environment and purchased some of the 3D assets for African animals, then flew around in simulation. this was awhile ago now so might not still work, but feel free to message me if you have any questions! cool idea!!

👍 Ben Weinstein, Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2025-01-27 23:53:16

I am speaking with this (very cool) author, simulated oyster detection in robotics. https://arxiv.org/pdf/2409.07003

🙌 Elizabeth Bondi-Kelly

Panayiotis Danassis (pdanassis@g.harvard.edu)

2025-01-22 10:52:08

Hello all! We invite you to submit any work related to AI for social impact to the Autonomous Agents for Social Good workshop @ AAMAS 2025!

DEADLINE: Feb 4, 2025. See you in Detroit!:)

For more information, please see: https://panosd.eu/aasg2025/

panosd.eu

Autonomous Agents for Social Good 2025

Original URL: https://panosd.eu/aasg2025/

Tiffany Vlaar (Tiffany.Vlaar@glasgow.ac.uk)

2025-01-22 15:37:40

Hi everyone! I wanted to highlight some fully-funded interdisciplinary PhD opportunities at the University of Glasgow (and partners) that are likely of interest to this community. Please help us spread the word 🙂

1. NETGAIN DFA – Interdisciplinary challenge-focused research around the science and practice of net gain and nature markets. Each project has supervisors from at least two NETGAIN university partners and one non-academic partner.

Website with projects: https://netgain.wp.st-andrews.ac.uk/phd-projects/

Applications close: 31 January 2025.

2. Exascale Computing for Earth, Environmental, and Sustainability Solutions, NERC Doctoral Landscape Award (ExaGEO DLA).

Website with projects: https://www.exageo.org

Applications close: 17 February 2025.

3. Leverhulme Programme for Doctoral Training in Ecological Data Science.

Website with projects: https://ecological-data-science.github.io

Applications close: 7 March 2025.

🌍 Oisin Mac Aodha

Tiffany Vlaar (Tiffany.Vlaar@glasgow.ac.uk)

2025-01-22 15:38:18

Since the deadline is coming up soon, in particular I want to highlight this project on Marine Rewilding, led by Dr Laurence de Clippele and with non-academic partner CreditNature, offered within the NETGAIN DFA https://www.findaphd.com/phds/project/the-marine-rewilding-effect-building-a-cost-effective-framework-for-measuring-marine-net-gain/?p181117

Please feel free to forward this to anyone you think may be interested

Carly Batist (cbatist@gradcenter.cuny.edu)

2025-01-24 15:09:35

Nature Week is coming to London end of March! Consisting of:

Nature Action Dialogues (Cambridge, 25-26 March)

Nature Tech Summit (London, 27 March)

Nature Tech Unconference (London, 28 March)

https://www.naturetechweek.com/

25-28 March 2025, UK

Nature Tech Week | 25-28 March 2025, UK

Original URL: https://www.naturetechweek.com/

👍 Ștefan Istrate, Juan Sebastián Cañas, Aleksis Pirinen

Aude Vuilli (aude.vuilli@gmail.com)

2025-01-27 03:27:01

Thanks for sharing! It seems the event is still very much in planning. @Carly Batist are you involve with the organisation? I am thinking it might interest some of the people in my lab 😇

Carly Batist (cbatist@gradcenter.cuny.edu)

2025-01-27 09:00:42

I’m part of the Nature Tech Collective which is mostly co-running the Unconference, but I would recommend reaching out to NTC directly about any questions you have. The Nature Action Dialogues are pretty planned already, that’s run by UNEP. The Summit is still in the works from what I understand as you said. The unconference is open for proposals now, because of the ‘unconference’ format, people submit proposals and then attendees vote on which ones to have. But also note that attendance is application-based as well.

👍 Aude Vuilli

Aude Vuilli (aude.vuilli@gmail.com)

2025-01-27 11:22:45

Great, thanks a lot for all the details information 🙂

Lydia Beaudrot (beaudrot@msu.edu)

2025-01-28 10:51:46

Open PhD position in applied ecology for a data scientist to develop ML models to detect, quantify, and map cryptic mortality in large carnivores with fantastic faculty including Richard Bischof: https://www.jobbnorge.no/en/available-jobs/job/274296/phd-scholarship-within-environmental-data-science

Jobbnorge.no

PhD scholarship within environmental data science (274296) | Norwegian University of Life Sciences (NMBU)

Job title: PhD scholarship within environmental data science (274296), Employer: Norwegian University of Life Sciences (NMBU), Deadline: Saturday, March 15, 2025

Original URL: https://www.jobbnorge.no/en/available-jobs/job/274296/phd-scholarship-within-environmental-data-science

👍 Oisin Mac Aodha, Sara Beery, Malte Pedersen, stefano puliti, Aleksis Pirinen

stefano puliti (stefano.puliti@nibio.no)

2025-02-03 03:43:53

Hi everyone,
here is a new dataset we just published for benchmarking tree species classification from proximal laser scanning data (TLS/MLS/ULS):

https://www.linkedin.com/feed/update/urn:li:activity:7292094140379504640/

linkedin.com

Stefano Puliti on LinkedIn: #openscience #ai #deeplearning #biodiversity #forestry #environment…

📣 𝗙𝗢𝗥-𝘀𝗽𝗲𝗰𝗶𝗲𝘀𝟮𝟬𝗞 it&#39;s officially out in 𝘔𝘦𝘵𝘩𝘰𝘥𝘴 𝘐𝘯 𝘌𝘤𝘰𝘭𝘰𝘨𝘺 𝘈𝘯𝘥 𝘌𝘷𝘰𝘭𝘶𝘵𝘪𝘰𝘯 British Ecological Society 🥳! With 20K trees of 33 species, this is currently the largest dataset for benchmarking tree species classification from proximal laser scanning data (TLS/MLS/ULS). 🌲 🌳 🌴 🎄 In addition to contributing to #OpenScience we provided a solid baseline with both 3D and 2D deep learning architectures benchmarked. Check out the links 🔗 to some of the goodies 👇 📰 paper: <https://lnkd.in/dfXCGKQX> 💾 data: <https://lnkd.in/dYjEtgaZ> 💻 code: <https://lnkd.in/dyRFG46m> 🎯 benchmark against the withheld test data: <https://lnkd.in/dVEsGiUU> It has been truly inspiring to witness the power of collaboration! 🤗 In a world where tech inequalities are increasingly evident, the scientific community must work toward a future where science is inclusive and accessible to all—free from barriers. A heartfelt thank you to the incredible 3DForEcoTech community (and beyond!) for coming together to make this vision a reality! An effort made possible by SmartForest4.0 (Norges forskningsråd) and COST Association - European Cooperation in Science and Technology A massive thanks 🙏 to all of the researchers around the 🌍 who made this possible: Emily Lines, Jana Mullerova, Julian Frey, Zoe Schindler, Adrian Straker, Matt Allen, Lukas Winiwarter, Natalia Rehush, Hristina Hristova, Brent Murray, Kim Calders, Louise Terryn, Nicholas Coops, Bernhard Höfle, Liam Irwin, Samuli Junttila, Martin Krůček, Grzegorz Krok, Kamil Král, Shaun Levick, Dr Linda Luck (Lück), Azim Missarov, 🌳 Martin Mokroš, Harry Owen, Krzysztof Stereńczak, Timo Pitkänen, Nicola Puletti, Ninni Saarinen, Chris Hopkinson, Chiara Torresan, Enrico Tomelleri, Hannah Weiser and Rasmus Astrup Climate AI Nordics #AI #DeepLearning #Biodiversity #Forestry #Environment #MachineLearning #OpenScience

Original URL: https://www.linkedin.com/feed/update/urn:li:activity:7292094140379504640/

👍 Binbin Xiang, Aleksis Pirinen, Nico Lang, Josafat-Mattias Burmeister, Sara Beery, Timm Haucke, David Russell, Andy Viet Huynh, Jane Wu

🌲 Binbin Xiang, Aleksis Pirinen, Nico Lang, Sara Beery, Timm Haucke, Dan Morris, Andy Viet Huynh, Jane Wu, Alan Stenhouse

Sara Beery (sara.beery@gmail.com)

2025-02-03 08:03:18

@Timm Haucke :)

🙌 Timm Haucke

Ben Weinstein (benweinstein2010@gmail.com)

2025-02-03 13:25:00

I am starting to sniff around OpenAI's deep research module. It was announced recently and can use the browser. I prompted:

I am looking for datasets that might include bounding boxes, points or polygons of annotations for machine learning projects for biodiversity science. I am looking for airborne imagery, such as drones, UAV data and other pictures of animals or trees from the sky, make me a csv file of the names, links, sources and description of the dataset. Here are examples of existing dataset I know about https://lila.science/ amd https://zenodo.org/records/13328697.

and I got

Dataset Name,Link,Source,Description
LILA (Long-Term Ecological Research),https://lila.science/,LILA Science,"LILA offers multiple datasets including UAV and drone-based images with annotated data for biodiversity research. Includes species detection and environmental monitoring datasets."
Zenodo (Record 13328697),https://zenodo.org/records/13328697,Zenodo,"This dataset includes airborne imagery and annotations for trees, vegetation, and animals, with bounding boxes and polygons. Ideal for machine learning applications related to forest and biodiversity monitoring."
xView2,https://xview2.org/,xView,"xView2 is a satellite and aerial imagery dataset annotated for object detection. Though it's more focused on infrastructure, it includes images that could be applicable to ecological studies, such as vegetation or wildlife habitats."
UAV-based forest inventory data,https://www.sciencedirect.com/science/article/pii/S235293851830498X,ScienceDirect,"This dataset consists of high-resolution UAV imagery annotated with polygons for individual trees, along with other forest features. It is designed for forest inventory and biodiversity applications."
Caltech UAV,https://data.caltech.edu/records/16,Caltech,"This dataset contains UAV images used for object detection, with annotations focused on a variety of objects, some of which could overlap with biodiversity applications (e.g., vegetation, trees)."
BIOMASS,https://data.unh.edu/dataset/biomass,University of New Hampshire,"The BIOMASS dataset contains UAV-based imagery annotated with points and polygons representing vegetation and forest cover, useful for biodiversity research."
FAIRY (Forest Area Imaging and Remote Sensing),https://www.fairy-data.org,FAIRY,"The FAIRY dataset offers aerial imagery from drones for ecological studies, including annotated images of trees and plant species for biodiversity conservation projects."
Biodiversity Monitoring Dataset,https://www.kaggle.com/datasets,Open Access,"A Kaggle dataset for biodiversity monitoring that contains UAV imagery, annotated with bounding boxes and points identifying various species and habitat types."
Biodiversity UAV Dataset (UAV4U),https://www.uav4u.com/dataset,uav4u,"This dataset provides UAV images focused on animal species, with polygon annotations for detecting various wildlife in natural habitats."
BirdsEyeView,https://www.nature.com/articles/s41598-019-44135-3,Nature,"BirdsEyeView dataset offers UAV imagery of bird populations, annotated with points and bounding boxes for species identification and population monitoring."

Ben Weinstein (benweinstein2010@gmail.com)

2025-02-03 13:34:45

Thread Reply:

Ben Weinstein (benweinstein2010@gmail.com)

2025-02-03 17:51:01

I love the promise and hype of the AI cycles, this feels like a great way to end the end. From my annotator Brier -> " Okay most of the links do not work - there are a handful that do, but only 3 that actually have data of interest for us. This would be for Lila Science, Zenodo (Record 13328697) and kaggle datasets.

Going to try to key-term search for the data that had links that did not work and see if I can find it, or if it does not really exist."

👀 Nico Lang, Chris Lange, Elizabeth Bondi-Kelly, Sepand Dyanatkar

stefano puliti (stefano.puliti@nibio.no)

2025-02-04 08:29:15

Can it parse all labels to a common format (e.g. COCO?). Now that would be a win 😀

DEV MEHROTRA (dmehrotra@wisc.edu)

2025-02-03 14:00:34

Hi everyone! I’m Dev, a master's student at UW-Madison. I’ve developed and worked on multiple conversation projects with IoT and ML, and mainly on a device called the Insect Eavesdropper, which uses contact microphones to measure biodiversity in insects! Right now, I’m working in a medical image analysis lab with foundation models. I’ve missed the outdoors, so I joined this community to stay productive at my desk while learning something new. Looking forward to connecting with you all!

👍 Sara Si-Moussi, Jose Ruiz-Munoz, Aleksis Pirinen, Seven Song, Owen Martin, Andy Viet Huynh

👋 Chris Lange, Owen Martin, Andy Viet Huynh, Helena Russello, Thomas Radinger

Eric Orenstein (eorenstein@mbari.org)

2025-02-05 08:50:08

I was just made of aware of the Locus Charter on responsible, ethical use of geospatial location data. Has anyone here engaged with the organization or otherwise thought about the ethical use of location data? The material on the Locus Charter website was posted c. 2021. I'm sure there are more recent discussions around this sort of thing. I'd love to get your pointers on where to look!

EthicalGEO

Locus Charter | EthicalGEO

The Locus Charter proposes that wider, shared understanding of risks and solutions relating to uses of location data can improve standards of practice, and help protect individuals and the public interest. We hope the Charter can improve understanding of risk, so those can be managed, and the many benefits of geospatial technologies can be realized ...

Est. reading time

10 minutes

Original URL: https://ethicalgeo.org/locus-charter/

👀 hannah miro

Ben Weinstein (benweinstein2010@gmail.com)

2025-02-05 18:15:07

Thanks to a user, we updated the DeepForest airborne livestock detection model. Nice notebook here. If there are those of you out there that want to play with it and see how it generalizes to new scenes, I'd be interested to hear how it performs. We've tested it out on 7 or 8 datasets, it could use more data, but provides a decent start. Definitely needs more diverse scenes for small objects.

👏 Lasha Otarashvili, Sara Beery, Julia Chae

🐄 Dan Morris, adam dunlop

Ted Schmitt (teds@allenai.org)

2025-02-05 18:41:32

@Howard L Frederick, this could be interesting to you

Nate Harada (nharada1@gmail.com)

2025-02-05 18:45:32

interesting, curious Ben are these data are also in ortho form or if because livestock move around too much that’s less practical?

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2025-02-05 20:24:26

@Lasha Otarashvili @Kirk Larsen

🐄 Kirk Larsen, Lasha Otarashvili

Howard L Frederick (simbamangu@gmail.com)

2025-02-05 22:49:46

Interesting - I'm forwarding to TAWIRI guys who have some datasets to test on (I'm travelling at present). Will dive in as soon as possible though

Ted Schmitt (teds@allenai.org)

2025-02-05 22:54:43

TAWIRI is the Tanzania Wildlife Research Institute (for those who may be wondering) 🤔

Howard L Frederick (simbamangu@gmail.com)

2025-02-05 23:04:50

Oops yes

Aakash Gupta (aakash@thinkevolveconsulting.com)

2025-02-08 10:49:38

I was recently reviewing some of the agentic framework for object detection and thought I'll test them on these images. Sharing the screengrab. Just for reference....

Ben Weinstein (benweinstein2010@gmail.com)

2025-02-08 11:16:29

cool, the future is in labeling a single object and being able to propagate labels into the rest of the image, but beyond the trivial images.

Hannah Kerner (hkerner@umd.edu)

2025-02-14 18:21:46

@adam dunlop I wonder how it would work for deer 😄

adam dunlop (adunlop621@gmail.com)

2025-02-14 22:01:40

The data definitely looks similar enough for it to work on the deer! And it looks like we can finetune it with a pretty small amount of data!

adam dunlop (adunlop621@gmail.com)

2025-02-15 15:12:31

@Ben Weinstein Out of curiosity, would this be able to track livestock throughout a video?

Ben Weinstein (benweinstein2010@gmail.com)

2025-02-15 20:29:50

Not out of the box, but could use the model features to train a typical tracking algorithm

🙌 adam dunlop

Stephen Formel (sformel@usgs.gov)

2025-02-07 09:43:20

It’s that time of year where again when GBIF is offering money to support grad students! The Graduate Researcher Awards (GRA) is an annual prize for Grad students to show how they’re using open biodiversity data from the GBIF network in their graduate research. What is the prize? A cool €5,000! The deadline for US applications is May 27, 2025. https://www.gbif.us/post/2025/gbif-graduate-researchers-award-call/

gbif.us

GBIF Graduate Researchers Award Call - GBIF.us

<a href="http://GBIF.US">GBIF.US</a> is committed to mobilizing and providing access to species occurrence (presence and absence) data that adheres to Findable, Accessible, Interoperable, Reusable (FAIR) data principles.

Original URL: https://www.gbif.us/post/2025/gbif-graduate-researchers-award-call/

Lukas Picek (lukaspicek@gmail.com)

2025-02-07 10:03:06

Hi there, I assume this is only for US students, right?

Tor Henrik Ulsted (tor.henrik@ulsted.no)

2025-02-07 10:07:49

It does seem like it is for all countries contributing to GBIF:
"""
Eligibility

Candidates must provide a letter of support verifying their affiliation with a university graduate programme (as a currently enrolled master’s or post-doctoral student) to be eligible for the award. In addition, candidates must be either:

a) citizens of a country participating in GBIF or
b) students at an institution located in a GBIF participant country.
"""

👍 Lukas Picek

Lukas Picek (lukaspicek@gmail.com)

2025-02-07 10:09:01

Oh, I see. There is also global call --> https://www.gbif.org/news/4AAE2KOKBRJ0iS2DE2fT65/call-for-nominations-opens-for-2025-gbif-graduate-researchers-award

gbif.org

Call for nominations opens for 2025 GBIF Graduate Researchers Award

Deadline for GBIF member countries to nominate graduate students whose innovative research relies on biodiversity data: **27 June 2025**

Original URL: https://www.gbif.org/news/4AAE2KOKBRJ0iS2DE2fT65/call-for-nominations-opens-for-2025-gbif-graduate-researchers-award

💯 Tor Henrik Ulsted

Stephen Formel (sformel@usgs.gov)

2025-02-07 11:29:21

Yes, you got it right. Each national node can nominate students based on their citizenship or association with an institution in that country. I’m the US node manager, so please feel free to send any questions my way.

Stephen Formel (sformel@usgs.gov)

2025-02-07 11:30:09

I would also be happy to introduce anyone to other node managers if that’s helpful.

Rebecca Kleinberger, PhD (rebecca.kleinberger@gmail.com)

2025-02-07 12:03:53

Hi everyone

I will be giving a virtual talk tonight hosted by the Peabody Essex Museum on my lab work on using technology to give more control and agency to animal in managed care (zoos, rescue, homes, labs) .

Animals Take Control: Technology for Enrichment and Connection
https://www.pem.org/events/animals-take-control-technology-for-enrichment-and-connection

Free and open to all ;)
🦜🐈🦔🐄🦍📱🖲️🕹️🎚️

pem.org

Animals Take Control: Technology for Enrichment and Connection

Can technology support the well-being of birds and other animals? In this virtual program, researcher Rébecca Kleinberger shares recent results

Original URL: https://www.pem.org/events/animals-take-control-technology-for-enrichment-and-connection

🦜 Oisin Mac Aodha, Kalindi Fonda, Ben Weinstein, Hannah Robertson, Aleksis Pirinen, Ben Seleb, Cameron Trotter, Rebecca Wilks

Stephen Formel (sformel@usgs.gov)

2025-02-10 12:16:34

Another opportunity for grad students (US citizens)!

The Earth Science Information Partners (ESIP) awards the Raskin Scholarship annually to an individual in the Earth or computer sciences who has an interest in community evolution of Earth science data systems. Special attention will be given to applicants demonstrating an interest in semantics, GIS, cyberinfrastructure and computing in the geosciences.

Award: $5000 + attendance of ESIP summer meeting (Seattle, WA)
Deadline: March 14, 2025
https://www.esipfed.org/student-opportunities/raskin-scholarship/?mc_cid=c91650ae50&mc_eid=e688cb4b33

Please note that even though some people don’t consider biology and ecology, ‘Earth Sciences’, ESIP does. I co-chair the Biological Data Standards cluster, and would be happy to give feedback to anyone who is interested in submitting.

PILI - ESIP

Raskin Scholarship - ESIP

ESIP's Raskin Scholarship seeks to promote collaboration, research support, and exposure for talented students in the Earth or computer sciences. Apply now.

Time to read

Less than a minute

Original URL: https://www.esipfed.org/student-opportunities/raskin-scholarship/?mc_cid=c91650ae50&mc_eid=e688cb4b33

Mike Volkin (mike@mikevolkin.com)

2025-02-11 14:49:08

My new nonprofit, Wild Tech Alliance, helps fund and support conservation technology projects. If you have a project that needs funding—or know someone who does—I’d love to connect. Let me know if this is relevant!

🙌 Lloyd Hughes, Tiziana Gelmi Candusso, Selena Sun, Anna Boser, Juliana Gomez Consuegra, Arthur Caillau, An Yu, Meredith Palmer, Elena Sierra, Quentin Geissmann, Carly Batist

❤️ Sara Beery, Ankita Shukla, Sasha Syunkova, Aleksis Pirinen, Alfredo Lozada Fuentes

👍 Sol Milne

Ben Weinstein (benweinstein2010@gmail.com)

2025-02-11 14:51:00

Can you tell us all a bit about your scope, mission, size of funding? Thanks!

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2025-02-11 14:55:49

Thanks for sharing @Mike Volkin, I echo @Ben Weinstein's question. Also, is this the website?
https://www.wildtech.org

Valentin Lucet (valentin.lucet@gmail.com)

2025-02-11 15:01:53

Or is it https://www.wildtechalliance.com/? Also curious like Ben!

Christine (christine@whaleseeker.com)

2025-02-11 15:05:29

Curious to learn more as well

Mike Volkin (mike@mikevolkin.com)

2025-02-11 15:28:11

Thanks guys just to add a little bit more clarity here, the website is wildtechalliance.com, it is not done, but the basic pages are up. We’re integrating the tech stack (the crowd funding component )at the end of the month. For the business model, our goal is to be the central hub for all conservation technology projects who need funding. We have about a dozen categories that the donor can filter from if they’re interested in donating in a particular category, and then we also have stages from idea all the way through implementation and capability that they can also filter projects from. My team will try to secure corporate donors first and then individual donors second. When you create a project you’ll get a link that you can share and try to bring in your own donations as well. Feel free to ask any more questions

✔️ Jon Van Oast, Eric Greenlee

Mike Volkin (mike@mikevolkin.com)

2025-02-11 15:28:43

I have a waitlist I’m keeping for those who want to be notified when I’m able to post projects, if you wanna be on it just let me know

Christine (christine@whaleseeker.com)

2025-02-11 15:37:43

yes, that sounds interesting. I would like to be informed thanks

👍 Mike Volkin

Kalindi Fonda (kalindi.fonda@gmail.com)

2025-02-11 15:53:05

Awesome initiative, so the funding would mainly come from private donors or also organizations/other forms?

Lukáš Adam (lukas.adam.cr@gmail.com)

2025-02-12 03:40:50

Be led by a U.S.-based individual or team, though the project can have global applications.

Is this a hard requirement? I am asking because the webpages seem not to be finished yet.

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2025-02-13 04:06:59

@Mike Volkin thank you so much for explaining! Appreciate your efforts. ❤️
Honest question: How does wildtechalliance.com compare to the long-running https://experiment.com with an established track record???
They fund many conservation tech projects, accepts corporate and individual donors (including major non-profit donors), and supports international projects (doesn't need US-based applicants).
Are we risking re-invention of wheels and diluting efforts? Maybe you can team up with experiment.com in some way?
I promise I'm not trying to put down this project, just genuinely asking! Thanks for your understanding.

Blair Costelloe (blaircostelloe@gmail.com)

2025-02-17 04:43:26

@Mike Volkin Thanks for posting this, very interesting! I am also interested to know the size of projects you envision supporting - i.e. is there a range that the project budget should fall into, or is this completely open?

Mike Volkin (mike@mikevolkin.com)

2025-02-17 12:31:01

@Christine @Kalindi Fonda @Lukáš Adam @Pen-Yuan Hsing @Blair hey guys, thanks for the comments, responded to all of them with this video: https://www.loom.com/share/464b407689424519ab7e5a21a8b2d534?sid=4aea8e0c-b1c3-4f34-8cac-0baef4ec345c

Loom

Wild Tech Alliance Update 🌍

Hey everyone, I wanted to give you an update on our efforts with the Wild Tech Alliance. We're focusing on corporate and private donations to support wildlife conservation projects, and I appreciate your patience as I catch up on messages. I also want to clarify that we're open to projects from around the world, not just the U.S. Please reach out to me via email for quicker responses, and let’s keep the momentum going in our marketing efforts!

Original URL: https://www.loom.com/share/464b407689424519ab7e5a21a8b2d534?sid=4aea8e0c-b1c3-4f34-8cac-0baef4ec345c

Jiangyue Wang (jyuewang@uw.edu)

2025-02-11 15:29:29

Hi everyone, I am Jiangyue Wang from UW Quantitative Ecology and Resource Management, working on mountain lion mortality and movement on the Olympic Peninsula. I am a first-year graduate student and would love to engage in this community to learn and discuss AI for conservation!

👋 Nico Lang, Taiki Sakai - NOAA Affiliate, Shir Bar, Elena Sierra, Kate Rakelly, Marie Tosa, gyeongsu, Quentin Geissmann, Viktor Domazetoski, Alexander Robillard, Angela Zhu, Stephanie Mitchell

🎉 Catherine Breen, Tessa Rhinehart, Catalina Bernal, Alexander Robillard, Angela Zhu

👋:skin_tone_3: Jess Tam

👋:skin_tone_2: Patrik Reiske, Cara Appel

Jiangyue Wang (jyuewang@uw.edu)

2025-02-11 15:30:18

Thanks @Catherine Breen for inviting me!

❤️ Catherine Breen

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2025-02-12 19:54:05

We would love to try face ID for mountain lions with our existing models at Wild Me if you're interested

❤️ Jiangyue Wang

Stephanie Mitchell (smitch88@uw.edu)

2025-03-10 17:59:33

Hi Jiangyue - I’m a 2nd year PhD student at UW using tech and AI in wildlife health surveillance. I’d love to connect!

❤️ Jiangyue Wang

Arthur Caillau (arthur@caillau.me)

2025-02-12 04:15:53

Hi everyone!
As the founder of EarthToolsMaker, I very often need to train and finetune ML models on a GPU. I have been doing this in the cloud for the past year and half which is not ideal for many different reasons (the data needs to be transferred there, the per hour pricing is sometimes ridiculous especially when trying out new models, etc). In order to bring down the compute costs, I am currently setting up a workstation and I will be hunting for a GPU (something like a RTX 4090).
Do you guys have any contacts, resellers, sponsors I could reach out to?
Thanks a lot!

Bernie Boscoe (boscoeb@sou.edu)

2025-02-12 11:57:39

For our project, we are happy with our Alienware (Dell) with an RTX 4090. (I realize this would only be at discount but I have failed to find anyone to donate 4090s)

Piotr Tynecki (piotr@tynecki.pl)

2025-02-12 12:59:41

NVIDIA GPU grant?

Nick Deere (nicolas.deere@gmail.com)

2025-02-12 04:48:28

Hi everyone, I'm Nicolas Deere, a postdoc at the Durrell Institute of Conservation and Ecology (DICE), University of Kent, exploring the long-term impacts of ecological and socio-economic change on mammal populations across Indonesia using camera trap data. I'm really keen to explore options for integrating AI across all stages of our workflow, from data processing to analysis, and look forward to learning from this amazing community :)

👋 Chris Lange, Sara Beery, Dan Morris, Timm Haucke, Tiziana Gelmi Candusso, Elena Sierra, Sol Milne

👋:skin_tone_3: Alan Stenhouse

Patrik Reiske (nailed-flow-0a@icloud.com)

2025-02-12 05:34:13

Hi everyone! I’m Patrik Reiske, a PhD student at Technische Universität Berlin, and my focus is visual representation learning in context of intelligent (rodent) behavior.

🐀 Chris Lange, Alan Stenhouse

👋 Chris Lange, Sara Beery, Timm Haucke, Elena Sierra, Andre Telfer

Andrea Zampetti (andrea.zampetti@uniroma1.it)

2025-02-12 10:51:37

Hi everyone! I'm Andrea Zampetti, a PhD student in animal biology at Sapienza University of Rome, and my research focuses on integrating automated approaches for efficient mammals monitoring and conservation, particularly through camera-traps. I've always been interested in AI applied to conservation biology, and in my latest works I started developing and incorporating deep learning algorithms for mammals and birds species classification in camera-trap projects. Excited to be here in this community!

👋 Sara Beery, Elena Sierra, Justin Kay, Diego Kiedanski, Dan Morris, Jiangyue Wang, Stephanie Roilo

👋:skin_tone_3: Alan Stenhouse

Filippo Varini (fppvrn@gmail.com)

2025-02-12 12:19:31

Help Build the Most Comprehensive Open-Source Fish Detection Model! 🐟🌊
Hey everyone! With @Oc and @Dan Morris, we are launching a community-driven project to create a universal fish detector, an AI model capable of detecting any fish-like creature, anywhere! 🦈🐠🐡
A ton of fish detection datasets are publicly available online, but they’re scattered across different formats. To train a powerful model, we need to gather, analyze, and standardize these datasets—and that’s where you come in!
:female_technologist: How can you help?
We are keeping a list of fish datasets to process here. Choose a dataset, analyze its metadata, and create a notebook to visualize annotated images.
💡 Why join?

Create an AI model that disrupts ocean conservation like the MegaDetector model did on land 🚀.

Gain hands-on experience with computer vision & dataset curation 🔬

Get credited as a contributor to a high-impact conservation effort 🙌

📌 Start here:
Read this Notion guide to start contributing to the project!
Additional resources:
📝 Tracking sheet: Dataset Excel
📂 GitHub repo: Fish Datasets
Any questions? Reach out to us on slack or email at fppvrn@gmail.com.
Let’s build this together! 🛠️💙

P.S. Let us know if we missed any dataset, by filling this form!

🙌 Carly Batist, Devi Ayyagari, Malte Pedersen, Timm Haucke, Levi Cai, Shir Bar, Talen Rimmer, Laura Chrobak, Jason Holmberg (Wild Me), Dante Wasmuht, Sonny Burniston, Yuerou Tang, Xiaoxuan (Shirley) Ma, Juliana Gomez Consuegra

👏 Lukas Picek, S M Rayeed, Timm Haucke, Jason Holmberg (Wild Me), Victor Anton

🐠 Shir Bar, Mitchell Rogers, Jason Holmberg (Wild Me), Alexander Merdian-Tarko, Kalindi Fonda, Alexander Robillard, Sofía Miñano, Alan Stenhouse

🙌:skin_tone_5: Prabath Gunawardane

❤️ Ben Williams, Alexander Robillard, Ghjulia Sialelli, Valerie

🙌:skin_tone_3: Alan Stenhouse

Ben Weinstein (benweinstein2010@gmail.com)

2025-02-12 12:59:02

Have you talked to fathomnet? They have lots of models and data to start from?

Ben Weinstein (benweinstein2010@gmail.com)

2025-02-12 12:59:20

https://huggingface.co/FathomNet

huggingface.co

FathomNet (FathomNet)

Original URL: https://huggingface.co/FathomNet

Levi Cai (levi@levicai.com)

2025-02-12 13:18:34

My team has scripts to handle FathomNet, AIMs OzFish, DeepFish, Brackish, NOAA puget, Kakadu, VIAME (approximately 1 million images in total) and transforming them into YOLO-format datasets. They are not very clean, but we are happy to share if you have a repo somewhere.

👍 Sara Beery, Filippo Varini

👍:skin_tone_4: Oc

Sara Beery (sara.beery@gmail.com)

2025-02-12 13:35:56

@Talen Rimmer

❤️ Talen Rimmer, Oc

Filippo Varini (fppvrn@gmail.com)

2025-02-12 14:32:51

Thanks @Ben Weinstein, yes I have talked to them. They do have generalizable models but there are more datasets that such models could be trained with. So the goal is to improve the SOTA that exists currently, with the same approach.
CC @Laura Chrobak

✅ Laura Chrobak

Filippo Varini (fppvrn@gmail.com)

2025-02-12 14:33:40

Hi @Levi Cai,thank you so much, that would be amazing!! You can push them to the repo with a PR or send them to me via email as well fppvrn@gmail.com

Filippo Varini (fppvrn@gmail.com)

2025-02-12 14:35:08

CC @Sonny Burniston

Kalindi Fonda (kalindi.fonda@gmail.com)

2025-02-13 02:15:37

I am in!

🙌 Filippo Varini

Kalindi Fonda (kalindi.fonda@gmail.com)

2025-02-13 02:21:29

@Victor Anton check this out

👍 Victor Anton

Cameron Trotter (cater@bas.ac.uk)

2025-02-13 05:08:35

@Filippo Varini unsure if it will be of help, but I updated the scraper used to compile the FishNet dataset. The code they used is here: https://github.com/Trotts/spiderfish

CC: @Andrew Temple

Filippo Varini (fppvrn@gmail.com)

2025-02-13 07:03:29

Thanks @Cameron Trotter! Do you know if those images contain bounding boxes? Or are they just image classification?

Cameron Trotter (cater@bas.ac.uk)

2025-02-13 07:13:24

@Filippo Varini running that scraper again will only give images, fishbase doesn't store bboxes etc. Looking at the paper, scraped images were manually cleaned further and then combined with some stuff from iNaturalist, but @Andrew Temple would be your point of contact for that!

Filippo Varini (fppvrn@gmail.com)

2025-02-13 08:01:24

Thanks!!

Yuerou Tang (yuerou.tang@berkeley.edu)

2025-02-13 17:55:34

i am having some troubled accessing the notion page -- it says The page you are looking for cannot be found!

Filippo Varini (fppvrn@gmail.com)

2025-02-14 09:03:06

Should be up now!

:gratitude_thank_you: Yuerou Tang

Ben Weinstein (benweinstein2010@gmail.com)

2025-02-12 13:43:18

i'm sure i'm not alone in feeling like the release of tech dev is really outpacing our community's ability to make sense of our use cases. I really encourage everyone to drop in snippets of experience with new tools, what is fad and posturing, what feels real and what kinds of costs we are engaging with. For example, I am playing with Vision Agent https://github.com/landing-ai/vision-agent?tab=readme-ov-file and watching it recapitulate like 6 months of DeepForest dev work is pretty remarkable. I asked it to detect cattle, and it wrote me code. I am considering trying to integrate this chat like service into our codebase to help users. Understanding pricing is still very murky. Here is agentic object detection which was on par with DeepForest's tree model. How costs connect to credits, whether this includes a LLM cost. VisionAgent is open source, but connects to openAI/antropic, so I have no idea what kind of costs that would incur, but my gut says small dollars to process imagery for annotation to be reviewed by people.

😮 Jon Van Oast, Alexander Robillard, Ben Lopata, Louis Freeland-Haynes, adam dunlop

❤️ Jon Van Oast, Shir Bar, Alexander Robillard, Aakash Gupta

👍 Sepand Dyanatkar, Alexander Robillard, Nils Blum-Oeste, Cameron Trotter, Sandupal Dutta, Cole von Glahn, Xiaoxuan (Shirley) Ma

👀 Shir Bar, Alexander Robillard, Dante Wasmuht, ash, Hannah Yin

👍:skin_tone_5: Prabath Gunawardane

👍:skin_tone_3: Jess Tam, Alan Stenhouse

➕ Rachel Mason

Ben Weinstein (benweinstein2010@gmail.com)

2025-02-12 13:44:46

There is also the existential and ethics question about what our role in all of this, how much to rely on the large companies, data sharing, and what types of applications are useful, and what types of training data are still important for us to train on our own.

👍 Dan Stowell, Marion Richardot

👍:skin_tone_3: Hannah Yin

Chaim (c.elchik123@gmail.com)

2025-02-13 03:58:18

@Ben Weinstein How does the vision model it selected compare to your own trained and fine tuned model? And is it's main focus to make coding easier or is it eliminating object detection model training as a whole?

➕ Rachel Mason

Aakash Gupta (aakash@thinkevolveconsulting.com)

2025-02-13 04:05:57

There will be options to use local models for running workflows on proprietary/sensitive data

Dan Stowell (dan.stowell@naturalis.nl)

2025-02-14 08:05:08

We are pleased to announce "BioDCASE" - the Bioacoustics Data Challenge! https://biodcase.github.io/

BioDCASE will be a satellite event of the DCASE challenge and workshop 2025. We developed this initiative to help address the needs of people who monitor animal sounds, while also solving some fascinating computational problems!

The BioDCASE challenge will be in spring/summer. There will also be a one-day workshop on 29th Oct 2025 as part of DCASE.

We encourage YOU to take part! Three different tasks have been designed - decide which one you'd like to help solve. The newly-assembled audio datasets will be released in April, giving you time to work with the data before the challenge deadline of 15th June.

The BioDCASE 2025 challenge tasks are:
1. Multi-channel alignment
2. Supervised detection of strongly-labelled Antarctic blue and fin whale calls
3. Bioacoustics for tiny hardware

More details: https://biodcase.github.io/
Please forward this announcement to anyone who might be interested. Students, data scientists, colleagues.

biodcase.github.io

BioDCASE

About BioDCASE - Evaluation & Benchmarking in...

Original URL: https://biodcase.github.io/

🙌 Sara Beery, Oisin Mac Aodha, Ben McEwen, Timm Haucke, Xiaoxuan (Shirley) Ma, Céline Angonin, Owen Martin, Arthur Caillau, Alessandra Vidal Meza, Thomas Radinger, Ben Weinstein, Juan Sebastián Cañas, Benjamin Hoffman, Jose Ruiz-Munoz, Aude Vuilli, Santiago Martinez Balvanera, Burooj Ghani, Tessa Rhinehart, Yasmine Benhamadi, Alexander Robillard, Carla Peterson, Georgia Atkinson, Carly Batist

🐳 Taiki Sakai - NOAA Affiliate, Kate Rakelly, Brian Geuther, Alexander Robillard, Arthur Wandzel, Alexander Merdian-Tarko

🔉 Maddie Cusimano, Ben Seleb, Arthur Wandzel

🤩 Aude Vuilli, Arthur Wandzel

🧠 Arthur Wandzel

🌿 Arthur Wandzel

Eugene Galaxy (eugene@animaldetect.com)

2025-02-19 02:59:33

Hey everyone!

I want to share a fun free app I made for the community. Find your spirit animal using AI in 5 seconds!

To promote awareness of different animal species, I made a simple app that generates your spirit animal and gives it a fancy name! Also, it’s been a great fun and exercise for my web dev skills.
https://deepfocused.me

No sign-up needed.

Free

Takes 5 seconds

Download button

DO NOT PRESS the “DO NOT PRESS” button

10% chance to generate a chimera (crossbreed of 2 animals), BEWARE!

The results can be absolutely unhinged, no filters or blocks applied.

I was surprised how easy it is to host your own AI-model on a PC and create a 1-page app.
Almost 0 expenses (minus electricity). Nowadays, there are services with free tiers generous enough to keep you running free forever, given small volumes.

Supabase for database, Umami for visitor counting, Vercel for web-hosting, Ngrok for server tunneling.
I am planning to write a simple guide on how you can create your own simple server+app in 1 evening.

Post what you got here :^)

❤️ Mitchell Rogers, Gaspard Dussert, Rebecca Wilks, Laura Madrid, Chris Llorca

👏 Mamoon, Nora Gourmelon

👏:skin_tone_4: Chris Llorca

Chris Llorca (llorca@terraformation.com)

2025-03-10 16:46:36

Love it! 🙂

Funny Cormorant of Shadow Self

🙌 Eugene Galaxy

Jeremy Fix (jeremy.fix@centralesupelec.fr)

2025-02-19 05:11:56

Hi, if this can be useful for you , I just wrote a little python GUI app wrapping SAM2.1 for labeling images; It supports prompting by positive/negative points and box. On the github , you will find illustrations on diatoms and plankton.

Available there : https://github.com/jeremyfix/sambiotic

Feel free to provide feedbacks.

jeremyfix/sambiotic

Language

Python

Last updated

5 days ago

👏 Mamoon, Chaim, Shir Bar, Xiaoxuan (Shirley) Ma, Otto Brookes, Jason Holmberg (Wild Me)

❤️ Sara Si-Moussi, Justin Kay, Sara Beery, Angus Galloway, Jason Holmberg (Wild Me)

Steve Haddock (haddock@mbari.org)

2025-02-22 00:39:23

Hi @Jeremy Fix This is not to do with your nice GUI, but I had a question about SAM itself. It looks like it works well for the copepods and chaetognaths in your plankton demo. I wonder how it does on the salps and if it picks up that calycophoran siphonophore in the middle..?

Jeremy Fix (jeremy.fix@centralesupelec.fr)

2025-02-22 06:18:38

It may not be exactly the species you are looking for but a colleague at Villefranche asked me for some of the semi-transparent plankton . I do have the following examples using SAM2.1 large :

I'm not myself expert of plankton but I have the feeling that some of these match the species you requested

Jeremy Fix (jeremy.fix@centralesupelec.fr)

2025-02-22 06:19:47

And the result of a quick and dirty annotation run , just adding one positive point per individual :

Steve Haddock (haddock@mbari.org)

2025-02-24 15:05:06

Thanks @Jeremy Fix. Looks like it does a pretty good job even on the salps.

👍 Jeremy Fix

Steve Haddock (haddock@mbari.org)

2025-02-24 15:59:25

cc @Danelle Cline 👀

Danelle Cline (dcline@mbari.org)

2025-03-11 12:51:23

Great! I will try SAM on our plankton images. This is encouraging.

Jeremy Fix (jeremy.fix@centralesupelec.fr)

2025-03-11 14:08:52

Great !Let me know if this works for you

Santiago Martinez Balvanera (santiago.balvanera.20@ucl.ac.uk)

2025-02-19 08:23:23

Hi Everyone 👋, I have a question for you,

I'm Santiago, and although I've been part of this channel for a while, I've mostly been lurking in the background. I work on AI for bioacoustics and am particularly interested in developing tools that support continual AI development for biodiversity monitoring—integrating everything into a coherent, iterative workflow.

I'm reaching out because I'm collaborating with the 🦇 Bat Conservation Trust (BCT) in the UK to design a public annotation platform that connects data collection with model training and evaluation. Given the expertise in this group, I was wondering if anyone has experience with or knows of systems that handle these kinds of operations. It would be incredibly helpful to discuss what works, what doesn't, and any best practices.

Specifically, I'm thinking about cloud-based and mostly hands-off systems that include:

- Data ingestion & management – Organizing data into a structured database with metadata.
- Expert annotation – Labeling of collected material.
- Automated model training & evaluation – Tracking performance changes over time.
- Model repositories – Managing versions and data sources.
- Post-processing validation – Verifying model predictions.

If you've worked on something similar or have insights to share, I'd love to hear from you! We are focused on audio, but I know there’s a lot to learn from those working with other modalities like camera traps 📷 or remote sensing 🌎. Many of the challenges are likely similar, just with different data, and I'm sure we can benefit from those experiences.

Feel free to reply here or send me a DM!

Thanks in advance!

🦇 Oisin Mac Aodha, Tiziana Gelmi Candusso, Namitha Suresh, Juan Sebastián Cañas, Sofía Miñano, Jason Holmberg (Wild Me)

Elly Knight (ecknight@ualberta.ca)

2025-02-19 10:11:06

Hi Santiago, wildtrax.ca is working on developing tools for ultrasonic annotation. I’d be happy to connect you with the folks leading that project if you like

Santiago Martinez Balvanera (santiago.balvanera.20@ucl.ac.uk)

2025-02-19 10:36:18

Hi Elly! That would be awesome, thanks 🙏! My BCT email is SBalvanera@bats.org.uk

Elly Knight (ecknight@ualberta.ca)

2025-02-19 10:49:32

Great! I’ve sent your contact over to Cami, she’ll reach out soon 🙂

❤️ Santiago Martinez Balvanera

Nicolas Arrieta Larraza (n.arrieta.larraza@gmail.com)

2025-02-20 07:20:16

If it helps, this open source data version control is terrific for data and version management: https://dvc.org/

It's like git for data, basically 😊

dvc.org

Data Version Control · DVC

Open-source version control system for Data Science and Machine Learning projects. Git-like experience to organize your data, models, and experiments.

Original URL: https://dvc.org/

❤️ Santiago Martinez Balvanera

Nicolas Arrieta Larraza (n.arrieta.larraza@gmail.com)

2025-02-20 07:22:52

Also, given that it's a classification task, you might use entropy as a metric to have an estimate of how good your model is doing without human intervention. It might be useful for the post-processing validation part of things.

Santiago Martinez Balvanera (santiago.balvanera.20@ucl.ac.uk)

2025-02-20 08:10:20

Thanks @Nicolas Arrieta Larraza! DVC is great and could be a nice part of the system.

Xiaoxuan (Shirley) Ma (m2mshirley@gmail.com)

2025-02-19 21:39:37

Hello everyone,
We are pleased to announce that we will host the
5th CV4Animals: Computer Vision for Animal Behavior Tracking and Modeling
at CVPR 2025. It will take place in person in Nashville, TN, USA on June 11 or 12, 2025.
The CV4Animals workshop aims to bring together experts from computer vision, neuroscience, biology, animal husbandry, ecology, and animal conservation to advance computer-vision-based animal behavior understanding.

We welcome submissions in 2 tracks:

unpublished work up to 4 pages

papers published within last 2 years

with a submission deadline of March 28, 2025.

For more details, stay updated via our website: CV4Animals
We look forward to an engaging workshop and hope to see many of you there!

Best,
The CV4Animals Organizing Team

cv4animals.com

Towards real-time monitoring of insect species populations

About

Original URL: https://www.cv4animals.com/

❤️ murilo, gyeongsu, Andre Telfer, Ted Schmitt, Yseult Hb, Oisin Mac Aodha, Urs, Mitchell Rogers, Otto Brookes, Alex Rood

🙌 murilo, gyeongsu, Andre Telfer, Nico Lang, Mamoon, Malte Pedersen

🙌:skin_tone_3: Jess Tam

🙌:skin_tone_5: Prabath Gunawardane

🎉 Jenna Kline

Ghjulia Sialelli (ghjulia.sialelli@inf.ethz.ch)

2025-02-20 07:13:52

👋 Hi everyone, I'm Ghjulia, a PhD student at ETH Zurich using AI and remote sensing for above-ground biomass estimation. With some colleagues, we're organizing the AI+Environment Summit, that will take place on October 1st, in Zurich. We don't have a lot of people working on conservation around us, so if you have any speakers/topics/workshops suggestions, feel free to fill in this form: https://forms.gle/AuCtaRDY8EVCzLfw6 ! 🙂

❤️ Samantha Biegel, Sonny Burniston, Sara Si-Moussi, Olivier Dietrich, Nico Lang, Jon Van Oast, Emilia Arens, Taiki Sakai - NOAA Affiliate, Mamoon, Jonathan Sauder, Robin Zbinden, stefano puliti, Lily Xu, Alan Stenhouse, Valerie

💯 Lily Xu

🌍 Lily Xu

Brittany Aguilar (baguilar@schmidtsciences.org)

2025-02-20 07:52:49

Hi Ghjulia, are you looking for more ecology type people, AI for Animals, evolutionary biologists, climate researchers? I have a lot of people in my network but want to send you the right ones

Ghjulia Sialelli (ghjulia.sialelli@inf.ethz.ch)

2025-02-20 08:00:58

we're open to all suggestions! we haven't chosen specific topics yet, so anyone that works with AI fits! and if it's more convenient for you, feel free to send me the recommendations directly instead of the form 🙂

Danielle Celone (daniellecelone@gmail.com)

2025-02-20 14:04:21

Hi everyone! I'm a master's student in computational linguistics at the University of Washington. Although my current training in in NLP, I'm interested in exploring other machine learning applications to conservation, especially within the realms of animal communication and wildfire prevention. I'm excited to learn more from the conversations happening here :)

👋 Andy Viet Huynh, Dan Morris, Maddie Cusimano, Shir Bar, Taiki Sakai - NOAA Affiliate, Sam Lapp, Mamoon, Owen Martin, Arthur Caillau, Jason Holmberg (Wild Me)

🤩 Andy Viet Huynh, Owen Martin, Jason Holmberg (Wild Me)

Brittany Aguilar (baguilar@schmidtsciences.org)

2025-02-20 15:54:48

hey there, hopefully you know about ESP? https://www.earthspecies.org/

Danielle Celone (daniellecelone@gmail.com)

2025-02-21 15:02:44

I do! Im a big fan of their work.

Aakash Gupta (aakash@thinkevolveconsulting.com)

2025-02-22 08:58:29

I was watching the Wild Robot on Disney with my kid today. Seems that will soon be reality

Stephen Formel (sformel@usgs.gov)

2025-02-21 17:49:11

Examples of AI using GBIF and OBIS data

Hello, your friendly US national node manager again 🙂. I'm working on gathering examples of AI using GBIF and / or OBIS data. My goal is to put together some support materials for the utility of GBIF/OBIS data about the United States and GBIF/OBIS use by US scientists. If folks could share publications they think are relevant, that would be great! I'm also happy to collect information about tools, platforms, code packages, etc.

🎉 Jon Van Oast, Owen Martin, Jason Holmberg (Wild Me), Elizabeth Campolongo, Talia Speaker

Åse Håtveit (aase@mit.edu)

2025-02-25 05:24:51

Hi! we are actively using GBIF data with a python package to enable people to train their own AI detection model for insect monitoring 🐛 https://github.com/Tvenver/Bplusplus/tree/package

Publication: https://www.nature.com/articles/s41598-024-68502-8

Note that we have made quiet some changes to the python package since the publication - let me know if you have any questions! 🙂

Nature

Scientific Reports - Towards real-time monitoring of insect species populations

Original URL: https://www.nature.com/articles/s41598-024-68502-8

🔥 Elizabeth Campolongo, Leonardo Buitrago

❤️ Stephen Formel, Leonardo Buitrago

Stephen Formel (sformel@usgs.gov)

2025-02-26 07:21:24

Thank you! To put a finer point on my request, I’m trying to prepare evidence that GBIF participation is a worthwhile investment for the United States. Since AI is a current priority, I want to emphasize any connections between GBIF and AI.

👍 Åse Håtveit, Leonardo Buitrago

👍:skin_tone_3: Alan Stenhouse

Stephen Formel (sformel@usgs.gov)

2025-02-26 07:22:29

Please note that I’m preparing this evidence out of an abundance of caution, not because it has been requested.

🌍 Alan Stenhouse

Lily Xu (lily.x@columbia.edu)

2025-02-22 15:23:23

Kora Labs is a startup based in Zürich, Switzerland developing reinforcement learning algorithms for precision ag. They're doing super cool stuff, and they're hiring for an Earth Observation / ML / ML Ops internship!

https://www.linkedin.com/posts/matteo-turchetta_hiring-activity-7298039670809006081-U[…]m=member_desktop&rcm=ACoAABCUOkwBTYFIoJMiS9LW4QUz9H1mg3s89Sg

Jack Edney (mjackedney@gmail.com)

2025-02-24 16:23:23

⚠️ New AI/ML x Nature Job ⚠️

Hi everyone,

I'm one of the ML engineers at Nature-tech startup Pivotal Future, and I'm excited to say that we are currently looking for talented ML engineers to join our team. 👀

We are developing state-of-the-art AI models for detecting and identifying plant and animal species, for gathering actionable nature intelligence across a wide range of remote data sources

So far we've built:

Multiple image and acoustic detector models across a wide range of taxa groups 🧐

Zero-shot species classification and retrieval models across taxa for multiple image sensor types 📸

A highly scalable data collection methodology and client sites to build unprecedented nature datasets 🌴

The infrastructure to run these at scale... ⚖️

Many more in the pipeline and a lot more on the way... 🔜

But we have a loads more to do... which is why we are looking for talented ML engineers & scientists 💪

We need strong computer vision and/or audio processing experience, experience with deploying models to the cloud and a desire to work with nature! ☁️

The job is fully remote, with plenty of perks, you can apply here:

https://pivotalfuture.bamboohr.com/careers/56

Please feel free to DM me or reach out on my LinkedIn for more information:

https://linkedin.com/in/jackedney

🤞

Pivotal

Machine Learning Engineer (remote role)

Machine Learning Engineer (remote role) Role Overview: Want to solve unique, large scale, highly complex technical problems that improve outcomes for nature? Pivotal is seeking an experienced Machine Learning Engineer to play a crucial role in the development of solutions that scale machine learning to

Original URL: https://pivotalfuture.bamboohr.com/careers/56

uk.linkedin.com

Jack Edney - Pivotal Future | LinkedIn

Hi, I&#39;m Jack. I am an AI developer with a specialism in computer vision. I am passionate… · Experience: Pivotal Future · Education: King&#39;s College London · Location: Winchester · 440 connections on LinkedIn. View Jack Edney’s profile on LinkedIn, a professional community of 1 billion members.

Original URL: https://linkedin.com/in/jackedney

❤️ Sara Beery, Céline Angonin, Jason Holmberg (Wild Me), Alexander Merdian-Tarko, Alan Stenhouse

👍 Marzi

Sara Beery (sara.beery@gmail.com)

2025-02-24 18:46:17

Sharing from Capgemini, new water forecasting challenge/benchmark:

Capgemini Invent Lab is launching an international Data Science Hackathon that aims to create solutions to rethink water management.

The purpose ? Develop models to predict water flow within a watershed and optimize water allocation between different economic players in the most sustainable way for ecosystems.

🔗 Hackathon takes place on Codabench: Here
🚀 A hackathon launch key note took place on Feb, 18th. The replay is available online.
📅 Starts Now ! 100% online | Ends in May 2025
🏆 3 Prizes for the 3 winning teams : :firstplacemedal: 3 000 € | :secondplacemedal: 1 500 € | :thirdplacemedal: 500 €

💧 4 reasons to take part:

Explore innovative predictive models for estimating water flows.

Deepen your knowledge by sharing experiences with experts in hydrology, data science and climatology.

Make an impact on the future by contributing to sustainable and accessible open source solutions.

Add value to your work with the unique opportunity to publish alongside the scientific committee and organizing team.

👉 Register here for more information ! Feel free to share ! 😊

For any question, please contact hackathonwaterscarcity@capgemini.com or directly in the Forum Page.

YouTube

Capgemini France (https://www.youtube.com/@capgemini-france)

Hackathon Water Scarcity – Launch key note

Original URL: https://youtu.be/cgm_XCAmPjI

Capgemini France

Hackathon Water Scarcity Hackathon

Imaginez, codez et créez des solutions pour repenser la gestion de l’eau. Ce Hackathon international est l’opportunité de contribuer à un avenir durable et de repousser les frontières de la Data Science.

Est. reading time

8 minutes

Original URL: https://www.capgemini.com/fr-fr/actualites/evenements/hackathon-water-scarcity/#en

❤️ murilo, Xiaoxuan (Shirley) Ma

🙌 murilo, Mamoon

🌊 murilo, Jess Tam, Aakash Gupta

Raj (therajsmile@gmail.com)

2025-02-24 22:17:22

Dear All,
I am trying YOLOv11 for detection and tracking fish in my experiments. So far, it’s giving good results once I trained the model with large number of custom images datasets, however, still need to correct tracking data manually once the tracking data exported into csv file format.
Does anyone has experience how to correct the object ids once the data exported?
Tried to search but couldn’t find a reliable info so would appreciate any inputs here. Thanks in advance.

DEV MEHROTRA (dmehrotra@wisc.edu)

2025-02-24 23:51:59

Can you explain more about what do you mean by correcting the object ids? Are they in the wrong order or format?

Raj (therajsmile@gmail.com)

2025-02-25 01:08:46

Thanks @DEV MEHROTRA, I meant that I successfully ran the tracker and got the tracking ids of fish I used, but some frames got re-assigned or missed or swapped, for example, fishid 1 got new id as 21 from frame number 151 or so. And then it kept continue as fishid 21 instead of 1 as originally. Now I want to correct fishid 21 as fishid 1 but do not know how to do for large number of frames.
Just to make clear that, it’s an example and I have large number of data like thousands of frames in which manual correction for all frames one-by-one in csv file is not possible.
So what I was asking if there is a tool where I can correct this id at frame 151 (from this frame got new id) and then it can correct the following ids for later frames. Thanks

Chaim (c.elchik123@gmail.com)

2025-02-25 01:51:48

@Raj I have done something very similar in my master thesis you can have a read and check out my method for solving this problem. https://scripties.uba.uva.nl/search?id=record_54895 If you have any questions feel free to ask.

scripties.uba.uva.nl

Scripties - Bibliotheek - Universiteit van Amsterdam

Scripties van studenten van de Universiteit van Amsterdam.

Original URL: https://scripties.uba.uva.nl/search?id=record_54895

Filippo Varini (fppvrn@gmail.com)

2025-02-25 01:53:07

@Raj we do something similar in SharkTrack https://github.com/filippovarini/sharktrack?tab=readme-ov-file#sharktrack

We accept that ids might be swapped around, but have a process to manually map id to species/individuals, so ecologica metrics (maxN) are not affected

Filippo Varini (fppvrn@gmail.com)

2025-02-25 01:54:32

Btw, we are collecting fish datasets for a universal fish detector!
if you have any datasets that can be shared, it would be amazing if you could include them!

Filippo Varini (https://aiforconservation.slack.com/team/U0660QDJAFL)

Help Build the Most Comprehensive Open-Source Fish Detection Model! 🐟🌊 Hey everyone! With @Oc and @Dan Morris, we are launching a community-driven project to create a universal fish detector, an AI model capable of detecting any fish-like creature, anywhere! 🦈🐠🐡 A ton of fish detection datasets are publicly available online, but they’re scattered across different formats. To train a powerful model, we need to gather, analyze, and standardize these datasets—and that’s where you come in! :female_technologist: How can you help? We are keeping a list of fish datasets to process <a href="https://docs.google.com/spreadsheets/d/1hd24tHF1cYDses96-iPREuf0NR9mOGHUb0TKok4ME7M/edit?gid=0#gid=0">here</a>. Choose a dataset, analyze its metadata, and create a notebook to visualize annotated images. 💡 Why join? • Create an AI model that disrupts ocean conservation like the <a href="https://github.com/agentmorris/MegaDetector">MegaDetector</a> model did on land 🚀. • Gain hands-on experience with computer vision & dataset curation 🔬 • Get credited as a contributor to a high-impact conservation effort 🙌 📌 Start here: <a href="https://www.notion.so/MegaFishDetector-Dataset-Creation-18bb4b3047ee80e197cfdda8cabe096a?pvs=21">Read this Notion guide to start contributing to the project!</a> Additional resources: 📝 Tracking sheet: <a href="https://docs.google.com/spreadsheets/d/1hd24tHF1cYDses96-iPREuf0NR9mOGHUb0TKok4ME7M/edit?gid=0#gid=0">Dataset Excel</a> 📂 GitHub repo: <a href="https://github.com/filippovarini/fish-datasets">Fish Datasets</a> Any questions? Reach out to us on slack or email at <a href="mailto:fppvrn@gmail.com">fppvrn@gmail.com</a>. Let’s build this together! 🛠️💙 P.S. Let us know if we missed any dataset, by <a href="https://forms.gle/Fa6oWfZtd7cYA9de8">filling this form</a>!

Original URL: https://aiforconservation.slack.com/archives/CLWGQ4BJ6/p1739380771104069

Raj (therajsmile@gmail.com)

2025-02-25 02:48:22

Thanks for your input @Chaim, but would appreciate if you could just briefly state here what tools or methods did you use. It would be easy for others too who look for similar approaches. Thanks.

Raj (therajsmile@gmail.com)

2025-02-25 02:50:18

Thanks for your input and suggestion for SharkTrack @Filippo Varini, I will surely look into it, though I am using way too much smaller fish, just about 3-5 cm. Thanks anyway.

Filippo Varini (fppvrn@gmail.com)

2025-02-25 02:54:58

Ah fair. Maybe worth looking into TAPIR? https://deepmind-tapir.github.io

deepmind-tapir.github.io

TAPIR: Tracking Any Point with per-frame Initialization and temporal Refinement

TAPIR provides fast and accurate tracking of any point in a video

Original URL: https://deepmind-tapir.github.io

Chaim (c.elchik123@gmail.com)

2025-02-25 03:02:13

@Raj
The method is post model tracking and uses a pandas dataframe of objects detected with their ID per frame as initial input. Fully made in Python.

In short it works the following way:

Re-Identification (Solving ID Switching) Solution

Identify disappearance & reappearance points.

If disappearance is near an edge check if reappearance is at the same edge.

Calculate max possible travel distance.

Compare similarity between disappearing & reappearing objects.

Assign IDs based on highest similarity + lowest distance.

Use a similarity threshold based on some experimentation. The similarity is visual similarity so you can use either the entire frame if you are dealing with a moving camera. Or of just the bounding box area if using a static camera. You can analyze movement patterns to calculate max distance and so on.

Raj (therajsmile@gmail.com)

2025-02-25 03:31:28

Thank you so much for your input @Chaim, greatly appreciated. I’ll certainly try it out. Thanks a ton.

Chaim (c.elchik123@gmail.com)

2025-02-25 03:32:34

@Raj No problem, if you have any questions feel free to send me a message. Currently working on using this method on re Identifying Elephants in drone videos.

Raj (therajsmile@gmail.com)

2025-02-25 03:58:03

Thanks a lot @Chaim, surely will reach out to you if any doubts happen.

Raj (therajsmile@gmail.com)

2025-02-26 21:59:43

Thanks @Filippo Varini for sharing TAPIR as well, surely will prioritize to look into it, Thanks.

🙌 Filippo Varini

Raj (therajsmile@gmail.com)

2025-02-26 22:00:44

Thanks once @Chaim for your kind reply, surely will connect with you if found any difficulties to do the analysis, thanks a ton.

Filippo Varini (fppvrn@gmail.com)

2025-02-25 01:56:34

Hi everyone, have you experimented with foundational models (SAM/CLIP/BioCLIP/BLIP/Inquire etc) for ecological computer vision applications?

What has your experience been so far?

👀 Ben Williams, Jason Holmberg (Wild Me), Alan Stenhouse, Jane Wu

Filippo Varini (fppvrn@gmail.com)

2025-02-25 01:57:32

On my side, I have been playing around with BioCLIP. It’s surprisingly accurate at the phylum/class level, but I noticed you need to crop out organisms and might struggle for finer-grained taxonomy

🙌 Sam Stevens

Roberta Hunt (r.hunt@di.ku.dk)

2025-02-25 02:15:23

I played around with SAM for wing cell segmentation and it gave a decent initial result. It was a little difficult that you needed to give it coordinates, but since our images were pretty well standardized it worked okay as a first quick pass.

Filippo Varini (fppvrn@gmail.com)

2025-02-25 02:56:06

@Roberta Hunt makes sense! By coordinates do you mean the (point or bounding box) prompts that you need to give SAM?
Did you try the AutomaticMaskGenwrator?

👍 Roberta Hunt

Roberta Hunt (r.hunt@di.ku.dk)

2025-02-25 03:08:36

Yes, I mean the points.
Since it wasn’t obvious from the images which cells we were interested in the auto generation didn’t work so well unfortunately. But looking forward to playing with models which combine textual understanding and segmentation in the future.

Filippo Varini (fppvrn@gmail.com)

2025-02-25 03:09:57

Makes sense, thanks for sharing your experience!

Arthur Caillau (arthur@caillau.me)

2025-02-25 06:25:04

I have some experience with SAM and BioCLIP. SAM works great when paired with something like an object detector to build some mask datasets that I tend to use as a training set for other models. BioCLIP did not seem to be worth in the last experiment we ran. It was quite inaccurate on our dataset.

Kate Rakelly (krakelly@gmail.com)

2025-02-25 16:55:09

I’m experimenting right now with groundingDINO to detect animals in camera trap data. https://github.com/IDEA-Research/GroundingDINO
Very preliminarily and qualitatively, it seems on par with MegaDetector in terms of finding animals.

Aakash Gupta (aakash@thinkevolveconsulting.com)

2025-02-26 06:32:47

I have tried Florence2 and it worked great in identifying animal species as well pointing out their behavior characteristics (grazing, drinking, resting..etc.)

Sam Stevens (stevens.994@buckeyemail.osu.edu)

2025-02-26 12:43:23

@Arthur Caillau I worked on bioclip and am developing a set of benchmarks for evaluating foundation models (https://github.com/samuelstevens/biobench). Can you share any details about your specific use case for bioclip? I would love to include it as a specific evaluation for our team as we develop the next version

samuelstevens/biobench

Computer vision benchmark for evolutionary biology-related tasks.

Website

<http://samuelstevens.me/biobench/>

Stars

Arthur Caillau (arthur@caillau.me)

2025-02-26 14:27:50

Hey Sam! My use case was to combine BioCLIP with MegaDetector to analyze camtrap data to automatically label detected animals down to the species level. It fails pretty hard on very simple things like confusing birds and cows.

👍 Sam Stevens

Filippo Varini (fppvrn@gmail.com)

2025-02-26 14:43:24

@Sam Stevens thanks for putting together the Benchmark, that is very helpful!

Filippo Varini (fppvrn@gmail.com)

2025-02-26 14:43:59

@Aakash Gupta super interesting you used Florence2 - did you fine tune it or did it classify behavior in zero-shot?

Aakash Gupta (aakash@thinkevolveconsulting.com)

2025-02-26 22:08:47

this was zero-shot behaviour.

Filippo Varini (fppvrn@gmail.com)

2025-02-27 05:48:16

impressing!

Aakash Gupta (aakash@thinkevolveconsulting.com)

2025-02-28 06:45:55

results of some of the inference. I have highlighted the one where it was not working properly. but it can be worked around, with some post-processing. Other behaviour like eating a carcas and mating were also identified.(cant share those images here)

Filippo Varini (fppvrn@gmail.com)

2025-02-28 08:23:42

👌 👌

Ben Williams (ben.williams.20@ucl.ac.uk)

2025-02-25 07:29:10

Hi all, I've got several hundred gigabytes of coral reef soundscape data from across the globe that I’m hoping to open-source alongside a paper submission. The dataset includes useful metadata (e.g habitat type, coordinates, timestamps etc) and fish sound detections from our classifiers.

I see several possible routes for making this available to the community, but each has some downsides. For example:

Zenodo or Figshare: Would require splitting the data into many smaller sub-repositories.

Dryad: Involves a small fee and requires a CC0 license which means attribution is not required (attribution is important to our partners who contributed data).

Schemes like the AWS Open Data Sponsorship Programme: Requires an application and approval process.

Essentially, I’m looking for the path of least resistance to get this data online. If anyone has any experience with the above or other recommendations I'd love to know! Thank you in advance 🙏

❤️ Owen Martin, Jess Tam, Abigail Oppong

💚 ash, Marius Miron, Alan Stenhouse

Stephen Formel (sformel@usgs.gov)

2025-02-25 08:22:31

So you’re looking to to share the raw data and the inferred occurrences? Any licenses that need to be respected on any of the data?

Stephen Formel (sformel@usgs.gov)

2025-02-25 08:25:30

Regardless of what you decide, I would be happy to help you publish the inferred occurrences to OBIS and GBIF, where they can be more easily used with other data. If they are already well annotated with metadata and in machine actionable formats, it should be relatively painless.

Stephen Formel (sformel@usgs.gov)

2025-02-25 08:26:00

NCEI would probably be interested in the raw data, I could put you in touch with people there who can help.

Stephen Formel (sformel@usgs.gov)

2025-02-25 08:32:45

If you do end up publishing to Dryad or Zenodo (for easiness) it would still be great if you could share the data on OBIS and GBIF, since it sounds like it’s really a wealth of knowledge.

👍 Leonardo Buitrago

Brian Geuther (Brian.Geuther@gmail.com)

2025-02-25 09:23:01

Dataverse has a handful of hosting locations (e.g. https://dataverse.harvard.edu/) which have different licensing and costs options depending upon who's depositing it. The number has changed over the years, but currently is free for totals below 1TB with "custom licenses" (generally any open license, but prefer CC0) for "researchers".

Matt Weldy (matthewjweldy@gmail.com)

2025-02-25 09:44:25

Zenodo will sometimes lift the file size limits for some repositories. You need to reach out and make the request though.

👍 Vincent Lostanlen

Jess Tam (j.tam@unsw.edu.au)

2025-02-25 20:04:30

Splitting them into smaller chunks on Zenodo might actually be a good idea for users who only want to do some small-scale explorations (if you don't mind the multiple uploads). You could also split them by country / habitat / year / etc. You can link them all in your github repo with download instructions, license, etc.

👍 Benjamin Algreen Adler, Aakash Gupta

Juan Sebastián Cañas (jscanass@gmail.com)

2025-02-26 04:43:12

Maybe @Dan Morris could help you with https://lila.science/

LILA BC

LILA BC (Labeled Image Library of Alexandria: Biology and Conservation)

Original URL: https://lila.science/

Aakash Gupta (aakash@thinkevolveconsulting.com)

2025-02-26 06:34:57

Have you checked Kaggle and HuggingFace datasets

Ben Williams (ben.williams.20@ucl.ac.uk)

2025-02-26 07:26:57

Thank you all, some useful suggestions! Any others are welcome

@Stephen Formel The NCEI could be a good home given its environment focus, although the 81 page approval handbook is a bit overwhelming 👀 If you have any contacts who could share the practical details that would be useful, either DM or email ben.williams.20@ucl.ac.uk
I'll be working on curating the detections next week so will add to my to do list to put together a short summary on this so you can consider if and how this could work for OBIS/GBIF!

@Brian Geuther - I was not aware of this, looks like a great option if one of the environmental specific ones doesn't work!

Re Zenodo - I am considering the split across multiple repo's option, though I've done this in the past and found it tricky as uploads fail a lot and its not as accessible to users who end up having to navigate multiple repo's. Its definitely my back up, but ideally I think it would be good to have it all in one place under the same DOI, where the data can still be split into sections (e.g a zip per site). Simply asking for an expanded repo might be nice like you say @Matt Weldy, have you ever tried this?

Thanks @Juan Sebastián Cañas, in my ignorance I hadn't realised LILA hosted raw data, think I've always skipped to the 'other' datasets for audio in the past! @Dan Morris if audio (raw unlabeled data + detections) are suitable then it could definitely be a nice home due to its environmental focus. I could share the summary with more details that I'll put together for Stephen if so?

@Aakash Gupta looks like Kaggle and Hugging Face have a 200gb and 300gb limit respectively. Though it may work depending how much the data compresses down. I think it might be nice to find a home where it sits amongst other scientific or ideally environmental data.

Dan Morris (agentmorris@gmail.com)

2025-02-26 10:32:09

@Ben Williams I prefer to keep LILA focused on labeled data. There's a gray area; if you had 50GB of strongly-labeled data and you thought another 200GB of soundscape data really added utility to the labeled data, I would consider it, but it sounds like this is an unlabeled dataset in spirit, so LILA is probably not the best fit.

If you used Zenodo, but the real home for documentation was, e.g., on GitHub, maybe you could ease the hassle of downloading from multiple datasets by providing .sh and .bat files to download and unzip all the relevant datasets? I'm not sure how Zenodo feels about that, but overall Zenodo seems like a good fit other than the size limit.

Matt Weldy (matthewjweldy@gmail.com)

2025-02-26 12:42:34

@Ben Williams I have not had to ask for an extended repository size yet. However, I looked into it for a dataset we are preparing.

I've also found that the uploads to zenodo fail quite often. I have found that breaking up datasets into smaller zipped chunks and then uploading them into the same repository was a bit easier (less frustrating) than trying to upload a 15GB file and having it fail at the very end.

Stephen Formel (sformel@usgs.gov)

2025-02-27 09:44:47

@Ben Williams yes, the NCEI handbook is a little over the top. I’ll send you an email, but I’m also going to share the info here in case it is useful to others:

NCEI

To submit data to the archive, contact pad.info@noaa.gov. Data submitted to the archive are “packaged” using PassivePacker, a data packaging and metadata gathering software tool. This archive accepts data from stationary marine, mobile marine, and terrestrial platforms.

If they don’t respond, you can reach out to Carrie Wall Bell, who is the ‘Fisheries Acoustics Archive Lead’: carrie.wall@noaa.gov

OBIS and GBIF

An example of data in GBIF and OBIS is the SanctSounds data we helped publish. The links on GBIF and OBIS are:

https://www.gbif.org/dataset/b3fc1d76-a50d-4a57-a2f3-425adfc59f9f

https://obis.org/dataset/7a4427f6-67ee-4cc1-b95f-3045523420a1

Notebooks about the crosswalk to the Darwin Core standard and the publishing process are here: https://github.com/ioos/bio_data_guide/tree/main/datasets/SanctSound

Inferences from acoustic data are still a newish frontier for publishing. If you don’t see how your data fits, please don’t walk away! I’d be happy to demonstrate how the standard is flexible, or discuss how to evolve the standard to accommodate your needs. After you’ve had a chance to peruse these materials, let’s get together and chat to see what works and what doesn’t. You (or anyone) can email me at sformel@usgs.gov

National Centers for Environmental Information (NCEI)

PassivePacker

PassivePacker is a data packaging and metadata gathering software tool NCEI developed to simplify data submission preparation for passive acoustic data.

Original URL: https://www.ncei.noaa.gov/node/2075

National Centers for Environmental Information (NCEI)

SanctSound: Studying the Underwater World of Sound

NOAA, the U.S. Navy, and partners launched the Sanctuary Soundscape Monitoring Project portal to listen and learn about underwater sounds.

Original URL: https://www.ncei.noaa.gov/news/sanctsound-studying-underwater-world-sound

gbif.org

Sanctuary Soundscape Monitoring Project (SanctSound) Daily Aggregated Species Detections

NOAA and the U.S. Navy are working to better understand underwater sound within the National Marine Sanctuary System. From 2018-2021, these agencies worked with numerous scientific partners to study sound within seven national marine sanctuaries and one marine national monument, which includes waters off Hawai‘i and the east and west coasts. Standardized measurements were used to assess sounds produced by marine animals, physical processes (e.g., wind and waves), and human activities. Collectively, this information will help NOAA and the Navy measure sound levels and baseline acoustic conditio…

Original URL: https://www.gbif.org/dataset/b3fc1d76-a50d-4a57-a2f3-425adfc59f9f

obis.org

Sanctuary Soundscape Monitoring Project (SanctSound) Daily Aggregated Species Detections - Ocean Biodiversity Information System

Original URL: https://obis.org/dataset/7a4427f6-67ee-4cc1-b95f-3045523420a1

🙌 Ben Williams

Sam Lapp (sam.lapp@pitt.edu)

2025-02-27 12:33:43

Just chiming in that I don’t know the answer but if you find a solution, please make a new post sharing your experience! This is something myself and plenty of others have wondered about as well.

👌 Ben Williams

➕ Abram Fleishman

Ben Williams (ben.williams.20@ucl.ac.uk)

2025-02-27 13:31:36

Thank you all 🙏
@Stephen Formel this definitely sounds like a good option, thanks for the email, I'll be in touch!

@Dan Morris agree it sounds a bit out of scope for LILA given those reqs, will definitely put a script together to access all the data if I fall back to the zenodo route

@Matt Weldy I've had a similar experience:
i) Large files tend to fail during upload
ii) Uploading many smaller zip files still results in many failures
iii) To handle (ii), I have to awkwardly check which failed and attempt to reupload these
iv) Repeat (iii) again with the new failers

@Sam Lapp will be sure to! Sounds like the NCEI repo suggested by Stephen could work for my use case given its ocean data, if no luck there I'll be checking the harvard dataverse suggestion next

Leonardo Buitrago (leobuitragoh@gmail.com)

2025-03-03 18:39:16

Hi Ben and colleagues!
This is my first post on this Slack channel—thanks for the invitation, by the way! I’m Leonardo from Colombia, currently part of the GBIF regional support team for Latin America and the Caribbean. Your question was very interesting to me, and I completely agree with @Stephen Formel’s response about using GBIF as an effective mechanism to share your dataset. My technical expertise focuses on the steps to mobilize data through GBIF, so I’d be happy to assist you or anyone else on this matter.
As @Stephen Formel mentioned in another thread, I’m also very interested in the links between GBIF and AI applications—both in terms of using GBIF data but also facilitating data integration and quality processes to enhance its usability.
I look forward to continuing this conversation in person in Medellín or answering any GBIF-related questions here for anyone who’s interested. 😉

❤️ Stephen Formel, Shir Bar, Sara Beery

Abram Fleishman (abram@conservationmetrics.com)

2025-03-14 14:11:04

How timely, Just this week I made a list of some online data repos, with very very brief notes about each. I am looking to archive ~3.8 TB of maine underwater recordings from Moorea with an open license with attribution. All these seem to fall short of what I need but maybe this list will be helpful to others with smaller datasets. @Ben Williams let me know what you end of doing as I think we would probably want to follow a similar path!

Also a question: has anyone ever uploaded data to the Internet Archive?

Here is my little list

Zenodo allows up to 50GB per data set but you can request a 1-time increase to 200GB for a single large dataset. https://help.zenodo.org/docs/deposit/manage-files/

Data Dryad allows files up to 10GB and a total of 300GB per “publication” https://datadryad.org/requirements

SEANOE limit of 100GB https://www.seanoe.org/html/publish-your-data.htm

Kaggle allows 200GB https://www.kaggle.com/docs/datasets#technical-specifications

OSF limit 50GB https://help.osf.io/article/137-osf-storage-caps

Harvard Dataverse limit 1 TB but might need to pay depending on (...not sure)? https://support.dataverse.harvard.edu/researchers

Mendeley Data 100GB (if institutional subscriber) https://data.mendeley.com/faq?q=size%20limit

Figshare private accounts allow 20 GB but some institutional accounts allow up to 5TB https://info.figshare.com/user-guide/uploading-and-managing-files/#file-size

Open data repository 1 GB lol https://www.opendatarepository.org/odr-cloud-services/

Internet Archive no limits specified https://archive.org/

🙌:skin_tone_4: Oc

Leonardo Buitrago (leobuitragoh@gmail.com)

2025-03-17 17:12:04

Hi @Abram Fleishman,
Thanks for the list you created– it's very useful!
I have helped some GBIF users upload images to Internet Archive (e.g. https://archive.org/details/JBGP-HERBARIO_VIRTUAL) to link them to specimens records published on GBIF (e.g. https://www.gbif.org/occurrence/4139010318). The process is quite simple, though I haven't tried with other data types.

Regarding Internet Archive limits, its help center states "Currently, there is no limit on the size of files nor the number of files. However, from a systems perspective, we do not recommend files larger than 50 GBs to be uploaded or more than 1000 files, per single page." You can find more details in their uploading tips here https://help.archive.org/help/uploading-tips/.

Eugene Galaxy (eugene@animaldetect.com)

2025-02-27 03:32:43

Hey everyone,

❗ 3 MORE SPOTS TO https://animaldetect.com CLOSED-TEST PHASE. ❗

On Monday, we are starting the test phase of our first MVP version!
We are looking for a limited number of testers and still have 3 places left.
Just write to me if you would like a spot reserved for you. Only your email is needed.

The test phase will continue for two weeks. Join us! 😄

Eugene Galaxy (eugene@animaldetect.com)

2025-02-27 06:36:10

2 spots left

Filippo Varini (fppvrn@gmail.com)

2025-02-27 15:08:19

Hi all, does anyone have experience with AI for plankton ID?
We are starting to explore the space and would really appreciate some advice on promising approaches and pitfalls to avoid!
We are interested in classifying phytoplankton and zooplankton from Flowcam imaging

Jeremy Fix (jeremy.fix@centralesupelec.fr)

2025-02-27 15:16:12

Hi , you may find interesrting to get in touch with people such as Jean Olivier Irisson https://lov.imev-mer.fr/web/member/jean-olivier-irisson/. They do have automated pipelines and a lot of annotated data on zooplankton , as far as I can tell. I'm not a specialist on this topic. I did use some of their data to build a challenge for my students (https://www.kaggle.com/competitions/3md4040-2025-challenge/overview). The device from which I used their data is a Zooscan. But they definitely also have flowcam data.

kaggle.com

3MD4040 - 2025 Challenge

This competition is about segmenting living organisms from non-living debris in large ZooScan images captured at Point B.

Original URL: https://www.kaggle.com/competitions/3md4040-2025-challenge/overview

Filippo Varini (fppvrn@gmail.com)

2025-02-27 15:27:57

Thank you so much @Jeremy Fix

Ben Weinstein (benweinstein2010@gmail.com)

2025-02-27 18:21:03

@Eric Orenstein

Laura Chrobak (lchrobak@mbari.org)

2025-02-28 12:54:42

Two quick thoughts- this dataset might be relevant ecotaxa and Ben Grassian from WHOI gave a presentation at the Marine Imaging Workshop last year on this topic. Hope this helps!

Laura Chrobak (lchrobak@mbari.org)

2025-02-28 13:01:21

Also, Patrick Daniel at UCSC has worked on few projects related to this too: https://essd.copernicus.org/articles/16/2971/2024/essd-16-2971-2024.html

First release of the Pelagic Size Structure database: global datasets of marine size spectra obtained from plankton imaging devices

Abstract. In marine ecosystems, most physiological, ecological, or physical processes are size dependent. These include metabolic rates, the uptake of carbon and other nutrients, swimming and sinking velocities, and trophic interactions, which eventually determine the stocks of commercial species, as well as biogeochemical cycles and carbon sequestration. As such, broad-scale observations of plankton size distribution are important indicators of the general functioning and state of pelagic ecosystems under anthropogenic pressures. Here, we present the first global datasets of the Pelagic Size Structure database (PSSdb), generated from plankton imaging devices. This release includes the bulk particle normalized biovolume size spectrum (NBSS) and the bulk particle size distribution (PSD), along with their related parameters (slope, intercept, and R2) measured within the epipelagic layer (0–200 m) by three imaging sensors: the Imaging FlowCytobot (IFCB), the Underwater Vision Profiler (UVP), and benchtop scanners. Collectively, these instruments effectively image organisms and detrital material in the 7–10 000 µm size range. A total of 92 472 IFCB samples, 3068 UVP profiles, and 2411 scans passed our quality control and were standardized to produce consistent instrument-specific size spectra averaged to 1° × 1° latitude and longitude and by year and month. Our instrument-specific datasets span most major ocean basins, except for the IFCB datasets we have ingested, which were exclusively collected in northern latitudes, and cover decadal time periods (2013–2022 for IFCB, 2008–2021 for UVP, and 1996–2022 for scanners), allowing for a further assessment of the pelagic size spectrum in space and time. The datasets that constitute PSSdb's first release are available at <https://doi.org/10.5281/zenodo.11050013> (Dugenne et al., 2024b). In addition, future updates to these data products can be accessed at <https://doi.org/10.5281/zenodo.7998799>.

Original URL: https://essd.copernicus.org/articles/16/2971/2024/essd-16-2971-2024.html

Filippo Varini (fppvrn@gmail.com)

2025-02-28 16:55:31

Thank you @Laura Chrobak!

Isabel Fenton (ifenton@turing.ac.uk)

2025-03-03 04:39:18

Maybe https://www.frontiersin.org/journals/marine-science/articles/10.3389/fmars.2024.1513463/full is useful? Work done by @Noushin Eftekhari

🙏 Noushin Eftekhari

👀 Noushin Eftekhari

🙌 Noushin Eftekhari

Urs (urs.waldmann@uni-konstanz.de)

2025-03-03 13:39:13

A former colleague of mine is collaborating with biologists to classify plankton from microscope images. I don't know details but it seems to me that it might be the same capturing device. So, it might be worth reaching out to him about your project: https://connect.helmholtz-imaging.de/expert/176

Noushin Eftekhari (neftekhari@turing.ac.uk)

2025-03-07 06:31:14

Hi @Filippo Varini , I would be happy to help. Please feel free to reach out. Here is also a colleague of mine at NOC who is also working on zooplankton: https://github.com/Mojtabamsd/Plankton-Fusion

Mojtabamsd/Plankton-Fusion

Stars

Language

Python

Filippo Varini (fppvrn@gmail.com)

2025-03-07 12:33:43

Thank you @Noushin Eftekhari!!

Serge Wich (sergewich@gmail.com)

2025-03-20 03:35:31

At this lab they are working on this as well: https://www.mba.ac.uk/what-we-do/our-science/ocean-biology/?_gl=1d73zai_upMQ.._gaMjA4MTk0Nzc1My4xNzQyNDU1OTkw_ga_PXG31YFK9X**MTc0MjQ1NTk5MC4xLjEuMTc0MjQ1NTk5OC4wLjAuMA..

Marine Biological Association

Ocean Biology | Marine Biological Association

The ocean plays a central role in mitigating climate change. MBA researchers focus on the impact this has on ocean biology. Find out more!

Time to read

2 minutes

Original URL: https://www.mba.ac.uk/what-we-do/our-science/ocean-biology/?_gl=1*d73zai*_up*MQ..*_ga*MjA4MTk0Nzc1My4xNzQyNDU1OTkw*_ga_PXG31YFK9X*MTc0MjQ1NTk5MC4xLjEuMTc0MjQ1NTk5OC4wLjAuMA

Yihang She (yihshe@student.ethz.ch)

2025-02-28 09:12:25

This list of upcoming schools may be of interest to people doing ML-related stuff https://ellis.eu/news/ellis-winter-summer-schools-2025-schedule

European Lab for Learning & Intelligent Systems

Travel and study in Europe: 2025 Schedule of ELLIS PhD Winter and Summer Schools announced

The ELLIS mission is to create a diverse European network that promotes research excellence and advances breakthroughs in AI, as well as a pan-European PhD program to educate the next generation of AI researchers. ELLIS also aims to boost economic growth in Europe by leveraging AI technologies.

Original URL: https://ellis.eu/news/ellis-winter-summer-schools-2025-schedule

❤️ Angela Zhu

Tiffany Vlaar (Tiffany.Vlaar@glasgow.ac.uk)

2025-02-28 12:14:31

The Leverhulme Programme for Doctoral Training in Ecological Data Science is offering interdisciplinary fully-funded PhD positions at the University of Glasgow for an October 2025 start.
In year 1, students undertake a group rotation project in 3 projects with different supervisors, before selecting a single project as their PhD topic.
Projects include:

What do wild insects do? Leveraging AI to describe wild insect behaviour and to uncover their functions for fitness

Text mining with large language models to enhance infectious disease ecology

Using animal movement data to understand and predict human-wildlife conflicts

Collective animal movement and resource selection in changing environment

Extreme value theory for predicting animal dispersal and movement in a changing climate

Leveraging large language models to provide insights into global plant biodiversity

Implications of mosquito community composition changes for vector control

The impact of deep learning optimization and design choices for marine biodiversity monitoring

Computationally efficient inference methods for integrated spatial capture-recapture and movement models

All information can be found on: https://ecological-data-science.github.io
Deadline for applications is 5pm (GMT) on 7 March.

📣 Oisin Mac Aodha, Sara Beery, Valerie

Laura Chrobak (lchrobak@mbari.org)

2025-02-28 13:51:37

Optimizing the Use of Low-Quality Images in Training Data 🌊

I’m posting because some of my colleagues at the Monterey Bay Aquarium Research Institute (MBARI) are tackling a compelling (and likely not uncommon) 📸computer vision challenge in this space: how to optimally include low-quality images in model training.

📎 I’ve attached an example of some of the ROIs they’re working with to illustrate the challenge. Essentially, some verified species detections are clear and easily identifiable, while others are blurry or more ambiguous. They’re interested to know the optimal ratio of high- vs. low-quality images when training their models to balance accuracy and real-world robustness.
They’ve also experimented with automated methods to quantify image quality (such as Laplacian variance, Sobel filtering) but these haven’t been particularly effective at distinguishing between high vs. low quality examples.

If anyone has insights, relevant literature, or even just ideas worth exploring, I’d love to hear your thoughts. Thanks in advance!

🙌 Sara Beery, Suzanne Stathatos, Taiki Sakai - NOAA Affiliate, murilo, Ben Weinstein, Cameron Trotter

👏 Malte Pedersen, murilo

🐟 murilo

❤️ murilo

Klára Janoušková (Klara.janou@gmail.com)

2025-02-28 14:05:07

It seems somewhat related to the work we are doing on independent FG-BG modelling. The simplified intuition is that if the object is clearly visible (for us that means we can detect/segment it and the FG prediction is obtained by cropping the object), the foreground should be used to make the decision (possibly aided by the context=BG in difficult, ambiguous cases), while when it is not, you are left with the image context to make your decision.

Then you do not really look for the optimal ratio of these during training to learn a single model for all the cases, because we learn a different model for FG (high quality) and a different one for BG (low quality), ...

I am not sure how much the detection/segmentation viewpoint is applicable here but maybe this gives a slightly different perspective at least :)

DEV MEHROTRA (dmehrotra@wisc.edu)

2025-03-02 14:54:42

I have used self supervise learning with foundation models to encode the images in latent space and then use dimensionality reducing techniques to determine the quality of the images. It's a cheap way to determine if the images are any good or not and a even better way for comparisons/ablation studies.

Ben Weinstein (benweinstein2010@gmail.com)

2025-03-03 12:54:22

agree with @DEV MEHROTRA, I've asked similar ideas, even literally asking the LLM to tell me if its clear or fuzzy.

Laura Chrobak (lchrobak@mbari.org)

2025-03-03 12:57:02

Thanks @DEV MEHROTRA and @Ben Weinstein those sounds like great strategies to try out!

❤️ DEV MEHROTRA

Ben Weinstein (benweinstein2010@gmail.com)

2025-03-03 13:00:23

in general, i'm looking for ideas on how we integrate LLMs directly into the preprocessing organization parts of the workflow, which I think are under appreciated for our users https://deepforest.readthedocs.io/, report back if you end up exploring this more, what LLM did you use, how much code did you write, did you go through an API (did you pay for it), and how we could automate such a process for users. I have google of summer of code students interested in this kind of thing.

➕ Rachel Mason

Ed Miller (ed@hypraptive.com)

2025-03-06 22:56:30

I am also interested in this space, but have been reluctant to rely on LLMs. How cheap are they, both financially and environmentally? Are you running these locally or using services?

DEV MEHROTRA (dmehrotra@wisc.edu)

2025-03-06 23:01:43

I am using foundation models like DinoV2 which are pretty cheap for interference so I can run them locally

👍 Ed Miller, Ben Weinstein

Sankalpa Ghose (sg@anart.ai)

2025-02-28 14:19:43

Hi everyone – just a friendly reminder that the AI FOR ANIMALS 2025 Conference is taking place this weekend at UC Berkeley! We have more than 60 speakers, 250+ in-person attendees, and 400+ virtual attendees.

And we are lucky to have @Sara Beery speaking on AI for Conservation at 10AM PST on March 1st!

Please do register to get free access to the virtual streams (which will also be available after on YouTube). Thank you!

aiforanimals.org

AI for Animals

Advancing the development of AI to create a better future for all Animals.

Original URL: https://www.aiforanimals.org

🎉 Jon Van Oast, Wenxin Yang, Aamir Ahmad, Brittany Aguilar, Shir Bar, Elena Sierra, Mamoon

❤️ Owen Martin, Sara Beery, Elena Sierra, Abigail O'Neill, Alan Stenhouse

Owen Martin (owma6084@colorado.edu)

2025-02-28 14:38:48

I’ll be speaking as well, about a direct application of AI to insect conservation! Tune in on Sunday at 11am pst if you’re curious how we’re building datasets and using neural networks for identification of rare fireflies from their flash patterns!

🐞 Sara Beery, Wenxin Yang, Alan Stenhouse

⚡ Sankalpa Ghose

Dan Morris (agentmorris@gmail.com)

2025-02-28 18:27:21

We released a small benchmark dataset today that may be of interest to AI4C folks:

https://github.com/google-research/ecology-georeferencing

Lots of information in the conservation literature is represented in maps (e.g., the exact survey area associated with a population estimate), so if LLMs are going to be useful for discovery and review in this area, they're going to have to get good at understanding not only the content of maps, but the spatial context of maps. That is, given a map, what area on the actual Earth does this map represent? This is a small dataset in which we manually georeferenced >100 maps from ecology papers, so we could test current LLMs on this task. That assessment is outside the scope of this data release, but tl;dr: all LLMs we tried are "OK but not quite good enough to use" at this task. But we think with some clever prompting and pseudo-tool-use, this is tractable. So, prompt away, and let us know what works!

google-research/ecology-georeferencing

Stars

Language

Jupyter Notebook

❤️ Sara Beery, Timm Haucke, Ian McGregor, Nate Harada, Robin Sandfort, Mamoon, Owen Martin, Lily Xu, Alan Stenhouse, Johannes Dollinger

😎 Jon Van Oast, Timm Haucke, Nate Harada, Ted Schmitt, Shir Bar, Owen Martin, Valerie

🗺️ Burak Ekim, Carly Batist, Björn Lütjens

Matt Weldy (matthewjweldy@gmail.com)

2025-03-02 10:32:34

I don't mean to put a tin hat on here, but could advancements down this path potentially expose lat long coordinates of listed species detections?

Tyler Marino (tylermarino8@gmail.com)

2025-02-28 19:23:46

This is so great! So perhaps when writing an ecology paper and developing a map, it is also beneficial to include a shape file or some geo-referenced file somewhere in the associated content portfolio

Ștefan Istrate (stefan.istrate@gmail.com)

2025-03-03 11:36:19

We're extremely excited to announce (and open source) SpeciesNet today, our AI model for species recognition in camera trap images, capable of recognising more than 2000 animals species around the world. To our knowledge this is currently the largest publicly available model out there, and ever since 2019, increasingly more complex versions of SpeciesNet have been serving the conservation community through the Wildlife Insights platform (https://www.wildlifeinsights.org/). We hope that by offering this powerful tool freely, we will enable even more researchers, conservationists and individual practitioners to scale up their global efforts to protect our fragile Earth and its biodiversity. Try it from the comfort of your own computer, in your office or in the field, without needing to upload your images to the cloud, and let us know what you think. Feel free to raise new issues and contribute with code and ideas. We're looking forward to seeing where the community will take SpeciesNet.

GitHub repo: https://github.com/google/cameratrapai

Our 2024 paper describing the model: https://doi.org/10.1049/cvi2.12318

Google's official announcement: https://blog.google/outreach-initiatives/entrepreneurs/ai-nature-climate-accelerator-nonprofits-speciesnet/

Even though ~90% of the open source code was written by me, this was the result of a solid team effort over the past few years and everyone deserves a great deal of recognition 👏 @Tomer Gadot, @Dan Morris, @Tanya Birch, Hyungwon (Clint) Kim, @Sara Beery, Jorge Ahumada.

Personally, this has been the most impactful project I've worked on in my career, and despite the uncertainty of the arrangement and the bureaucratic hurdles we had to fight every single year, the risks I took to be able to contribute to something meaningful to me paid off in the end. But these opportunities are extremely rare, and even more so for highly skilled engineers, in a time when we desperately need more of them in a whole range of disciplines if we're serious about solving any planetary crisis. As I'm looking for my next tribe, I'm rooting for more initiatives like this to reach the maturity to be released in the public space.

🎉 Timm Haucke, Oisin Mac Aodha, Lukas Picek, Ian McGregor, Dan Morris, Justin Kay, Ben Weinstein, Jon Van Oast, Sonny Burniston, Declan, Tanya Birch, Jonathan Sauder, Shir Bar, Filippo Varini, Cara Appel, Urs, Roshni Biswas, Elena Sierra, Taiki Sakai - NOAA Affiliate, Arthur Caillau, Subhransu Maji, Sara Beery, Matt Weldy, Arjun Subramonian (they/them), Carly Batist, Gaspard Dussert, Chaim, Don Cosseboom, Owen Martin, Tomer Gadot, Viktor Domazetoski, Mamoon, Seven Song, Piotr Tynecki, Angela Zhu, Xiaoxuan (Shirley) Ma, Nicolas Arrieta Larraza, Lily Xu, Fagner Cunha, Ed Miller, Abram Fleishman, Vijay Karthick, Antón Álvarez

❤️ Timm Haucke, Lukas Picek, Jon Van Oast, Sonny Burniston, Tanya Birch, Elena Sierra, Sara Beery, Arjun Subramonian (they/them), Don Cosseboom, Owen Martin, Davide Coppola, Santiago Martinez Balvanera, Valentin Gabeff, Angela Zhu, Lily Xu, Prabath Gunawardane, Alan Stenhouse, Jenna Kline, Antón Álvarez

🙌 Timm Haucke, Ragib Nihal, Judith Dekkers, Elizabeth Bondi-Kelly, Laura Chrobak, Shir Bar, Elena Sierra, Sara Beery, Arjun Subramonian (they/them), Don Cosseboom, Owen Martin, Aakash Gupta, Lily Xu, Björn Lütjens, Antón Álvarez

Sara Beery (sara.beery@gmail.com)

2025-03-03 11:54:59

I am so, so, so excited that the model is finally released!! This has been in the works for years, and represents a truly amazing effort by the whole team and especially Stefan!!!

❤️ Owen Martin, Valentin Gabeff

Sara Beery (sara.beery@gmail.com)

2025-03-03 11:56:05

If you want to see the model in action, come join the EFI-ESA Stats seminar starting now!!

Register here: https://bit.ly/4dbtALA

Zoom

Welcome! You are invited to join a webinar: EFI-ESA Stats Seminar: Ecological Modeling with AI and Python. After registering, you will receive a confirmation email about joining the webinar.

This Statistical Methods seminar by the Ecological Forecasting Initiative and the Ecological Society of America Statistical Ecology Section is hosting Sara Beery and Timm Haucke (Massachusetts Institute of Technology) to talk about Ecological Modeling with AI and Python.

Original URL: https://bit.ly/4dbtALA

Dan Morris (agentmorris@gmail.com)

2025-03-03 11:56:38

I'm super-biased, but FWIW, I've been playing with this OSS version of the model for a couple of weeks, and it's the best new toy I've ever gotten my hands on (even better than guitars and/or Voltron that would have been my favorite toys at previous phases of my life). Clearly the species classification problem isn't "solved" by any stretch, but IMO this model represents a step change in the number of users who can benefit from species classification without training a custom model.

🎉 Tanya Birch, Shir Bar, Carly Batist, Owen Martin, Timm Haucke, Valentin Gabeff, Jason Holmberg (Wild Me), Alan Stenhouse

Ben Weinstein (benweinstein2010@gmail.com)

2025-03-03 11:56:40

Thanks everyone!

Ben Weinstein (benweinstein2010@gmail.com)

2025-03-03 11:58:16

Is there tooling around fine tuning? What are the next step needs in getting this to community? Connections to Timelapse, eco assist, etc. how can we be of use?

Dan Morris (agentmorris@gmail.com)

2025-03-03 12:05:22

Nothing is really required of Timelapse; there is already a script in the repo to convert the output to Timelapse format; I've been piloting with a few external users who are already using the results in Timelapse, and that is going smoothly.

There is one new feature Saul will likely add to the Timelapse integration: currently in Timelapse you can filter by category name (in fact, filtering by category name is the entire goal of using AI), but since the category names have no concept of taxonomy, there's not a way to say "show me all the cats, regardless of whether the classifications were at family, genus, or species level". So I've added a new field to the MegaDetector/Timelapse .json format called "classificationcategorydescriptions", which are just per-category free text, but in practice are used by the SpeciesNet --> MD format converter to store taxonomic strings. And I talked to Saul about adding a Timelapse feature that allows you to filter for "description contains", which will allow you to say, e.g., "description contains felidae". No specific schedule on that, but I think it's a pretty good compromise that allows taxonomy-related functionality but does not require a complicated concept of taxonomy on the Timelapse side.

Of course, if someone is inspired to try to add that in the Timelapse repo, go for it!

Dan Morris (agentmorris@gmail.com)

2025-03-03 12:07:06

Peter and I met today to talk about AddaxAI (formerly EcoAssist) integration. Stay tuned!

Dan Morris (agentmorris@gmail.com)

2025-03-03 12:07:21

IMO fine-tuning would be a great area for community contributions.

Ben Weinstein (benweinstein2010@gmail.com)

2025-03-03 12:44:53

Thanks @Dan Morris, does feel like a place where connecting to LLM is simple, a bit overkill perhaps.

Ben Weinstein (benweinstein2010@gmail.com)

2025-03-03 12:48:43

@Ștefan Istrate how much should we read into the tensorflow + pytorch double dependency here. Should we be worried about maintaining this long term? I gather the underlying data isn't going to be released (understandable). I am planning on making a test drive video of this for my workshop students (https://weecology.github.io/AI_for_ecology_workshop/) and I'll post it here too.

Filippo Varini (fppvrn@gmail.com)

2025-03-03 13:19:08

Congrats @Ștefan Istrate that seems super helpful!!
One question, is there support for marine species ID or is SpeciesNet mainly designed for camera-trap/land-based data?

➕ Laura Chrobak, Shir Bar

Dan Morris (agentmorris@gmail.com)

2025-03-03 14:26:05

@Filippo Varini SpeciesNet is for camera trap data. When we're done with the new fish detector you are leading the charge on, we'll have to make SpeciesNet for fish too. :)

🐟 Sara Beery, Malte Pedersen, Owen Martin, Shir Bar

🙌 Filippo Varini, Malte Pedersen, Shir Bar

Ștefan Istrate (stefan.istrate@gmail.com)

2025-03-04 06:47:42

@Ben Weinstein Ideally we could (as a community) converge to one single framework in the near future (e.g. pytorch), but the double dependency (tensorflow + pytorch) shouldn't be an issue even if we don't converge. We have ways of running each component separately (detector, classifier, ensemble) and we can decouple python dependencies fairly easy if necessary. Dan documented in the README how create separate environments for classification and detection, which were necessary for his CUDA setup. What I'm mostly worried about is if future TensorFlow versions not being able to support models trained with older versions. If I remember correctly, this was the reason we can't support the latest TF version at the moment.

👍 Ben Weinstein

Ștefan Istrate (stefan.istrate@gmail.com)

2025-04-22 05:32:19

Quick update on SpeciesNet: As of today (Apr 22, 2025), SpeciesNet will not depend on TensorFlow any more. We have decided to serve both models (classifier + detector) via PyTorch from now on, and we hope this will make things much easier for anyone dealing with Python environments and GPUs. The conversion from TensorFlow to PyTorch went smoothly and the differences in predicted scores were minimal in our tests, but if you encounter any new issues feel free to raise your concerns on GitHub (https://github.com/google/cameratrapai).

🎉 Ben Weinstein, Sara Beery, Dan Morris

🤩 Sara Beery

Laura Chrobak (lchrobak@mbari.org)

2025-03-04 15:07:31

Navigating the Depths – New FathomNet Kaggle competition is live!

In the field of marine ecology, accurate taxonomic classification is essential for addressing fundamental questions: What species exist in a particular place? What is the ecosystem biodiversity and how does it change over time?

Questions like these motivate the focus of our 2025 FathomNet Kaggle Competition, aiming to push the boundaries of taxonomic accuracy and inspire innovative solutions.

Learn more: here

kaggle.com

FathomNet 2025 @ CVPR-FGVC

Navigating the Depths: Advancing Hierarchical Classification of Ocean Life

Original URL: https://www.kaggle.com/competitions/fathomnet-2025

🐟 Oisin Mac Aodha, Levi Cai, Chaim, Nico Lang, Angela Zhu, Jason Holmberg (Wild Me), Sara Beery, Malte Pedersen, Sonny Burniston, Alan Stenhouse, Genevieve Patterson, Elizabeth Campolongo

🙌 Nico Lang, Angela Zhu, Mamoon, Jason Holmberg (Wild Me), Sara Beery, Aditya Shirvalkar, Malte Pedersen, Sonny Burniston, Genevieve Patterson, Elizabeth Campolongo

👀 Lilli, Jason Holmberg (Wild Me), Devi Ayyagari, Filippo Varini

Kaitlyn Gaynor (kaitlyn.gaynor@ubc.ca)

2025-03-06 22:45:18

📣 Seeking postdoctoral researcher: Bridging AI and wildlife ecology 📣

The Gaynor Lab at the University of British Columbia :flag_ca: is recruiting a postdoctoral researcher to lead a self-directed project that bridges the fields of AI and wildlife ecology. This position is supported by the AI for Biodiversity Change (ABC) Global Center, which brings together ecologists and computer scientists from six universities in the United States and Canada to develop new AI-enabled, data-supported approaches for understanding the impacts of environmental changes on biodiversity (https://www.abcresearchcenter.org/).

abcresearchcenter.org

ABC Global Center

The NSF and NSERC-funded Artificial Intelligence and Biodiversity Change (ABC) Global Center brings together ecologists and computer scientists from six universities in the United States and Canada to develop new AI-enabled, data-supported approaches for understanding the impacts of environmental

Original URL: https://www.abcresearchcenter.org/

👀 Alan Stenhouse, Oisin Mac Aodha, Jennie Hatch, Lily Xu

👍 Ian McGregor, Valentin Gabeff

💚 Lily Xu, Elizabeth Campolongo

Kaitlyn Gaynor (kaitlyn.gaynor@ubc.ca)

2025-03-06 22:45:29

The postdoc will be expected to develop an independent research project that falls within this broad research theme and the foci of the Gaynor Lab (https://gaynorlab.weebly.com/), leveraging existing datasets of their choice. This project may include the development of algorithms for extracting ecological information from raw sensor data (i.e., camera traps, acoustic recorders), or any other cutting-edge applications of AI for understanding the effects of human disturbance on wildlife populations and communities. See our recent ABC review paper in Nature Reviews Biodiversity (Pollock et al. 2025) for the types of questions that would fall within the scope.

Research in the Gaynor Lab examines the effects of human disturbance on global biodiversity, with emphases on the behavioral responses of large mammals to human presence, the effects of anthropogenic disturbance on predator-prey and other species interactions, and the socio-ecological dynamics of conservation and coexistence. Our work involves large-scale data synthesis and meta-analyses, and field studies in Canada, the US, Mozambique, and Nepal. There will be opportunities for the postdoc to collaborate with other members of the lab on ongoing and new projects.

The position will be in person, based at the Biodiversity Research Centre at UBC’s Vancouver campus (https://biodiversity.ubc.ca/). In addition to joining the ABC community, the postdoc will also be integrated into other intellectual communities at UBC, including the Department of Zoology, Department of Botany, Data Science Institute, and Interdisciplinary Biodiversity Collaboratory. Salary is $70,000 CAD annually, plus standard UBC postdoctoral fellow benefits. The position is for one year, with renewal for a second year contingent on satisfactory progress. The preferred start date is September 1, 2025, but is flexible.

The ideal candidate will have a PhD in ecology or related discipline, and have a strong computational background, with prior experience applying AI tools for biodiversity monitoring. Interested candidates should e-mail kaitlyn.gaynor@ubc.ca with the subject line “ABC Postdoc Application” and an attached CV and 2-page cover letter that describes (1) relevant qualifications and fit to the position and (2) the research question(s) and approaches you would be interested in pursuing. Applications submitted by April 15, 2025 will be given full consideration.

gaynorlab.weebly.com

GAYNOR LAB

Understanding the role of humans in ecosystems is critical and urgent for biodiversity conservation, especially given the rapid growth of the global anthropogenic footprint. Our research examines the...

Original URL: https://gaynorlab.weebly.com/

biodiversity.ubc.ca

Welcome | Biodiversity Research Centre

Biodiversity research aims to document the enormous diversity of life on earth and to identify the factors that generate and maintain this diversity.

Original URL: https://biodiversity.ubc.ca/

Kaitlyn Gaynor (kaitlyn.gaynor@ubc.ca)

2025-03-06 22:45:44

Ping me if you have any questions, and please share widely with your networks! 🙂

James Brown (jamesbrown@lincoln.ac.uk)

2025-03-07 05:32:01

Hi everyone! I wanted to make you all aware of a workshop that will be taking place at ICIP 2025, which may be of interest to a number of you here 🙂 Our workshop on "Computer Vision for Ecological and Biodiversity Monitoring (CV-EBM)" is now accepting regular paper submissions - please visit our website to find out more: https://cvebm.blogs.lincoln.ac.uk/. If you have any questions please feel free to ask here! Thanks, James (CV-EBM Organiser)

👍 Oisin Mac Aodha, Sara Beery, Chris Lange, S M Rayeed, Josafat-Mattias Burmeister, Ghjulia Sialelli, Jason Holmberg (Wild Me), Malte Pedersen

Hugo Magaldi (magaldi.hugo@gmail.com)

2025-03-10 10:45:19

Hi all !

I've read this question somewhere on the slack recently but I can't find it anymore 🙈 So apologies for the duplicate.

What is the proper way to quote MegaDetector in a paper ? Same question for the curation of the LILA and Widlife Insights databases ? I know @Dan Morris is part of it, I just want to make sure I'm crediting all the people involved.

Thanks

Lukas Picek (lukaspicek@gmail.com)

2025-03-10 11:32:04

To my best knowledge, it should be this one. Right @Sara Beery?

@article{beery2019efficient,
  title={Efficient pipeline for camera trap image review},
  author={Beery, Sara and Morris, Dan and Yang, Siyu},
  journal={arXiv preprint arXiv:1907.06772},
  year={2019}
}

👍 Hugo Magaldi

Sara Beery (sara.beery@gmail.com)

2025-03-10 12:00:34

Yep!

👍 Hugo Magaldi

Dan Morris (agentmorris@gmail.com)

2025-03-10 12:03:06

Most LILA datasets have recommended citation on the dataset page; if you use one that doesn't, I recommend just citing the URL of the dataset.

👍 Hugo Magaldi

Hugo Magaldi (magaldi.hugo@gmail.com)

2025-03-10 12:09:46

Yes, thank you. My question was more: who should I acknowledge for maintaining the LILA/WI websites (in addition to citing the datasets) ?

Dan Morris (agentmorris@gmail.com)

2025-03-10 12:26:31

Only citing the datasets is necessary. I will defer your offer of good Karma for citing LILA itself until I need some other OSS favor.

😇 Hugo Magaldi

😆 Sara Beery

Sara Beery (sara.beery@gmail.com)

2025-03-10 13:47:39

New BirdCLEF challenge!!!

https://www.kaggle.com/competitions/birdclef-2025

kaggle.com

BirdCLEF+ 2025

Species identification from audio, focused on birds, amphibians, mammals and insects from the Middle Magdalena Valley of Colombia.

Original URL: https://www.kaggle.com/competitions/birdclef-2025

🦉 Taiki Sakai - NOAA Affiliate, Subhransu Maji, Sonny Burniston, Dan Morris, Carly Batist, Nicolas Arrieta Larraza, gyeongsu, Juan Sebastián Cañas, Owen Martin, Irina Tolkova, Léonard Boussioux, Jiangyue Wang, Alan Stenhouse

🎉 Jon Van Oast, Hannah Robertson, Matt Weldy, Mamoon, Shir Bar, Juan Sebastián Cañas, Arthur Caillau, Santiago Martinez Balvanera, Owen Martin, Filip Dorm, Maddie Cusimano, Laura Chrobak, Léonard Boussioux

🐦 Hannah Robertson, Brian Geuther, Ilyass Moummad, Mitchell Rogers, Ben Weinstein, Burooj Ghani, Juan Sebastián Cañas, Owen Martin, Léonard Boussioux

👍 Holger Klinck, Lukas Picek, Juan Sebastián Cañas, Owen Martin

❤️ murilo, Mitchell Rogers, Maria Jose Guerrero, gyeongsu, Juan Sebastián Cañas, Owen Martin, Maddie Cusimano, Alan Stenhouse

🐸 Juan Sebastián Cañas, Owen Martin, Maddie Cusimano, Yves Bas, Vijay Karthick

🦗 Juan Sebastián Cañas, Owen Martin, Brian Geuther, Maddie Cusimano

Holger Klinck (hk829@cornell.edu)

2025-03-10 14:01:31

You are fast, Sara :)

Sara Beery (sara.beery@gmail.com)

2025-03-10 15:40:50

I'm excited!!

🤩 Juan Sebastián Cañas

Holger Klinck (hk829@cornell.edu)

2025-03-10 16:01:13

Makes two of us. The first multi-taxa one 🙂

Lukas Picek (lukaspicek@gmail.com)

2025-03-10 14:56:59

Yes, the BirdCLEF challenge is live! 🎉
But that’s not all! There are also some other exciting challenges coming up:

AnimalCLEF 🦁: Multi-species animal identification

PlantCLEF 🌿: Plant identification in vegetation quadrat images

FathomNet 🌊: Hierarchical classification of ocean life

FungiCLEF 🍄: Few-shot classification with rare fungi species

GeoLifeCLEF 🌍: Location-based species presence prediction

For more details, check out the FGVC Workshop website!

👍 Holger Klinck, Sonny Burniston, Sara Beery, Jan Dirk Wegner, Shir Bar, gyeongsu, Namitha Suresh, FF, Hugo Magaldi, Robin Zbinden, Kostas Papafitsoros, Thijs van der Plas, Owen Martin, Stephanie Mitchell, Laura Chrobak, Léonard Boussioux

❤️ murilo, Sonny Burniston, Subhransu Maji, Sara Si-Moussi, Ilyass Moummad, Sara Beery, Mitchell Rogers, Shir Bar, Jess Tam, Roberta Hunt, Thijs van der Plas, Owen Martin, Prabath Gunawardane, Filip Dorm, Léonard Boussioux, Navodita Mathur, Kakani Katija, Alan Stenhouse

😮 Jon Van Oast, Léonard Boussioux

🐢 Mitchell Rogers, Robin Zbinden

🐦 Oisin Mac Aodha

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2025-03-10 16:13:02

can one still register and participate? it says it started 28th feb.

Lukas Picek (lukaspicek@gmail.com)

2025-03-10 16:22:39

All just started. Anyone can “register” and participate 🙂

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2025-03-10 16:23:08

👍 thnks

Jeremy Fix (jeremy.fix@centralesupelec.fr)

2025-03-11 06:18:20

The GeoLifeClef link seems down ? It points to https://www.kaggle.com/competitions/geolifeclef-2025 but the competition does not seem to exist

Lukas Picek (lukaspicek@gmail.com)

2025-03-11 06:58:49

@Jeremy Fix It was supposed to be running yesterday. Unfortunately, we have encountered some “interesting” bugs on Kaggle which must be resolved before launch. I will ping you once it is up and running.

👍 Jeremy Fix, Navodita Mathur

Owen Martin (owma6084@colorado.edu)

2025-03-11 19:42:19

Id love to pair up with some other practitioners here for birdclef! Started some eda today and it seems like a rad dataset

🤞 Lukas Picek, Navodita Mathur

Sam Lapp (sam.lapp@pitt.edu)

2025-03-11 15:07:43

Has anyone tried training / fine-tuning a LLM chatbot to answer questions on a specific code base / repository / package?

👀 Ben Williams, ash

Kalindi Fonda (kalindi.fonda@gmail.com)

2025-03-11 15:18:34

I haven't tried it myself, but isn't this what Cursor does? And I know a few people who communicate with various LLMs about their code from within the terminal/ide, I can ask them more about it, if you have any specific questions or connect you to them directly.

Matt Weldy (matthewjweldy@gmail.com)

2025-03-11 15:41:38

Not a complete answer to your question... But, there are a few tools out there for pulling a codebase into a single markdown document, which can be fed into LLMs as a reference document. In a couple of situations I found it useful. But in fair warning I've had it fail miserably a few times.

Sam Lapp (sam.lapp@pitt.edu)

2025-03-11 15:43:30

yes in my experience, both providing the code as a single document as “context” and using VS Code with the Copilot chat (which is supposed to use your whole active project as context), have given unsatisfactory results

Matt Weldy (matthewjweldy@gmail.com)

2025-03-11 15:44:53

I had Aider recommended to me as well. But I haven't tried it yet. https://aider.chat/

aider

/llms.txt—a proposal to provide information to help LLMs use websites – Answer.AI

aider is AI pair programming in your terminal

Original URL: https://aider.chat/

Ben Weinstein (benweinstein2010@gmail.com)

2025-03-11 17:14:48

I'm a huge fan of cursor

Sam Lapp (sam.lapp@pitt.edu)

2025-03-11 17:19:06

I’ll check it out. My hope is to provide a chatbot that provides correct advice on using OpenSoundscape (python library), but so far chatGPT with context/hints is pretty bad at this. Given extensive package documentation and tutorials, I feel like an LLM should have an easy time providing code solutions that use the package correctly (at least for things where there’s already an example in the tutorials!)

💪 Kalindi Fonda, Sara Beery

👍 Ben Weinstein

Mike Gallimore (mgallimore88@gmail.com)

2025-03-11 18:31:06

Hi Sam, Yes I’ve got some experience in this area - want to arrange a call sometime soon?

Sam Lapp (sam.lapp@pitt.edu)

2025-03-11 18:41:20

That sounds great @Mike Gallimore - Sam.lapp@pitt.edu

Dan Morris (agentmorris@gmail.com)

2025-03-11 18:56:42

IMO you don't want to fine-tune a model, as others have suggested, you just want to make it easy to get the entire code base into the model as context.

So if you like VS Code, +1 for suggestions to try Cursor (which is VS-Code-based). IMO the magic of Cursor is this operator:

https://docs.cursor.com/context/@-symbols/@-folders

...which says "hey, whatever LLM I'm talking to through Cursor, I'm about to ask you a question, and here's a whole folder of code I want you to read before I ask you that question".

But if you're not into VS Code, or you don't really feel like installing anything, there are some great OSS tools whose entire raison d'etre is to take your GitHub repo and smush it into a text file that you can upload to a chat interface before you ask a question. I.e., "hey, LLM, I'm attaching a GitHub repo as a text file, and I'm going to ask you questions about it". This one is the most popular:

https://github.com/yamadashy/repomix

It's pretty fantastic, you can even skip running anything locally, and just go to their Web page, put in your repo URL, and get a giant text or XML file:

https://repomix.com/

(The Web page seems to be down right now, but you'll get the idea, and you don't need the Web page, the command-line version does the same thing.)

The bottom of that Web page has nice examples of the kinds of prompts you might use after uploading your packed repo file.

Also once you go through the trouble of formulating a query and packing up your repo, definitely try all the chatbots you can try in different tabs, they all have strengths and weaknesses.

Let us all know how your experiments go!

👍 Sam Lapp, Sara Beery, Mike Gallimore, Justin Kay

👍:skin_tone_3: Alan Stenhouse

Mike Gallimore (mgallimore88@gmail.com)

2025-03-12 18:53:05

Hi,

It’s possible that you’ll run into issues with the context length of any LLM you choose to use.

Option 1 - Custom GPT
I think the quickest way to do this would be to make a custom GPT on chat GPT. You can upload up to 20 documents, each containing 2,000,000 tokens max. I’m not sure how they got the token length so high so I’m skeptical until I try it myself, but If this worked, it would be simplest because all the infrastructure is already set up. Anybody could use the custom GPT with a subscription.

llms.txt
For future models to be able to parse your codebase more efficiently from future web scrapes, you could consider making an LLMs.txt file. It’s a markdown file containing all the docs, called llms.txt or llms.md, which sits in the root of your repo.

Here is

an example llms.txt from an online repository,

the original documentation with all the extra formatting, images etc which make things easier for a human to read, but which an llm wouldn’t need, and hence would increase the context length and be harder to parse,

and a page outlining the rationale behind providing an llms.txt file.

Basically you make a dense representation of the knowledge in the documentation to make it smaller and easier for either a model to parse.

Option 2 - Cursor
The Cursor idea from Dan Morris is worth a try. I have not tried this myself, but it looks promising. I am unsure how this would scale to be used as a chatbot.

RepoMix is an awesome tool too! This would be a quick way to split the repo for GPT.

Option 3 - RAG
If the context length of the models is too short, or if the performance is still poor using a custom GPT then you could start looking into Retreival Augmented Generation. It’s just another way of providing relevant context to an LLM. The technique works like this:

split the documentation and code repo into segments,

create embeddings (numeric representation) for each segment. These can be stored in a database.

create an embedding for the question you want to ask

calculate the cosine similarity/dot product between the query embedding and each segment embedding, and pick the top, say, 5.

Feed the text versions of these top segments into a model along with the prompt to get an answer which has the right context.

It is a little fiddly deciding how to split the repo (by file, function name, paragraph?) and which model to use to create the embeddings: they also have their own context lengths, and will have been developed for varying tasks.

Hopefully you’d build a system which could retrieve just the most relevant sections / pages from the documentation and give them to a model with the prompt.

You’d need to think about deployment in this case too if it was for scale.

Here’s some stuff on RAG if you decided to go down that path:
https://parlance-labs.com/education/rag/ben.html

Option 4 - fine tuning
It is possible to fine tune large language model to learn new information and vocabulary. You can do this on your laptop now too - a technique called Parameter Efficient Finetuning (PEFT) and Low Rank Adaptation (LoRA) trains a set of new parameters which sit in between the model layers, so you don’t need to train the entire model- just these new weights. It’s a fast way to fine-tune and it also helps stop catastrophic forgetting. Check out the unsloth documentation for info.

It could be worth trying- I’ve played with it a bit but I had inconclusive results.

Hope you get it figured out!

I really enjoyed working with the OpenSoundScape library and thought the documentation was some of the best I’ve come across.

claudette.answer.ai

claudette

Claudette is Claude’s friend

Original URL: https://claudette.answer.ai/

Answer.AI

We propose that those interested in providing LLM-friendly content add a /llms.txt file to their site. This is a markdown file that provides brief background information and guidance, along with links to markdown files providing more detailed information.

Original URL: https://www.answer.ai/posts/2024-09-03-llmstxt.html

Parlance

Beyond the Basics of RAG – Parlance

LLMs are powerful, but have limitations: their knowledge is fixed in their weights, and their context window is limited. Worse: when they don’t know something, they might just make it up. RAG, for Retrieval Augmented Generation, has emerged as a way to mitigate both of those problems. However, implementing RAG effectively is more complex than it seems. The nitty gritty parts of what makes good retrieval good are rarely talked about: No, cosine similarity is, in fact, not all you need. In this workshop, we explore what helps build a robust RAG pipeline, and how simple insights from retrieval research can greatly improve your RAG efforts. We’ll cover key topics like BM25, re-ranking, indexing, domain specificity, evaluation beyond LGTM@few, and filtering. Be prepared for a whole new crowd of incredibly useful buzzwords to enter your vocabulary.

Original URL: https://parlance-labs.com/education/rag/ben.html

docs.unsloth.ai

Welcome | Unsloth Documentation

New to Unsloth?

Original URL: https://docs.unsloth.ai/

👍 Sam Lapp

Sam Lapp (sam.lapp@pitt.edu)

2025-03-15 16:59:40

Wow thank you all especially @Dan Morris and @Mike Gallimore for detailed thoughts! I didn’t know about llms.txt, that’s very good to know. I’m thinking I should make a little benchmark task set to compare these approaches. I will report back with results or questions.

Rachel Mason (rachel.e.mason1@gmail.com)

2025-03-12 21:03:19

Is anyone here affiliated with an organization using Planet's nonprofit program? What type of work are you using it for?

Ghjulia Sialelli (ghjulia.sialelli@inf.ethz.ch)

2025-03-13 04:53:59

@David ?

Lily Xu (lily.x@columbia.edu)

2025-03-14 19:37:08

NSF Postdoc position as a Forest Carbon Scientist, working with Earthshot Labs

Earthshot is a climate tech company focused on scaling carbon development efforts for reforestation and conservation, developing and financing nature-based carbon projects, and advancing automation in technology and science while ensuring regulatory compliance. Our goal is to make the easiest and most effective way to create and finance nature restoration and conservation carbon projects, and we need your help to do it.

Requirements include: PhD in the field of Ecology, Forestry, Remote Sensing, Modeling, Environmental Science or related field

Through NSF’s Innovative Postdoctoral Entrepreneurial Research Fellowship (IPERF) program

Details: https://docs.google.com/document/u/2/d/e/2PACX-1vScL1zRzKbr_3kEU1FcGLioraU4uNzgqaojA6w17SrmQHgxvL63Tr-66vKAO1w_yHACkQFwjrEnRUwK/pub

❤️ Owen Martin, Vincent Lostanlen, Eric Greenlee

Ben Williams (ben.williams.20@ucl.ac.uk)

2025-03-16 11:03:51

Hi all, I had the chance to give a talk at TEDx London in Jan on some of the work from my PhD using AI to help protect coral reefs 🪸 This was very much made for a general public audience so plenty of details I don’t go into, but I hope it might be of interest to some in this community! If anyone wants to find out more than please do reach out 🤗 I’m also finishing my PhD this year, so if you know of any opportunities to work on similar challenges then hmu! https://youtu.be/GnB8OoT2UAE?si=BlZjxBKPQv0AsOOJ

YouTube

TEDx Talks (https://www.youtube.com/@TEDx)

How AI is helping transform coral reef conservation | Ben Williams | TEDxLondon

Original URL: https://youtu.be/GnB8OoT2UAE?si=BlZjxBKPQv0AsOOJ

👏 Matt Weldy, Jeegn Dani, Mamoon, Kalindi Fonda, Oisin Mac Aodha, murilo, Rupa Kurinchi-Vendhan, Dan Morris, Gedeon, Timm Haucke, Alessandra Sellini, Jonathan Sauder, Céline Angonin, Ian McGregor, Juan Sebastián Cañas, Fagner Cunha, Chris Lange, Michael Procko, Björn Lütjens, Jarrett Blair, Nathaniel Rindlaub

❤️ Owen Martin, Nora Gourmelon, Juan Sebastián Cañas, Kishore Panaganti, Chris Llorca

👏:skin_tone_2: Cara Appel

👏:skin_tone_4: Chris Llorca

🪸 Chris Llorca, Alan Stenhouse

Ben Weinstein (benweinstein2010@gmail.com)

2025-03-17 17:29:04

I think one place that LLMs can start to play a role is in resurrecting historically annotated data. There are huge datasets from prior research efforts pre2020 that were not annotated with any goal of machine learning development. I am quite interested in unlocking these pieces if anyone wants to join me. Consider this image in which a non-profit directly 'dotted' the image, literally overtop the image pixel, obscuring the target classes. We have the original image, the counts, and a 'screenshot' of the final dotted image. I suspect that an object detector (like our bird detector) plus a LLM may be able to recover the positions of these annotations and convert them into COCO like annotations. I am going to play with this along with a google summer of code student. If other teams have similar kind of tasks in the flavor of 'old annotated imagery that is not suitable for machine learning development, but could be with an intern and 100 hours', message me. I think this would make a fun gemini benchmark (@Dan Morris)

😲 Mitchell Rogers, Shir Bar, Aleksis Pirinen, Alexander Robillard

👍 Sara Beery, Ankita Shukla, Bernie Boscoe, Alexander Robillard

🙌 Mamoon, Tiziana Gelmi Candusso, Alexander Robillard

👍:skin_tone_3: Jess Tam, Alan Stenhouse

Sara Beery (sara.beery@gmail.com)

2025-03-17 17:46:27

@Eddie Vendrow and I have been talking about this

👍 Ben Weinstein

Kate Rakelly (krakelly@gmail.com)

2025-03-17 18:34:47

Recognizing that the task could be much harder than appears from these screenshots….What about a simple color filter > cluster > report the mean pixel position of each cluster. If the shapes are always the same colors and size, seems like this could work well?

👍 David Russell

Ben Weinstein (benweinstein2010@gmail.com)

2025-03-17 20:18:22

Yes and no, we don’t have the dotted image itself, just a screenshot, so regardless of how you get the positions of markers there is still the logic portion of the number of categories and their estimated position relative to the image frame.

Kate Rakelly (krakelly@gmail.com)

2025-03-17 20:46:07

Oh I see…and the screenshots were taken by hand, so the original image isn’t at the same coordinates in all of them?

Kate Rakelly (krakelly@gmail.com)

2025-03-17 20:53:58

I can see how the point is probably a general-purpose tool to rehabilitate all sorts of old data. Just a bit skeptical of the capability and the compute tradeoff :)

Ben Weinstein (benweinstein2010@gmail.com)

2025-03-17 21:33:13

oh for sure. definitely questionable. Worth exploring. Just getting a feeling for what kinds of data are out there. My sense is that there are many esoteric use cases, it is only worthwhile if there is some unifying thread.

👍 Kate Rakelly

Sara Beery (sara.beery@gmail.com)

2025-03-18 09:05:01

Hey all! Help me put together a list of AI training programs for ecologists? Unless @Dan Morris already has this list somewhere 😂

https://docs.google.com/document/d/1fMNLFMKSsrBKSWcarBQAQh11cMfL1nn5j7clas0szZo/edit?usp=sharing

😂 Malte Pedersen

😎 Jon Van Oast, Carly Batist, Shir Bar

❤️ Kalindi Fonda, Reaz Masud, Nanticha Ocharoenchai (Lyn), Jennifer, Alan Stenhouse

💪 Kalindi Fonda

👍 Reaz Masud

Sara Beery (sara.beery@gmail.com)

2025-03-18 09:05:26

https://cv4ecology.caltech.edu/

Sara Beery (sara.beery@gmail.com)

2025-03-18 09:06:23

https://imageomics.osu.edu/news/2024/06/imageomics-institute-launches-new-ai-ecology-course

Imageomics Institute

Imageomics Institute Launches New AI-Ecology Course

Original URL: https://imageomics.osu.edu/news/2024/06/imageomics-institute-launches-new-ai-ecology-course

Sara Beery (sara.beery@gmail.com)

2025-03-18 09:06:58

@Ben Weinstein does your new course in spanish have a website?

Sara Beery (sara.beery@gmail.com)

2025-03-18 09:08:04

There are also masters programs, though these tend to be "data science" as opposed to AI:

https://www.ucl.ac.uk/biosciences/msc-ecology-and-data-science

https://bren.ucsb.edu/masters-programs/master-environmental-data-science

UCL Division of Biosciences

MSc Ecology and Data Science

Prepare to address critical global environmental challenges with a sought-after skillset spanning data science and ecology, on this one-year multidisciplinary MSc.

Original URL: https://www.ucl.ac.uk/biosciences/msc-ecology-and-data-science

UC Santa Barbara Bren - Bren School of Environment

Master of Environmental Data Science (MEDS) | UC Santa Barbara Bren - Bren School of Environment

The Master of Environmental Data Science degree program (MEDS) at the Bren School is a 1 year professional graduate degree program that connects data science and environmental science to advance solutions to global problems.

Original URL: https://bren.ucsb.edu/masters-programs/master-environmental-data-science

Ben Weinstein (benweinstein2010@gmail.com)

2025-03-18 10:25:43

https://weecology.github.io/AI_for_ecology_workshop/index.html ya my mini lectures will be in english.

Jamie Alison (jalison@ecos.au.dk)

2025-03-18 10:54:49

https://darsa.info/ACCESS-2025/ 🙂

darsa.info

InsectAI ACCESS-2025

Aarhus Comprehensive Computational Entomology Summer School, 2025

Original URL: https://darsa.info/ACCESS-2025/

👏 Malte Pedersen

Sara Beery (sara.beery@gmail.com)

2025-03-18 11:10:35

Feel free to add them directly!

👍 Reaz Masud

Reaz Masud (masud@altri.ai)

2025-03-19 23:26:18

Thank you for sharing this list, cool programs!

In the doc, I can see some resources were added regarding AI training, not directly related to ecology - very interesting!

We have an ocean of information now, and AI made access to the relevant content easier than ever.

Using different conversation tools like ChatGPT - I can ask a question, and AI can guide me to find relevant resources (e.g. youtube videos, research papers, websites, etc.).

If knowledge can be acquired solely through self-interest, to drive impact - collaboration with the right support group (community) is essential from my view. So that knowledge and skills can be shared and complemented.

Added a section to the doc.

👍 Sara Beery

Vincent Lostanlen (vincent.lostanlen@gmail.com)

2025-03-18 10:48:22

Hello everyone,
Is anyone going to the IEEE ICASSP conference in Hyderabad, India?
There is a special session on bioacoustics and program chair Geoffroy Peeters is looking for a chairperson. I was planning to attend in person myself but i can't go anymore due to visa complications.

Dan Stowell (dan.stowell@naturalis.nl)

2025-03-21 11:58:27

From our group, ask @Inês Nolasco?

🆗 Vincent Lostanlen

Björn Lütjens (bjoern.luetjens@gmail.com)

2025-03-19 10:08:02

Hi all! We just released the first lecture from this year's workshop on Computer Vision for Ecology at Caltech with @Sara Beery giving an Intro to Computer Vision - Deep Learning, CNNs, and Transformers. We're aiming to release the other lectures on a biweekly schedule. If you have any comments please let me know. Thanks! https://youtu.be/5rE2ogBh0zc?si=s06eRF6bMms3-yof

YouTube

CV4Ecology (https://www.youtube.com/@cv4ecology)

Caltech CV4E - Intro to Computer Vision - Deep Learning, CNNs, Transformers - Prof. Sara Beery

Original URL: https://youtu.be/5rE2ogBh0zc?si=s06eRF6bMms3-yof

😎 Brian Geuther, Jon Van Oast, Carly Batist

🙌 Kenza, Michael Procko, Ian McGregor, Aditya Shirvalkar, Mamoon, Dan Morris, Justin Kay, Elena Sierra, Hammed Akande, Shir Bar, Carly Batist, Mike Gallimore, Malte Pedersen, Reaz Masud, Yseult Hb, Gaspard Dussert, Rupa Kurinchi-Vendhan, Laura Chrobak, S M Rayeed, Ilyass Moummad, Ghjulia Sialelli, Aleksis Pirinen, Alexander Robillard, Carla Peterson

🍿 Burak Ekim, Christine Muthee, Ben Weinstein, Sara Beery, Elena Sierra, Lukas Picek, Yseult Hb, Alexander Robillard

🎉 Jon Van Oast, Riccardo de Lutio, stefano puliti, Lily Xu, Alexander Robillard

❤️ Kishore Panaganti, Laura Madrid, takatomi-k, Reaz Masud, Jess Tam, Vijay Karthick, Wenxin Yang, Ekaterina Nepovinnykh, Maria Napolitani, Alexander Robillard, Alan Stenhouse, Nathaniel Rindlaub

💙 Patrick Beukema, Alexander Robillard

👍 Lukas Picek, takatomi-k, Reaz Masud, Akshit Gupta, Alexander Robillard

🙌:skin_tone_3: Hannah Yin

Akshit Gupta (akshitgupta1695@gmail.com)

2025-03-24 11:53:11

The lecture is great to watch as it allowed basic-principles thinking and validation.

❤️ Sara Beery

Björn Lütjens (bjoern.luetjens@gmail.com)

2025-03-24 12:49:23

great to hear :)

Dan Stowell (dan.stowell@naturalis.nl)

2025-03-21 12:02:26

Job alert! Postdoctoral Fellow in AI for Ultrasonic Bioacoustic Monitoring 🦇 😎

All the details: https://www.naturalis.nl/en/about-us/job-opportunities/postdoctoral-fellow-in-ai-for-ultrasonic-bioacoustic-monitoring

Plus here it is for Bluesky and Mastodon (with pictures 🙂 re-boings appreciated)

📣 Oisin Mac Aodha, Justin Kay, Owen Martin, Ben Weinstein, Juan Sebastián Cañas

❤️ Sara Si-Moussi, Owen Martin, Juan Sebastián Cañas, Sara Beery

Lily Xu (lily.x@columbia.edu)

2025-03-24 04:03:05

Fully funded PhD and MS in the Environmental Informatics Hub, led by Prof. Iadine Chades, at Monash University (in Australia)!

We are offering fully-funded scholarships to support outstanding candidates interested in developing AI and computer science solutions to tackle critical environmental challenges, including a strong focus on biodiversity conservation and the nature-positive agenda.
Projects will involve close collaboration with the Faculty of IT and may include external supervisors and partners (e.g. NGOs, government, industry). We particularly welcome applicants from backgrounds in computer science, applied mathematics, environmental science, engineering, or related disciplines.

Deadline April 15; details attached

📢 Juan Sebastián Cañas

💚 Millie Chapman, Mamoon, Reaz Masud, Juan Sebastián Cañas

‼️ Tyler Marino, Juan Sebastián Cañas

Burooj Ghani (buroojghani@gmail.com)

2025-03-24 09:24:44

🚨 New dataset drop! 🦗🎶

We just released InsectSet459 - the first large-scale open dataset of insect sounds, featuring 26,399 audio clips from 459 species of Orthoptera and Cicadidae.
It's designed for deep learning in insect sound recognition and benchmarks show promise - but there's plenty of room for innovation 🧠⚡️
Check out the dataset and our arXiv preprint to dive into the details!

🎉 Roberta Hunt, Sara Beery, Aakash Gupta, Céline Angonin, Ben Weinstein, Santiago Martinez Balvanera, Dan Morris, Viktor Domazetoski, Carly Batist, Arthur Caillau, Sam Lapp, Morgan Langley, Taiki Sakai - NOAA Affiliate, Elizabeth Campolongo, Mélisande Teng, Owen Martin, Reaz Masud, Juan Sebastián Cañas, Vincent Lostanlen, Aleksis Pirinen, Antón Álvarez

❤️ Owen Martin, Reaz Masud, Juan Sebastián Cañas, Irina Tolkova, Vincent Lostanlen

Carly Batist (carly.h.batist@gmail.com)

2025-03-24 10:55:58

AWESOME! Love that you are working on such under-researched taxa (in PAM but also generally) and bringing them to the forefront!

❤️ Burooj Ghani, Dan Stowell, Antón Álvarez

Adrien Fontvielle (adrien@enlaps.fr)

2025-03-24 10:18:45

Hi all! I know this is not exactely a place to find glaciologist, but I thought it might have an interest for some of you:

PLANET WATCH BY ENLAPS – One technology. So many changes. One purpose: Protect our planet.

Today, we are launching Planet Watch by Enlaps!

As glaciers face unprecedented threats due to climate change, we are taking action.

We don’t just capture images—we document climate history.

With Tikee, our 100% autonomous, solar-powered, and connected camera, combined with the myTikee cloud platform, scientists can monitor glaciers with unmatched precision.

Our commitment:
- 1% of Tikee camera sales dedicated to environmental projects
- 40 scientific projects will be selected to track glacier evolution
- A panel of renowned experts to select impactful projects

Our expert jury:
- Dr. Dr. Heïdi Sevestre – Glaciologist, speaker, and author, and Tikee expert
- Prof. Dr. Michael Zemp – Director of the World Glacier Monitoring Service (WGMS)
- And myself to check the technical feasibility

📢 Call for projects open until June 6, 2025!
Join us and submit your project! 👉 https://enlaps.io/us/planet-watch

enlaps.io

Planet Watch by Enlaps: Glacier Monitoring

Planet Watch by Enlaps is an invitation for project submissions from glaciologists across the United States to highlight the impact of global warming. Indeed, in the domain of scientific research, timelapse photography is extensively employed to analyze glaciers.

Original URL: https://enlaps.io/us/planet-watch

🧊 Benjamin Hoffman

Dan Morris (agentmorris@gmail.com)

2025-03-24 15:43:02

New dataset on LILA:

https://lila.science/datasets/unsw-predators/

This dataset contains ~131k images of quolls, foxes, dingoes, goannas, and possums from New South Wales. This increases the amount of Australian data on LILA by exactly infinity percent, and the ~84k images of goannas increase the representation of reptiles in camera trap data on LILA by almost infinity percent. Big thanks to Brendan Alting and Neil Jordan from UNSW, who provided the data, and to @Peter van Lunteren who nudged us to get it out the door.

🎉 Timm Haucke, Jon Van Oast, Jason Holmberg (Wild Me), Elizabeth Campolongo, Shir Bar, Mitchell Rogers, Reaz Masud, Toshio Matsuoka, S M Rayeed, Jenna Kline, Carly Batist, Burooj Ghani, Ethan White, Owen Martin, Neha Hulkund, Jess Tam, Namitha Suresh

🙌 Peter van Lunteren, Dante Wasmuht, Jason Holmberg (Wild Me), Shir Bar

😎 Jon Van Oast, Jason Holmberg (Wild Me)

❤️ Reaz Masud, Tiziana Gelmi Candusso, Kalindi Fonda

🦘 Alan Stenhouse, Alexander Merdian-Tarko

Sara Beery (sara.beery@gmail.com)

2025-03-25 14:14:55

Hey everyone!

The AI for wildlife conservation community has been impacted by recent layoffs in the US public sector; many of the conservation scientists who either developed or used AI tools for wildlife monitoring have lost their jobs, and have reached out to AI practitioners for advice on building AI skills before re-entering the workforce.

To assist in that upskilling, and to foster a community in which those impacted can help each other build AI skills, the CV4Ecology program is organizing a two-week remote workshop in May of 2025 that will teach computer vision and machine learning methods to former US public sector conservation scientists, based loosely on the CV4Ecology workshop.

Please share broadly to your networks so this can reach its intended audience! And we are also looking for volunteers to help instruct/mentor. Details here: https://cv4ecology.caltech.edu/ai-upskilling-workshop-2025.html

👏 Oisin Mac Aodha, Ștefan Istrate, Jeegn Dani, Felipe Parodi, Timm Haucke, Dan Morris, Justin Kay, Mark Goldwater, Elie Alhajjar, Carly Batist, Kate Rakelly, Esther Rolf, Ted Schmitt, Luke Meyers, Gracie Ermi, Ian McGregor, Aditya Shirvalkar, Chris Cosma, Lauren Shrack, Rupa Kurinchi-Vendhan, Jose Ruiz-Munoz, Boyu Zhang, Mitchell Rogers, Eric Colson, Elena Sierra, Irina Tolkova, Mamoon, Shir Bar, Andy Viet Huynh, Matt Weldy, Alessandra Vidal Meza, Maria Jose Guerrero, Sam Lapp, Nina van Tiel, Yseult Hb, Robin Zbinden, Riccardo de Lutio, Vincent Lostanlen, Piotr Tynecki, Matthias Zuerl, murilo, Wenxin Yang, Marion Richardot, Yujie Guo, Toshio Matsuoka, Jennifer, David Russell, Ghjulia Sialelli, Roberta Hunt, Amritha Pallavoor

👏:skin_tone_2: Cara Appel

😍 Jon Van Oast, Mark Goldwater, Taiki Sakai - NOAA Affiliate, Chris Cosma, Elena Sierra, Catherine, Andy Viet Huynh, Burak Ekim, Yseult Hb, Marion Richardot, Aleksis Pirinen, Reaz Masud

✅ Patrick Beukema, Andy Viet Huynh, Marion Richardot

👏:skin_tone_5: Prabath Gunawardane

Stephen Formel (sformel@usgs.gov)

2025-03-26 11:51:39

@Sara Beery the form asks us to confirm that we have lost our govt job. I suspect I won’t have a job by May. Any chance the workshop can make room for people like me?

David Russell (davidrussell327@gmail.com)

2025-03-27 14:24:19

This is great! How much is this focused on wildlife vs. ecology more broadly? I know a couple forest ecologists who lost jobs who might be interested.

Sara Beery (sara.beery@gmail.com)

2025-03-27 14:38:23

David, ecology more broadly :)

🙌 David Russell

👍 Reaz Masud

Sara Beery (sara.beery@gmail.com)

2025-03-27 14:39:18

@Stephen Formel yes we are accepting this type of participation. You're not the only one in this position 💔

👍 Stephen Formel

Reaz Masud (masud@altri.ai)

2025-04-01 10:59:13

Wonderful initiative, thanks for sharing the details.

❤️ Sara Beery

Carly Batist (carly.h.batist@gmail.com)

2025-03-25 15:35:35

Does anyone have podcasts they'd recommend about AI? Not necessarily specific to conservation but outlets that discuss innovative AI use cases. Appreciate any insights, thanks!

Dan Morris (agentmorris@gmail.com)

2025-03-25 15:58:57

I don't know about AI for Good stuff, but in terms of keeping up with industry trends in LLMs/GenAI, my favorite is https://www.aiforhumans.show/. It's legit fun, and it's enough to make me feel like I know more or less what's happening without actually paying that much attention to the new models coming out every week. There's a vaguely-pirate-themed song in every episode that I can't get enough of.

The Data Skeptic podcast (another general AI podcast) did a season called "Animal Intelligence" that still wasn't about AI for good per se, but was definitely an AI podcast talking about animals, so that's pretty close: https://dataskeptic.com/podcast/animal-intelligence.

This podcast has sort of an AI for good vibe: https://podcasts.apple.com/us/podcast/in-the-blink-of-a-i/id1776534118, and at the very least, they had one episode that's squarely in our universe: https://podcasts.apple.com/us/podcast/leveraging-ai-for-biodiversity-conservation-with-jada/id1776534118?i=1000676116153.

👍 Carly Batist, Elizabeth Campolongo

🙏 Andy Viet Huynh

👍:skin_tone_4: Oc

Carly Batist (carly.h.batist@gmail.com)

2025-03-25 17:23:20

Awesome, thanks Dan!

murilo (murilogustineli@gatech.edu)

2025-03-26 09:04:12

I really like Machine Learning Street Talk https://www.youtube.com/@MachineLearningStreetTalk
it's also available on Spotify 🙂

YouTube

Machine Learning Street Talk

MLST is the top technical AI podcast on Spotify. Subscribe now! Welcome! We bring you the latest in advanced AI research, from the best AI experts in the world. Our approach is unrivalled in terms of scope and rigour – we believe in diversity of ideas (which is to say, not just LLMs!) and we also cover other promising alternative paths to AGI, as well as CogSci, CompSci, Neuro, Mathematics, Philosophy of Mind and Language. Support us on Patreon for early access, exclusive content, private Discord, biweekly calls and much more! <https://www.patreon.com/mlst> Donate here: <https://www.paypal.com/donate/?hosted_button_id=K2TYRVPBGXVNA> Please email us to learn about sponsorship packages and deals. tim at <a href="http://mlst.ai">mlst.ai</a> (please put your budget in the subject line) Podcast booking agencies - don't contact us under any circumstances, it's insulting to us (asking for "earned interviews" when you get paid top dollar to send automated GPT-invitations). We will also blacklist your clients.

Original URL: https://www.youtube.com/@MachineLearningStreetTalk

👍 Carly Batist

👍:skin_tone_4: Oc

Christine Muthee (cmuthee@andrew.cmu.edu)

2025-03-28 09:58:56

We also have The last week in AI for the latest gossip and some oddly deep conversations about AI.

Spotify

#204 - OpenAI Audio, Rubin GPUs, MCP, Zochi

Original URL: https://open.spotify.com/show/17HiNdxcoKJLLNibIAyUch

👍 Carly Batist

Carly Batist (carly.h.batist@gmail.com)

2025-03-26 09:40:42

Washington DC is hosting its first DC Climate Week this year, April 28-May 2!
Each day is themed: day 1 (climate innovation & tech), day 2 (policy & advocacy), day 3 (finance & economy), day 4 (nature), day 5 (mobility & water) and then there are a bunch of side events popping up too.
https://www.dcclimateweek.com/

DCCW

DC Climate Week | April 28 – May 2, 2025

DC Climate Week is the premier event for global climate action, taking place in Washington, D.C. from April 28 – May 2, 2025. Join leaders in climate technology, policy, and sustainability for a week of impactful discussions, innovative solutions, and strategic networking. Elevate your commitment to

Original URL: https://www.dcclimateweek.com/

🙌 Puneet Kollipara

Diogo Borges Provete (dbprovete@gmail.com)

2025-03-26 11:26:02

hello everyone! allow me to advertise a Special Collection I’m leading as guest editor for 70+ Wiley journals, focused on Climate Change and the 2030 UN SDGs. Unfortunately neither ESA nor BES journals are participating, but there’s plenty of great journals in the list, including society-owned ones like Austral Ecology and New Phytologist

Robert Dawes (robert.dawes@bbc.co.uk)

2025-03-27 13:02:33

I appreciate this is a bit of a MegaDetector specific question, but what happened to MD6a and MD6b? Are they essentially replaced by the current selection of models?

Dan Morris (agentmorris@gmail.com)

2025-03-27 13:07:25

Do you mean MDv5a and MDv5b?

Robert Dawes (robert.dawes@bbc.co.uk)

2025-03-27 13:11:10

Ha, perhaps that's what's thrown me and a and b were both on v5 (and my notes from using them are wrong!). They are both available via your https://github.com/agentmorris/MegaDetector/ I believe?

agentmorris/MegaDetector

MegaDetector is an AI model that helps conservation folks spend less time doing boring things with camera trap images.

Stars

144

Language

Jupyter Notebook

Dan Morris (agentmorris@gmail.com)

2025-03-27 13:13:36

Yes, MDv5a and MDv5b are alive and well, and yes, you can run them with code from that repo, or with the corresponding Python package:

https://pypi.org/project/megadetector/

👍 Robert Dawes

Dan Morris (agentmorris@gmail.com)

2025-03-27 13:14:06

Or with AddaxAI.

Robert Dawes (robert.dawes@bbc.co.uk)

2025-03-27 13:16:48

Is there a performance comparison anywhere? The Pytorch-Wildlife page talks about Animal Recall as a metric but I can't see any stats for the v6 models (or the v5 models). And I appreciate some work better in different scenarios.

Dan Morris (agentmorris@gmail.com)

2025-03-27 13:21:00

There are no published comparisons between MDv5 and MDv6 that I'm aware of. There are several independent evaluations of MDv4 and MDv5 (including comparisons of MDv5a and MDv5b). I list some of them here:

https://github.com/agentmorris/MegaDetector/blob/main/megadetector.md#have-you-evaluated-megadetectors-accuracy

I like to think I wasn't biased and that those are just the first few evaluations that I became aware of, but take it with a grain of salt that the person who worked on a model is sending you a list of evaluations, and consider doing some Google Scholar searches to make sure I'm not tricking you and/or tricking myself into having a rosy opinion of MDv5. To be fair to me, I do maintain a whole page telling you all the things MDv5 is terrible at:

https://github.com/agentmorris/MegaDetector/blob/main/megadetector-challenges.md

💡 Burak Ekim

Dan Morris (agentmorris@gmail.com)

2025-03-27 13:24:00

More generally, be cautious assuming that evaluations on Someone Else's Data (tm) will apply to your data. The only thing that matters is how a model performs on your data, so what I almost always recommend for new users is to run MD on your data, then generate a page like this one on a sample of your data (typically 7k-10k images), and in particular look carefully at the "non-detections", to make sure you are OK with a model's performance on your data:

https://lila.science/public/snapshot_safari_public/snapshot-safari-kar-2022-00-00-v5a.0.0_0.200/

Robert Dawes (robert.dawes@bbc.co.uk)

2025-03-27 13:58:59

Thank you for all this. I should say we have used MD in the past and MDv5a has been absolutely excellent for our usecase. So thanks for all your hard work.

👍 Dan Morris

❤️ Sara Beery

Jonathan Sauder (sauder@mit.edu)

2025-03-27 23:15:34

New dataset drop 🪸
The Coralscapes Dataset for semantic scene understanding in coral reefs! The first consistently and annotated dataset for dense general-purpose segmentation in reefs of the Red Sea! 2075 images, 174k segmentation masks, 39 classes across 5 countries in the Red Sea.

Coralscapes is deliberately designed to follow the structure of Cityscapes to make it easy to use!

Project website: https://josauder.github.io/coralscapes/
Preprint: https://arxiv.org/pdf/2503.20000

Big thanks to @Viktor Domazetoski, @Devis Tuia, the Transnational Red Sea Center at EPFL and its partners!

josauder.github.io

The Coralscapes Dataset

The Coralscapes Dataset: Semantic Scene Understanding in Coral Reefs

Original URL: https://josauder.github.io/coralscapes/

🙌 Rupa Kurinchi-Vendhan, Justin Kay, Shir Bar, Nina van Tiel, Filip Dorm, Alessandra Sellini, Robin Zbinden, Sara Beery, Timm Haucke, Elizabeth Campolongo, Diego Marcos, Andy Viet Huynh, Daniel Yang, Aleksis Pirinen, Owen Martin, Ghjulia Sialelli, Carly Batist, Akshit Gupta, Arthur Caillau, Roberta Hunt, Levi Cai, Nico Lang, Jenna Kline, Jason Holmberg (Wild Me)

🪸 Shir Bar, Filip Dorm, Alessandra Sellini, Robin Zbinden, S M Rayeed, Viktor Domazetoski, murilo, Dan Morris, Timm Haucke, Elizabeth Campolongo, Elena Sierra, Kate Rakelly, Andy Viet Huynh, Aleksis Pirinen, Ghjulia Sialelli, Josh Veitch-Michaelis, Nico Lang, Jason Holmberg (Wild Me), Alexander Merdian-Tarko

😎 Jon Van Oast, Andy Viet Huynh, Ghjulia Sialelli

❤️ Owen Martin, Ghjulia Sialelli, Navodita Mathur, Alan Stenhouse

Carly Batist (carly.h.batist@gmail.com)

2025-04-01 08:51:47

Conservation International is proud to announce the launch of the Nature Tech for Biodiversity Sector Map, developed in partnership with the Nature Tech Collective!!

Showcasing 600+ biodiversity tech organizations (many of which have strong AI components/products) this map will help practitioners, funders, and policymakers better use the technologies that exist to deliver these insights, and it highlights the gaps and barriers that still remain in order to drive collective action where it's needed most.

Key Insights
🔍 Tech gaps: 200+ orgs focus on monitoring, but data integration & interpretation lag - limiting real-world impact.
🧬📡 Adoption barriers: Despite impressive advancements in technologies like eDNA, bioacoustics, and remote sensing, practitioners face barriers to harnessing them fully, such as data interpretation, cost and implementation
🌍Uneven distribution: Biodiversity records are unevenly distributed across species (taxonomic bias) and locations (geographic bias), creating blind spots in conservation decision-making
🎯 Target-setting challenges: A lack of standardized guidance and diverse stakeholder needs make it harder for organizations to translate biodiversity data into actionable insights for different groups.
🤝 Democratization more a promise than reality? AI, camera traps & citizen science remain costly & complex.

🔗 Explore the map here: www.naturetechcollective.org/stories/biodiversity-sector-map

❤️ Sara Beery, Jenna Kline, Sara Si-Moussi, Wenxin Yang, Chris Cosma, Jon Van Oast, Gracie Ermi, Owen Martin, Elizabeth Campolongo, Jonathan Sauder, Jason Holmberg (Wild Me), Alexander Robillard, Alexander Shenkin, Alan Stenhouse

🎉 Malte Pedersen, Dan Morris, Jon Van Oast, Gracie Ermi, Jason Holmberg (Wild Me), Meredith Palmer, Alexander Robillard

Dan Stowell (dan.stowell@naturalis.nl)

2025-04-02 12:32:57

Just launched: the BioDCASE challenge! Monitor whales, birds and other animals, through their sounds? We're publishing new datasets, and new evaluations, to help you do so! Please join this new challenge, and please spread the word: https://biodcase.github.io/

👍 Holger Klinck, Oisin Mac Aodha, Benjamin Hoffman, Namitha Suresh, Reaz Masud, Jose Ruiz-Munoz, Sam Lapp, Navodita Mathur

🤩 Maddie Cusimano, Carly Batist, Sara Beery, Ben McEwen, Juan Sebastián Cañas

🔊 Benjamin Hoffman, Irina Tolkova

🐋 Nicolas Arrieta Larraza, Taiki Sakai - NOAA Affiliate, Dan Morris, Alexander Merdian-Tarko

🎉 Santiago Martinez Balvanera, Meredith Palmer, Ilyass Moummad, Aleksis Pirinen, Marius Miron, Elizabeth Campolongo

👍:skin_tone_3: Alan Stenhouse

Mani Ramaswami (mani.ramaswami@tcd.ie)

2025-04-03 03:34:45

Dear All, for including human-feedback in training AI to identify species (e.g iNaturalist/ Seek) I believe there was a built in mechanism for (a) rating the expertise of individuals giving feedback; and (b) assigning weights to feedback, based on the level of expertise.

Mani Ramaswami (mani.ramaswami@tcd.ie)

2025-04-03 03:36:15

Would someone have a reference to this? And also let me know if there one can access the level of expertise behind each "research grade" identification in Naturalist?

Leonardo Buitrago (leobuitragoh@gmail.com)

2025-04-03 12:08:01

Hi Mani,
That’s a very interesting question! This paper might be a good starting point: https://doi.org/10.1093/biosci/biad051.
I’ve also reached out to some GBIF colleagues who may have a deeper understanding of this. I’ll let you know if I find more specific information on the topic 😉

Leonardo Buitrago (leobuitragoh@gmail.com)

2025-04-03 12:18:39

One suggestion already came up, is to approach Grant Van Horn, for any detailed questions.

They summarize it each time they build it:
https://www.inaturalist.org/blog/91824-new-computer-vision-model-v2-12-with-1-983-new-taxa

The original paper I think is this, by Grant
https://openaccess.thecvf.com/content_cvpr_2018/papers/Van_Horn_The_INaturalist_Species_CVPR_2018_paper.pdf

iNaturalist

New Computer Vision Model (v2.12) with 1,983 new taxa

We released a new computer vision model today. It has 86,861 taxa up from 84,878. This new model (v2.12) was trained on data exported on February 02, 2024. Here's a graph of the models release schedule since early 2022 (segments extend from data export date to model release date) and how the number of species included in each model has increased over time. Thanks to some work by the team described here, we are going to start posting accuracy estimates with these model releases estimated against 1,000 random Research Grade observations in each group not seen during training time. The paired bars below compare average accuracy of model 2.11 with the new model 2.12. Each bar shows the accuracy from Computer Vision alone (dark green), Computer Vision + Geo (geen), and Computer Vision + Geo + Cropping Change (light green). 'Cropping Change' is a slight modification to the way images are prepared before they are sent to the CV model that resulted in an average 2.1% improvement. Overall t...

Original URL: https://www.inaturalist.org/blog/91824-new-computer-vision-model-v2-12-with-1-983-new-taxa

Leonardo Buitrago (leobuitragoh@gmail.com)

2025-04-03 12:19:54

In one of their early posts they (iNaturalist people) say species with at least 20 research-grade observations are being used for training:
https://www.inaturalist.org/pages/computer_vision_demo

iNaturalist

iNaturalist Computer Vision Explorations · iNaturalist

Original URL: https://www.inaturalist.org/pages/computer_vision_demo

Mani Ramaswami (mani.ramaswami@tcd.ie)

2025-04-04 02:41:40

Thanks Leonardo, I'll follow up on your the leads - but please do let me know if you hear more from your GBIF colleagues. (I am more an end user than a programmer). Thanks! Mani

Mani Ramaswami (mani.ramaswami@tcd.ie)

2025-04-04 02:50:39

Also Leonardo - I will follow up with Grant Van Horn. Best. Mani

😉 Leonardo Buitrago, Mani Ramaswami

stefano puliti (stefano.puliti@nibio.no)

2025-04-03 05:10:07

🚨 New forest vision paper on estimation of blueberry and lingonberry cover 🌲📷🫐

https://www.linkedin.com/feed/update/urn:li:activity:7313179747440635904/

linkedin.com

#mee #citizenscience #nfi_ai #deeplearning #biodiversity #forestry #nfi… | Stefano Puliti

🚨 NEW PAPER ALERT 🚨 In our latest PathFinder and SmartForest4.0 study published in British Ecological Society #MEE, we explored how AI can transform vegetation surveys by providing more objective and accurate estimates of understory cover—in our case for blueberry🍇 and lingonberry 🍒 species. Using smartphone images 🖼️ of vegetation plots (see 👇) collected as part of the Norwegian National Forest Inventory (NFI), we trained an Ultralytics YOLOv8 models to: 1) Detect vegetation plot frames; 2) Segment Vaccinium species, and 3) Estimate species-specific cover This is our first example of applying AI to NFI image data, and we&#39;re excited about the potential! Could this become the standard method for variables currently limited by subjectivity, high uncertainty, or lack of direct measurement—despite being clearly visible in field images? Many NFIs already collect field photos, yet these are hugely underused. Compared to #CitizenScience datasets, NFI images offer systematic, geographically balanced coverage at national scale—a major advantage for AI-driven ecological monitoring. 🌍📸 Stay tuned for more in our #NFI_AI series—there&#39;s much more coming through the Small4Good EU project! 💥 Huge congrats to Pauline Müller for leading this work and publishing her MSc thesis in Methods in Ecology and Evolution—an outstanding achievement! 🎉 📄 Paper: <https://lnkd.in/djF_u-iV> 📊 Data: <https://lnkd.in/dvZHNCmX> 💻 Code: <https://lnkd.in/drtMBxbd> #DeepLearning #Biodiversity #Forestry #NFI #SmartForests #AI4Nature #RemoteSensing #MEE

Original URL: https://www.linkedin.com/feed/update/urn:li:activity:7313179747440635904/

🫐 Binbin Xiang, Nico Lang, Justin Kay, Nora Gourmelon, Elizabeth Campolongo, Ethan White, Brian Geuther, Seven Song, Jan Dirk Wegner, Leonie Hodel, Oc, Alexander Robillard, Aleksis Pirinen, Alan Stenhouse, Alexander Merdian-Tarko

❤️ Sara Beery, Emily Lines, Josafat-Mattias Burmeister, Leonie Hodel, Oc, Alexander Robillard

Adrien Fontvielle (adrien@enlaps.fr)

2025-04-07 11:56:03

Hello, I am starting to scratch the surface on willdlife detection on images. So far I've tested SpeciesNetwhich is based on MDv5 (itself based on YOLO5) for the detection part. I am also looking on off-the-shelf "open object detection" networks which have the promise of detecting anything and are based on modern architecture (eg YOLOE, based on YOLO11). Some preliminary testing show it has some potential for my applications but I wanted to know if you've heard of some more formal comparison studies.

Kate Rakelly (krakelly@gmail.com)

2025-04-07 12:43:57

I’ve done some preliminary experiments with off-the-shelf YOLO-World for detecting animals in camera trap data. Qualitatively, I’d say it’s roughly on par with MD at detecting generic “animals”, but very bad at detecting different animal species. It was also decent at detecting “collars” on the deer.
I’m not aware of a VLM open-detection model like YOLO-World being fine-tuned on a big camera trap dataset, but that would be awesome if it exists or if someone is working on it.

AILab-CVC/YOLO-World

[CVPR 2024] Real-Time Open-Vocabulary Object Detection

Website

<https://www.yoloworld.cc>

Stars

5268

👍 Adrien Fontvielle

Hamzah Malik (hamzahmalik1221@gmail.com)

2025-04-07 12:17:32

Hi everyone! I am a Final Year MEng Computing (Visual Computing & Robotics) student at Imperial College London. My Final Year Project is on using machine learning within autonomous conservation drones to better plan routes based on wildlife movement so that they are not disturbed by the flight path and noise. I am looking for a suitable object detection model that is trained using the perspective of a drone that can identify wildlife animals from the environment below, if you can help me in my research in any way please do not hesitate to reach out! Thank you!
(I am also very new here and am not familiar with the layout of this Slack server so apologies if I need to look for help elsewhere 😅)

👋 Sara Beery, Elizabeth Campolongo

Ben Weinstein (benweinstein2010@gmail.com)

2025-04-07 12:19:57

sample image? https://deepforest.readthedocs.io/en/v1.5.0/user_guide/02_prebuilt.html#bird-detection-model starting from birds transfers well to finetuning for other wildlife.

David Russell (davidrussell327@gmail.com)

2025-04-07 14:44:44

This sounds like the same application that Rowan Converse (a collaborator of both Ben and I) is working, but more from the applied management perspective

Hamzah Malik (hamzahmalik1221@gmail.com)

2025-04-07 17:23:04

Oh nice thank you for pointing me towards DeepForest! I had not come across it before and i think this will be very useful!

Hamzah Malik (hamzahmalik1221@gmail.com)

2025-04-07 17:23:44

Oh interesting, I'll reach out to Rowan to see if I can get more details, thank you!

👍 David Russell

David Russell (davidrussell327@gmail.com)

2025-04-07 18:26:04

Sorry I didn't realize she was here @Rowan Converse

Roberta Hunt (r.hunt@di.ku.dk)

2025-04-08 05:02:10

Maybe there are some helpful references in this as well https://besjournals.onlinelibrary.wiley.com/doi/pdfdirect/10.1111/1365-2656.13904

Elizabeth Campolongo (e.campolongo479@gmail.com)

2025-04-14 11:44:31

@Jenna Kline has done a lot of work in this area (e.g., wildwing)

imageomics.github.io

wildwing

Autonomous UAS Software for In Situ Imageomics Missions

Original URL: https://imageomics.github.io/wildwing/

Hamzah Malik (hamzahmalik1221@gmail.com)

2025-04-14 14:36:45

Oh nice! Thank you all for the links theyre quite useful!

Hamzah Malik (hamzahmalik1221@gmail.com)

2025-04-14 14:39:39

My biggest hurdle right now is getting sufficient training/evaluating data that is of aerial drone footage that has bounding boxes on wildlife (bonus points for 3D data). I think I've downloaded the datasets used in the two papers mentioned above but either the videos are too short with not much in them or the file structures of the downloaded directory isnt super intuitive to navigate. Is it possible that you guys know where I can find a good dataset for my project? 🙂

Jenna Kline (jennamkline@gmail.com)

2025-04-14 15:20:12

I don't know of any 3D data, but I'm working on releasing a 155K image drone dataset that spans 3 locations and 5 species. You can find the details in our preprint here - https://arxiv.org/abs/2504.07744. We're working on making it publicly available on HF. If this looks like it could be useful to your project, I'd be happy to let you know once it's available and answer any questions about the dataset.

arXiv.org

MMLA: Multi-Environment, Multi-Species, Low-Altitude Aerial Footage Dataset

Real-time wildlife detection in drone imagery is critical for numerous applications, including animal ecology, conservation, and biodiversity monitoring. Low-altitude drone missions are effective for collecting fine-grained animal movement and behavior data, particularly if missions are automated for increased speed and consistency. However, little work exists on evaluating computer vision models on low-altitude aerial imagery and generalizability across different species and settings. To fill this gap, we present a novel multi-environment, multi-species, low-altitude aerial footage (MMLA) dataset. MMLA consists of drone footage collected across three diverse environments: Ol Pejeta Conservancy and Mpala Research Centre in Kenya, and The Wilds Conservation Center in Ohio, which includes five species: Plains zebras, Grevy's zebras, giraffes, onagers, and African Painted Dogs. We comprehensively evaluate three YOLO models (YOLOv5m, YOLOv8m, and YOLOv11m) for detecting animals. Results demonstrate significant performance disparities across locations and species-specific detection variations. Our work highlights the importance of evaluating detection algorithms across different environments for robust wildlife monitoring applications using drones.

Original URL: https://arxiv.org/abs/2504.07744

🙌 David Russell

David Russell (davidrussell327@gmail.com)

2025-04-14 15:24:50

The problem I was working on with Rowan was actually going from per-image detections across multiple views to 3D localization. The idea was to geometrically identify when the same individual was observed across multiple views and then triangulate based on all observations. Some visuals are summarized in this notebook here. It was definitely very preliminary work but I thought it was promising.

David Russell (davidrussell327@gmail.com)

2025-04-14 15:25:29

Sorry, the reason I said that was because the dataset you're releasing looks really cool and it seems like it could be directly used with this type of 3D localization.

👍 Jenna Kline

Jenna Kline (jennamkline@gmail.com)

2025-04-14 15:28:43

That's great! I previously tried some generic photogrammetry approaches to localize the animals, but the oblique imagery proved to be pretty challenging with drift, camera calibration, etc, so glad to hear others are working on this 🙂

David Russell (davidrussell327@gmail.com)

2025-04-14 15:35:24

Yeah the first step in our pipeline is photogrammetry as well, but the big takeaway for me is the photogrammetry mesh is not accurate enough for localization tasks especially if your animals are small and you have oblique views. So that's why we were exploring more triangulation approaches that only use the camera location/orientation from photogrammetry but not the mesh. We haven't been actively working on this recently but I'm happy to chat if you're interested and still working on this.

Hamzah Malik (hamzahmalik1221@gmail.com)

2025-04-14 21:27:18

I'm currently testing out the DeepForest Livestock model that predicts cattle-like animal bounding boxes. It seems to be able to do it, although when inspecting the data frame that comes out of the model, it seems to label everything as a "Tree"? I can't find anywhere in the docs referring to this, is this a bug or am I using this model wrong in some way?

Ben Weinstein (benweinstein2010@gmail.com)

2025-04-14 21:27:59

Just because you didn't specify a label.

Ben Weinstein (benweinstein2010@gmail.com)

2025-04-14 21:28:45

We can fix that, when you load the model, m.label_dict = {0:""}

Hamzah Malik (hamzahmalik1221@gmail.com)

2025-04-14 21:32:36

Unfortunately this didn't fix it, unless I also messed this up too haha

Ben Weinstein (benweinstein2010@gmail.com)

2025-04-14 21:33:35

I just made an issue, what about when you load the model

model = main.deepforest(label_dict={0:"Animal"})

Hamzah Malik (hamzahmalik1221@gmail.com)

2025-04-14 21:36:58

Unfortunately again this did not work 😞

Ben Weinstein (benweinstein2010@gmail.com)

2025-04-14 21:37:24

okay, I made an issue, we'll make it happen, let's continue the discussion here. https://github.com/weecology/DeepForest/issues/1027, we don't want these kind of things in general thread taking up other people's notifications.

Hamzah Malik (hamzahmalik1221@gmail.com)

2025-04-14 21:37:28

oh maybe it needs to be string: int

Hamzah Malik (hamzahmalik1221@gmail.com)

2025-04-14 21:38:39

nope didnt work

Ben Weinstein (benweinstein2010@gmail.com)

2025-04-14 21:38:53

please let's move to that issue.

🫡 Hamzah Malik

Hamzah Malik (hamzahmalik1221@gmail.com)

2025-04-15 06:43:39

The issue has been fixed by Keerthi Reddy 🙂

Alexander Robillard (RobillardA@SI.EDU)

2025-04-08 10:17:28

Hi All, proud to share a link to our new paper leveraging a large aerial image data set to identify routes across the Mojave desert tortoise's habitat.
Wildlabs Discussion thread: https://wildlabs.net/article/application-computer-vision-highway-vehicle-route-detection-case-study-mojave-desert
Direct Link to paper: https://zslpublications.onlinelibrary.wiley.com/doi/full/10.1002/rse2.70004
Looking for my next collaboration (please see below)-
Linkedin: https://www.linkedin.com/in/alexander-robillard-ph-d-17a6521bb/
Researchgate: https://www.researchgate.net/profile/Alexander-Robillard?ev=hdr_xprf

wildlabs.net

Application of computer vision for off-highway vehicle route detection: A case study in Mojave desert tortoise habitat | WILDLABS

Driving off-highway vehicles (OHVs), which contributes to habitat degradation and fragmentation, is a common recreational activity in the United States and other parts of the world, particularly in desert environments with fragile ecosystems. Although habitat degradation and mortality from the expansion of OHV networks are thought to have major impacts on desert species, comprehensive maps of OHV route networks and their changes are poorly understood. To better understand how OHV route networks have evolved in the Mojave Desert ecoregion, we developed a computer vision approach to estimate OHV route location and density across the range of the Mojave desert tortoise (Gopherus agassizii). We defined OHV routes as non-paved, linear features, including designated routes and washes in the presence of non-paved routes. Using contemporary (n = 1499) and historical (n = 1148) aerial images, we trained and validated three convolutional neural network (CNN) models. We cross-examined each model on sets of independently curated data and selected the highest performing model to generate predictions across the tortoise's range. When evaluated against a ‘hybrid’ test set (n = 1807 images), the final hybrid model achieved an accuracy of 77%. We then applied our model to remotely sensed imagery from across the tortoise's range and generated spatial layers of OHV route density for the 1970s, 1980s, 2010s, and 2020s. We examined OHV route density within tortoise conservation areas (TCA) and recovery units (RU) within the range of the species. Results showed an increase in the OHV route density in both TCAs (8.45%) and RUs (7.85%) from 1980 to 2020. Ordinal logistic regression indicated a strong correlation (OR = 1.01, P

Original URL: https://wildlabs.net/article/application-computer-vision-highway-vehicle-route-detection-case-study-mojave-desert

❤️ Sara Beery, Timm Haucke, Declan, Owen Martin, Vanesa Reyes, Elizabeth Campolongo, Talia Speaker

🎉 Dan Morris, Lukas Picek, Alexander Merdian-Tarko

👏 Kostas Papafitsoros, Kate Rakelly

Patrick Beukema (patrickb@allenai.org)

2025-04-08 13:59:21

I asked this last year, asking again to see if anything has changed. Are there any open source (or commercial) tools that supports GPS trajectory annotations in a “first-class” way (i.e. not as a hack/after-thought).

👀 Alan Stenhouse

Patrick Beukema (patrickb@allenai.org)

2025-04-08 14:00:49

Here is an example of what I am talking about in a maritime context:
https://www.datocms-assets.com/64837/1729029922-skylight-data-1.gif?fit=max&fm=webp&h=810&w=1550

Original URL: https://www.datocms-assets.com/64837/1729029922-skylight-data-1.gif?fit=max&fm=webp&h=810&w=1550

Patrick Beukema (patrickb@allenai.org)

2025-04-08 14:01:40

the annotator here is annotating each individual GPS message with some label like “anchored, fishing, transiting” etc. And they are able to filter this data/color code by things like Speed over ground, course over ground, gap, etc

Patrick Beukema (patrickb@allenai.org)

2025-04-08 14:10:24

Thread Reply:

Sara Beery (sara.beery@gmail.com)

2025-04-08 16:31:49

https://research.google/blog/geospatial-reasoning-unlocking-insights-with-generative-ai-and-multiple-foundation-models/

Pretty cool

🧡 Wenxin Yang, Elena Sierra, murilo, Jason Holmberg (Wild Me), Ian McGregor, Shravan Ambudkar, Shadman Tabib

😎 Jon Van Oast, Mitchell Rogers, Tyler Marino, Joe Ferdinando, Jason Holmberg (Wild Me)

:female_fairy: Burak Ekim

:female_fairy::skin_tone_3: Jess Tam

🙌 Carly Batist, Aleksis Pirinen

Leonardo Buitrago (leobuitragoh@gmail.com)

2025-04-09 14:46:46

Hi everyone,
I’d like to share a new opportunity for colleagues in the Pacific Islands and Latin America & the Caribbean 🌍
The Biodiversity Information for Development (BID) programme — funded by the EU and coordinated by GBIF — has launched a new call for proposals to support projects that mobilize biodiversity data and promote its use in research and policymaking.
🌱 Details for LAC region: https://www.gbif.org/news/psbfUTIrFlz0GBmfNfShW/2025-bid-call-for-proposals-latin-america-and-the-caribbean
🐠 Details for the Pacific: https://www.gbif.org/news/7CnFmwFVne2kYDXIWzKkQY/2025-bid-call-for-proposals-pacific

Feel free to share with anyone who might be interested! 😉

gbif.org

BID: Biodiversity Information for Development

This multi-year EU-funded programme aims to increase the amount of biodiversity information available in sub-Saharan Africa, the Caribbean and the Pacific

Original URL: https://www.gbif.org/bid

gbif.org

2025 BID call for proposals: Latin America & the Caribbean

Deadline for submissions: **27 May 2025**

Original URL: https://www.gbif.org/news/psbfUTIrFlz0GBmfNfShW/2025-bid-call-for-proposals-latin-america-and-the-caribbean

gbif.org

2025 BID call for proposals: Pacific

Deadline for submissions: **27 May 2025**

Original URL: https://www.gbif.org/news/7CnFmwFVne2kYDXIWzKkQY/2025-bid-call-for-proposals-pacific

🎉 Jon Van Oast, Mitchell Rogers, Sara Beery, Manuela Ramos, Alan Stenhouse

murilo (murilogustineli@gatech.edu)

2025-04-14 15:25:24

Hi everyone! I'll be attending and presenting a workshop at ODSC next month (May 13-15) in Boston. It'll be my first time in Boston and I'd love to visit MIT while I'm there. If you're at MIT and open to meeting up or chatting, feel free to reach out 🙂

I don't remember if I've introduced myself. My name is Murilo, I work as a senior AI software engineer at Intel, and I'm also pursuing a master's in CS at Georgia Tech. I'm passionate about deep learning, information retrieval, and biodiversity conversation research. Right now, I'm participating in the PlantCLEF competition on Kaggle, which focuses on multi-label plant species classification in images.

If you're curious, you can learn more about my work here: https://murilogustineli.com/

Looking forward to connecting!

Open Data Science Conference

ODSC East 2025

At the ODSC East 2024 data science conference, you can learn trending and in-demand skills in machine learning, deep learning, NLP, and more.

Est. reading time

167 minutes

Original URL: https://odsc.com/boston/

murilogustineli.com

Murilo Gustineli

A highly-customizable Hugo résumé theme powered by Hugo Blox Builder.

Original URL: https://murilogustineli.com/

❤️ Adrian Cheung

Carly Batist (carly.h.batist@gmail.com)

2025-04-15 07:06:52

Conservation Tech meet-up for DC Climate Week!
https://wildlabs.net/event/wildlabs-climate-week-conservation-tech-meetup-washington-dc

Date: Tuesday April 29, 2025

Time: 7pm - 10pm

Location: The Admiral (Dupont Circle)

wildlabs.net

WILDLABS Climate Week Conservation Tech Meetup: Washington, DC | WILDLABS

We're excited to participate in DC's first-ever climate week with a lively happy hour meetup hosted by WILDLABS! Join us at The Admiral in Dupont Circle to connect, unwind, and kick off inspiring conversations that will continue throughout the week.

Original URL: https://wildlabs.net/event/wildlabs-climate-week-conservation-tech-meetup-washington-dc

💚 Millie Chapman, Alex Rood, Catherine Breen, Andy Viet Huynh, Vanesa Reyes, Bernie Boscoe, murilo, Puneet Kollipara, Catherine, Lily Xu, Talia Speaker

Lily Xu (lily.x@columbia.edu)

2025-04-23 11:03:55

I'm hiring for a machine learning data scientist & research assistant for summer 2025!

Join me in working on a project on invasive species management, by predicting species risk and planning capture strategies. This work will be in partnership with an innovative startup that is working directly to capture environmentally destructive invasive species in the US.

This project will be working with a US-based conservation partner to build predictive machine learning models and design harvest strategies for invasive species removal. The primary goal is to develop usable ML models and optimization tools to inform practical environmental decision-making on the ground.

There will also be opportunities to extend this work into a research publication for a top-tier AI venue.

The project would be:

Paid, full-time internship for summer 2025

Possibility of extension beyond the summer (part-time or full-time)

Remote, but candidates should be authorized to work in the US

Supervised by Lily Xu, assistant professor at Columbia University

Ideal candidate background will include:

Strong background in CS, data science, and/or applied math

Experience developing machine learning models

Excellent coding skills in Python

Excellent writing and interpersonal communication skills

Genuine interest in conservation/sustainability

Nice to have: background in optimization methods

Nice to have: experience with GIS and geospatial data

Candidates would ideally have already completed an undergraduate degree, but exceptional undergrads will also be considered.

How to apply:
Please send a CV and 3–5 paragraphs of your background/interests via email to <lily.x@columbia.edu> with the subject line "Application: ML for invasive species management". Applications will be reviewed on a rolling basis.

🌱 Eric Greenlee, Timm Haucke, Mark Goldwater, Tyler Marino, Neha Hulkund, Sara Beery, Andy Viet Huynh, Elaine Steinberg, Ben Weinstein, Millie Chapman, Konstantin Klemmer, gyeongsu, Carlos Rodríguez - Pardo, Irina Tolkova, Puneet Kollipara, Jennifer

🙌 Carlos Rodríguez - Pardo

Konstantin Klemmer (konstantin.klemmer16@alumni.imperial.ac.uk)

2025-04-23 20:40:51

Do you have social media posts for this Lily? If so lmk and we can share from CCAI account!

Lily Xu (lily.x@columbia.edu)

2025-04-24 21:54:03

Thank you so much, @Konstantin Klemmer!!

I juts posted:
https://x.com/lilyxu0/status/1915580408366063790
https://bsky.app/profile/lilyxu.bsky.social/post/3lnm4bl4gm22f
https://www.linkedin.com/posts/lilylxu_im-hiring-for-a-machine-learning-data-scientis[…]mber_desktop_web&rcm=ACoAABCUOkwBTYFIoJMiS9LW4QUz9H1mg3s89Sg

🎉 Konstantin Klemmer, Puneet Kollipara

Jennifer (jzhuge@alumni.cmu.edu)

2025-05-30 15:15:07

just sent an email, looking forward to your response!

Marzi (marziyeh.mohammadi@fau.de)

2025-06-03 14:31:57

Hi, I wanted to ask if the position is still available.

Lily Xu (lily.x@columbia.edu)

2025-06-03 14:45:48

Thank you for your interest but the position has been filled.

🙏 Marzi

Justin Chung Clark (justin@drivendata.org)

2025-04-24 12:08:47

🎉 ZAMBA CLOUD NOW SUPPORTS IMAGES! 🎉

Hello conservation community!

We're excited to announce that Zamba Cloud—our wildlife identification platform that has been helping researchers analyze camera trap videos—now supports still images too!

What is Zamba Cloud?
Zamba Cloud is a user-friendly web application that uses advanced AI to automatically identify wildlife in camera trap footage. While we've previously focused on videos, we're now expanding to support still images based on your feedback!

New Features:
- Image Processing: Upload your camera trap photos to identify wildlife (in addition to our existing video support)
- Powerful AI Recognition: Our default model recognizes 178 different animal taxa in images
- Automatic Detection: The system finds animals in your images and draws boxes around them
- Customization Without Coding: You can "fine-tune" our model (teach it to better recognize your specific animals) by uploading labeled examples from your project

Why This Matters:
Many researchers use both video AND image-based camera traps. Now Zamba Cloud provides a unified solution for all your wildlife identification needs, saving countless hours of manual review and accelerating conservation work.

How to Get Started:
1. Visit zambacloud.com
2. Create an account or log in
3. Upload your images
4. Get results in an easy-to-use spreadsheet!

For Developers:
All Zamba Cloud features, including the new image support, are available in our open-source Zamba Python library for those who prefer to code: github.com/drivendataorg/zamba

Questions about our new image support? Happy to chat!

📸 🐘 🐆 🦊 🐻

🙌 David Bennett, Dan Morris, Sara Beery, Emily Dorne, Carly Batist, Robin Zbinden, Dante Wasmuht, Jason Holmberg (Wild Me), Owen Martin, Alexander Merdian-Tarko

❤️ Jennifer, Peter Bull, Cara Appel, Sara Beery, Gaspard Dussert, Emily Dorne, Dante Wasmuht, Jason Holmberg (Wild Me), Owen Martin

Patrik Reiske (nailed-flow-0a@icloud.com)

2025-04-25 04:48:07

Hi from Berlin, everyone. I’m new to the workshop and was wondering if the reviews will be published (to the authors) before the decision. Thanks

Lukas Picek (lukaspicek@gmail.com)

2025-04-25 05:43:11

Hi Patrik, could you clarify which workshop you’re referring to? This Slack channel is not tied to a specific workshop.

Patrik Reiske (nailed-flow-0a@icloud.com)

2025-04-25 14:29:20

CVPR’s CV4Animals

Lukas Picek (lukaspicek@gmail.com)

2025-04-26 04:25:19

Then, the best person to ask should be @Urs.

Urs (urs.waldmann@uni-konstanz.de)

2025-05-04 05:41:36

Hi Patrik,

Sorry I saw the thread only now.

We released the decisions and you should see your reviews now.

Btw, there is a CV4Animals channel here that you can join.

Best regards,
Urs

Patrik Reiske (nailed-flow-0a@icloud.com)

2025-05-04 07:54:21

Hi Urs. Thanks for letting me know. Joined the channel

Dan Morris (agentmorris@gmail.com)

2025-04-26 11:03:17

New dataset on LILA, courtesy of the MIT Sea Grant program:

https://lila.science/datasets/mit-sea-grant-river-herring/

~90k bounding boxes on freshwater fish, primarily herring.

❤️ Sara Beery, Peter Bull, Timm Haucke, Andy Viet Huynh, Mark Goldwater, Ted Schmitt, Malte Pedersen, Bernie Boscoe, Sonny Burniston, murilo, Arthur Caillau, Rita Pucci, Justin Chung Clark, Gracie Ermi, Edina, Devi Ayyagari

🐟 Shir Bar, Timm Haucke, Andy Viet Huynh, Mark Goldwater, murilo, Jonathan Sauder, Emily Dorne, Justin Chung Clark, Elizabeth Campolongo, Alan Stenhouse

🤩 Atul Ingle, Mark Goldwater

😎 Jon Van Oast

:fishcowboy: Laura Chrobak, Alexander Merdian-Tarko

Matthias Zuerl (matthias.zuerl@fau.de)

2025-04-28 02:40:45

@Talen Rimmer

Puneet Kollipara (kollipython@gmail.com)

2025-04-28 09:59:08

Folks in Washington, DC this week for DC Climate Week and/or MeasureDev: Biodiversity on Land and at Sea, there's still time to RSVP for Tuesday evening's Nature + Conservation Tech Meetup! Put on by WILDLABS.net and Society for Conservation GIS, this happy hour will take place 7:00–10:00 PM on Tuesday, April 29, at The Admiral in Dupont Circle. Join me and @Carly Batist for a lively evening of climate and conservation conversations and networking.

Please note that the registration page has changed since the event's announcement: click here to RSVP. Hope to see many of you there!

lu.ma

Nature + Conservation Tech Meetup · Luma

Join WILDLABS, the global conservation technology network, and the Society for Conservation GIS for a DC Climate Week happy hour meetup! We'll gather in Dupont…

Original URL: https://lu.ma/ndiq78j8?tk=TpAhiy&utm_source=aiforconservation_slack

🙌 Carly Batist, Jason Holmberg (Wild Me), Andy Viet Huynh

💚 Millie Chapman, Jason Holmberg (Wild Me), Andy Viet Huynh, Talia Speaker

Patrick Beukema (patrickb@allenai.org)

2025-04-28 13:50:34

Hey all,

We have been testing some new real-time GPS models for maritime intelligence/marine conservation use cases for the last several months. We have now open sourced the models/weights, training and inference code, as well as the service/package that is running live in production now in Skylight. This is a bespoke GPS transformer we created for this use case. It processes ~5B messages/day on modest hardware (5 T4 GPUs). You can read more about the model and our approach in this blog post or check out the technical paper which was just presented as a spotlight yesterday at ICLR (CCAI). Somewhat unique here is that the code we are open sourcing is the production code.

Ai2 Blog

GitHub

Paper

Technical talk by Henry Herzog (lead researcher)

We have some exciting work planned with some researchers in this group to extend these models to human wildlife conflict modeling.

Very grateful to this community 💙.

allenai.org

Introducing Atlantes: the first AI-powered GPS model for real-time global scale maritime intelligence | Ai2

Atlantes: a system of transformers for real-time GPS modeling.

Original URL: https://allenai.org/blog/atlantes

🙌 Justin Kay, Arthur Caillau, Sara Beery, Peter Bull, Shir Bar, Dan Morris, Andy Viet Huynh, Laura Chrobak, Elizabeth Campolongo, Alexander Merdian-Tarko

❤️ Sara Beery, Andy Viet Huynh, Ștefan Istrate, Edina, Ben Williams

🛳️ Shir Bar, Andy Viet Huynh, Filip Dorm

🎉 Christine Muthee

Puneet Kollipara (kollipython@gmail.com)

2025-04-28 15:59:46

Someone on the ESIP Slack just posted about the rapidly approaching deadline for the 2025 NCEAS Morpho Initiative RfPs.

Thought I'd cross-post it here: https://docs.google.com/document/d/19OzaLZbeV2aLUV1Mdncga5L_QpbulqUU/edit

National Center for Ecological Analysis and Synthesis

Morpho Initiative

Morpho Initiative Collaborative data science in service of environmental management working groups synthesis science team science

Original URL: https://www.nceas.ucsb.edu/morpho

Chris Cosma (chris.cosma@consbio.org)

2025-04-28 16:18:49

I’m co-leading one of the current groups if anyone has any question!

Eugene Galaxy (eugene@animaldetect.com)

2025-04-30 05:05:10

Hey everyone,

WE ARE FINALLY LIVE! Animal Detect platform is live! 🎉

After 8 months of hard work with @Hugo we are ready to present the first stable version of Animal Detect!

Animal Detect is built for ecologists and wildlife researchers who want to process and organize their wildlife data quickly, easily, and without any technical hassle.
With Animal Detect you can:

Create Wildlife projects and manage them

Upload your dataset of wildlife images

Use our AI tools to filter empty shots, label detections, sort & categorize animals

Download results

Instantly generate pie charts and label data for further AI-training

Share a dynamic map with detection locations (if your images have GPS tags)

Our goal is to make your research workflow smoother and more efficient, so you can focus on what matters most - your discoveries.

Ready to get started?
Log in to Animal Detect
New users start with 1GB of free data upload so you can try our services without any stress.

What's next?
Our pipeline is already loaded with BIG additions that will make your wildlife image processing even faster and easier:

SpeciesNet support

Fully Automated mode (from clicking “Go!” to results in your email in 1 click)

How it looks? See screenshots below. 🙂

If you have any questions or feedback, just write to us. We’d love to hear from you!

👍 Dante Wasmuht, Gaspard Dussert, Andy Viet Huynh, Navodita Mathur, Jason Holmberg (Wild Me)

👀 Irina Tolkova, Andy Viet Huynh, Jason Holmberg (Wild Me)

👏 Jose Ruiz-Munoz, Sara Beery, murilo, Andy Viet Huynh, Jason Holmberg (Wild Me)

🎉 Sara Beery, Yuyan Chen, Dan Morris, Andy Viet Huynh, Carly Batist, Helena Russello, Alan Stenhouse, Brian Geuther, Jason Holmberg (Wild Me), Alexander Merdian-Tarko

Steve Haddock (haddock@mbari.org)

2025-05-01 16:01:00

@Danelle Cline 👀

❤️ Eugene Galaxy

Danelle Cline (dcline@mbari.org)

2025-05-20 14:06:15

thanks @Steve Haddock Nice tool! Gave this a quick try with images that had batrays, sharks and birds. Lower confidence values produced more detections as expected. It would be great to try our yolo11x-uavs-detect model here trained by @Duane Edgington

huggingface.co

MBARI-org/yolo11x-uavs-detect · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Original URL: https://huggingface.co/MBARI-org/yolo11x-uavs-detect

👀 Steve Haddock

🎉 Eugene Galaxy

Eugene Galaxy (eugene@animaldetect.com)

2025-07-10 03:22:41

Hey @Danelle Cline and @Steve Haddock. Thanks for trying it out! We are constantly improving the pipeline using feedback from the users. Three weeks ago we have implemented SpeciesNet classification (2000+ animal classes) and added possibility to annotate detections using Wildlife Insight's taxonomy library (can be found here).

Great suggestion about your yolo11x-uavs model, we would be extremely curious to try it out and see if we can help you anyhow.

Would you like to chat over a call and introduce each other? 😊

Steve Haddock (haddock@mbari.org)

2025-07-10 04:08:16

Thanks for following up Eugene.
@Danelle Cline if you can arrange a call some time in the next couple weeks, I'd love to eavesdrop, but would not have anything technical to contribute.

Eugene Galaxy (eugene@animaldetect.com)

2025-07-10 08:56:05

@Danelle Cline you can use my calendly calendar to find a time that fits you: https://calendly.com/eugene-animaldetect/book-a-meeting. Its possible to invite guests to the call as well. 😊

Looking forward talking to you!

Niki Amini-Naieni (niki@adainnovations.ai)

2025-05-01 06:44:31

Free Tool (CountGD) to Automatically Count Animals in Images with AI

Hi Everyone, we are testing and improving our new open-world counting model CountGD (published at NeurIPS 2024). It can be used to automatically count in theory any specified object in images given a text description of the object, visual examples, or both. It does not require retraining or data annotation to generalize to new objects or images. Given an input image and a specification of the object to count, it outputs x-y coordinates of all the identified objects in the image and a final count (i.e., the number of x-y coordinates). All the code is open-source, and we have an app where you can upload and test your own images. Please share your experience on our WILDLABS discussion page here, so we can know for what applications the model works well and where we could improve it. Below I have included some useful resources:

CountGD WILDLABS Discussion Page

CountGD WILDLABS Product Page

CountGD App to Test Out the Model

CountGD Open-Source Code

Annolid Data Annotation Tool That Integrates CountGD

LinkedIn Post About CountGD's Accuracy Across a Wide Range of Applications (e.g., counting sea birds in aerial images, as a data annotation tool, counting buildings in satellite images, counting seals on beaches, etc...)

CountGD NeurIPS Conference Publication

robots.ox.ac.uk

CountGD: Multi-Modal Open-World Counting

Original URL: https://www.robots.ox.ac.uk/~vgg/research/countgd/

wildlabs.net

Counting Problems in Conservation | WILDLABS

We're actively exploring new applications for CountGD, our object counting model designed to automatically count instances in images. So far, we've partnered with bird experts to count birds in aerial imagery, helping assess the impact of climate change on population trends. We're now looking to expand into other domains where automated counting could make a difference—if you have a relevant use case or idea, we'd love to hear from you!Additionally, if you've tested CountGD on your own dataset, we’re eager to hear your feedback—whether it performed well or revealed opportunities for improvement. Your input will help guide future development and improvements.

Original URL: https://wildlabs.net/discussion/counting-problems-conservation

The Inventory | WILDLABS

CountGD -- count any object in images automatically on The Inventory | WILDLABS

Count any object in images automatically using a text description only, visual examples only, or both.

Original URL: https://wildlabs.net/inventory/products/countgd-count-any-object-images-automatically

huggingface.co

CountGD_Multi-Modal_Open-World_Counting - a Hugging Face Space by nikigoli

This app counts objects in images. You can use text to describe the object, draw boxes around examples in the image, or use both. After providing the inputs, click "Count" to get the predicted coun...

Original URL: https://huggingface.co/spaces/nikigoli/countgd

❤️ Ștefan Istrate, Constanza Andrea Molina Catricheo, Dante Wasmuht, Eugene Galaxy, Sara Beery, Elizabeth Campolongo, Roberta Hunt, Gustavo Perez, Chris Cosma, Aakash Gupta, Cody Kupferschmidt, Alan Stenhouse, Talia Speaker

👍 Dan Morris, Jason Holmberg (Wild Me), Chris Cosma

👏 Silvia Zuffi, Jason Holmberg (Wild Me), Chris Cosma, Malte Pedersen, Naveen Dhar

🙌 Owen Martin, Emilio Luz-Ricca, Mitchell Rogers, Alexander Merdian-Tarko

Dante Wasmuht (dante@conservationxlabs.org)

2025-05-01 07:39:53

@Lasha Otarashvili @Jason Holmberg (Wild Me)

Ben Weinstein (benweinstein2010@gmail.com)

2025-05-01 09:55:51

I'd be interested in comparing against this and discussing more. I'll have a try.

🙏 Niki Amini-Naieni

Niki Amini-Naieni (niki@adainnovations.ai)

2025-05-01 09:59:50

Awesome @Ben Weinstein thank you!

Ben Weinstein (benweinstein2010@gmail.com)

2025-05-01 14:27:29

Just glancing at the app and the repo and the paper. A couple things that jump out at me, 1) those install instructions might need to be vetted and perhaps maybe a docker container, or something to reduce the challenge, is python 3.9 really needed? From 2020? Is that particular version of gcc needed? 2) couple demos show general SAM performance, clear objects are well defined, occluded objects are missed (about half the birds in this image). 3) hit or miss on semantic concepts, 'cattle' missed entirely on this test. 4) Is the visual exemplar on the per crop level, as in the figure in the paper, or an example segmentation, as in the prediction. 5) How difficult is it to fine-tune, if we could produce millions of GT images of biodiversity objects in more realistic airborne biodiversity monitoring scenarios, I can see it has utility as a general annotation starting point, connecting to something like label-studio backend. We are writing code for this in other, less sophisticated models.

Niki Amini-Naieni (niki@adainnovations.ai)

2025-05-01 17:41:55

Thank you for your feedback!

RE 1) We have a docker image available in the app. I will reply with a picture of the command. The app is mainly for the user. The codebase is mainly for AI researchers who want to reproduce the results of the paper. We have a Colab notebook as well accessible via the GitHub repository. Additional packages and tooling are currently in the making.

RE 2) Could you clarify what you are referring to? SAM is the Segment Anything Model by Meta. Our model rarely uses SAM (usually it does not at all). The app does not use it at all.

Re 3) The text should be the name of one instance of the object. I would try "white animal." See the attached image. If the object does not look similar to common instances of the animal, it helps to describe what the object does look like (which is why I said "white animal"). A visual exemplar would also probably fix the issue since you are telling the model visually what the object looks like.

Re 4) the visual exemplar is on the per object level.

Re 5) We are currently working on instructions for fine-tuning on custom datasets. However, perhaps with the proper visual exemplars and text this may not be needed, unless the images are really hard. I think this cattle example does not look that hard.

Niki Amini-Naieni (niki@adainnovations.ai)

2025-05-01 17:44:58

Thread Reply:

Niki Amini-Naieni (niki@adainnovations.ai)

2025-05-01 17:48:41

Here are some examples of visual exemplars for the bird example:

Niki Amini-Naieni (niki@adainnovations.ai)

2025-05-01 17:49:25

Also, with the text you specify it for one object at a time. I need to make that clearer in the WILDLABS inventory

Niki Amini-Naieni (niki@adainnovations.ai)

2025-05-01 17:50:05

I picked 3 random objects I thought were birds. It was hard for me to tell.

Niki Amini-Naieni (niki@adainnovations.ai)

2025-05-01 18:10:39

I forgot to say, yes, if you train it on millions of biodiversity images, I imagine it would be very good for this application

Ben Weinstein (benweinstein2010@gmail.com)

2025-05-01 18:19:15

brilliant. We are developing such a dataset for airborne imagery, let's stay in touch, connecting to https://deepforest.readthedocs.io/, and an active learning workflow. We would of course cite and direct the user the appropriate attribution.

🎉 Niki Amini-Naieni

Ben Weinstein (benweinstein2010@gmail.com)

2025-05-01 18:21:18

the docker container is particularly interesting because label studio allows docker for the ML backend, you can imagine your model would be hugely useful in propelling initial predictions and prompts to facilitate annotation. @Dan Morris https://labelstud.io/guide/ml_create.html

labelstud.io

Label Studio Documentation — Write your own ML backend

Set up your machine learning model to output and consume predictions in your data science and data labeling projects.

Original URL: https://labelstud.io/guide/ml_create.html

Ben Weinstein (benweinstein2010@gmail.com)

2025-05-01 18:21:44

I'm going to write this up as a discussion piece and we can move the discussion there.

👍 Niki Amini-Naieni

Niki Amini-Naieni (niki@adainnovations.ai)

2025-05-01 18:25:51

Awesome!

Ben Weinstein (benweinstein2010@gmail.com)

2025-05-01 18:41:54

https://github.com/weecology/DeepForest/discussions/978

Sara Beery (sara.beery@gmail.com)

2025-05-01 15:00:08

Who will be at CVPR? Is there interest in another bird walk? If so, anyone want to help me organize?

👍 Silvia Zuffi, murilo, Mitchell Rogers, Daniel Grzenda, Lukas Picek, Valentin Gabeff, Valerie, Lasha Otarashvili, Luke Meyers, Tuan-Anh VU, Kevin Barnard

🐦 Burak Ekim, murilo, Mitchell Rogers, Jenna Kline, Oisin Mac Aodha, Malte Pedersen, Alex Chan, Jaycee Choi, Lukas Picek, Lasha Otarashvili

murilo (murilogustineli@gatech.edu)

2025-05-01 15:10:15

@Lukas Picek and I will be there. It will be my first time attending the conference 🙂

❤️ Sara Beery

Anselm Bradford (ans@anselmbradford.com)

2025-05-01 16:19:58

I was going to be, but I'm a federal employee and all our training budget and travel is frozen, so 99.99% sure my request will just sit in limbo 😞

😭 Sara Beery

Ted Schmitt (teds@allenai.org)

2025-05-01 17:05:19

@Patrick Beukema bird walk might be of interest to you.

Sara Beery (sara.beery@gmail.com)

2025-05-01 17:35:28

In the past they have been about 5% people who know and love birds, 95% people who want to go on a walk outside and talk about research in this area :)

🦆 Mitchell Rogers

Patrick Beukema (patrickb@allenai.org)

2025-05-01 18:26:21

yes Ill be at CVPR (very excited for Earth Vision! 💙 )

❤️ Sara Beery

Jenna Kline (jennamkline@gmail.com)

2025-05-01 19:13:47

I will be at CVPR, plus a bunch of us from Ohio State/Imageomics!

❤️ Sara Beery

Devis Tuia (devis.tuia@epfl.ch)

2025-05-02 02:22:14

no, enjoy the conference!

Oisin Mac Aodha (macaodha@caltech.edu)

2025-05-02 10:22:36

Happy to help organize.

❤️ Mitchell Rogers

Ronny Hänsch (rww.haensch@gmail.com)

2025-05-05 08:08:24

I wont be but @Ragini Bal Mahesh might be interested

Ragini Bal Mahesh (ragini.mahesh@dlr.de)

2025-05-05 08:12:30

I will be at EarthVision and happy to join

❤️ Sara Beery

Jaycee Choi (jaycee.choi@jax.org)

2025-05-05 14:17:37

I will be attending and would love to go on a bird walk! 🐦

Lukas Picek (lukaspicek@gmail.com)

2025-05-06 05:48:23

I’ll be there. Hopefully not jet-lagged and will join.
@Ryan Farrell is also coming and will be for sure interested in the bird walk :)

❤️ Sara Beery

Devis Tuia (devis.tuia@epfl.ch)

2025-05-06 06:24:06

from my team, @Valentin Gabeff and @Valerie will be there. I suppose they might be interested in birding

✅ Valentin Gabeff, Valerie

❤️ Sara Beery

Luke Meyers (lmeyers@SEATTLEU.EDU)

2025-05-06 18:04:49

Yes! I was so disappointed I missed bird walk a couple years ago

❤️ Sara Beery, Valentin Gabeff

Riley Knoedler (mknoedler@west-inc.com)

2025-05-19 11:41:44

Very excited to join the bird walk again and happy to help however I can!

Sara Beery (sara.beery@gmail.com)

2025-05-23 14:23:02

Ok, I propose we schedule the CVPR bird walk for the morning of June 13th from 7-9am, before the main conference starts.

What we need:

where is a good place to go look at birds in Nashville? Ideally not too hard to get to from near the conference center. Looking for any advice!

A volunteer to plan out recommended logistics given the walk location: where should we meet, what is the best way to get there (public transit? lyft?)

A volunteer to make promo materials for the walk so we can spread the word!

@Riley Knoedler @Oisin Mac Aodha @Ryan Farrell @Lukas Picek @Luke Meyers @Jaycee Choi @Jenna Kline @Ragini Bal Mahesh @murilo can I get a volunteer to take charge of each of 1-3?

🎉 Jon Van Oast, gyeongsu, murilo, Mohamed Elhoseiny

🐦 Mitchell Rogers, murilo

murilo (murilogustineli@gatech.edu)

2025-05-23 14:51:51

I’m happy to make the promo materials for the walk :) @Sara Beery

murilo (murilogustineli@gatech.edu)

2025-05-23 14:53:11

what does that entail? A poster that can be shared on socials?

Sara Beery (sara.beery@gmail.com)

2025-05-23 14:53:30

Yes! A poster and a blurb

👍 murilo

murilo (murilogustineli@gatech.edu)

2025-05-23 15:19:03

Sounds good! I’ll have that done by Monday and will send you a draft of the poster

Sara Beery (sara.beery@gmail.com)

2025-05-23 16:43:11

🥳🥳

Oisin Mac Aodha (macaodha@caltech.edu)

2025-05-24 11:49:17

RE 1 How about we go here:
https://maps.app.goo.gl/kQS5odQRsrt4aK5P8

It's about 10 mins drive from the conference venue, and even has an observation deck.

Sara Beery (sara.beery@gmail.com)

2025-05-27 10:40:27

This looks perfect

👍 Oisin Mac Aodha

Jennifer D'Souza (jenlindadsouza@gmail.com)

2025-05-05 07:08:27

Hi everyone,
We're now inviting submissions for the EcoDL Workshop (https://sites.google.com/view/ecodl2025/). As one of the co-organizers, I’m pleased to share the call for papers below.
If you have any related questions about submissions, feel free to get in touch.

🌿 𝗖𝗮𝗹𝗹 𝗳𝗼𝗿 𝗣𝗮𝗽𝗲𝗿𝘀: 𝗘𝗰𝗼𝗗𝗟 𝟮𝟬𝟮𝟱 🌿
𝗪𝗼𝗿𝗸𝘀𝗵𝗼𝗽 𝗼𝗻 𝗗𝗶𝗴𝗶𝘁𝗮𝗹 𝗟𝗶𝗯𝗿𝗮𝗿𝗶𝗲𝘀 𝗮𝗻𝗱 𝗔𝗜-𝗯𝗮𝘀𝗲𝗱 𝗜𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻 𝗦𝘆𝘀𝘁𝗲𝗺𝘀 𝗳𝗼𝗿 𝗘𝗰𝗼𝗹𝗼𝗴𝗶𝗰𝗮𝗹 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 𝗮𝗻𝗱 𝗣𝗿𝗮𝗰𝘁𝗶𝗰𝗲 (𝗘𝗰𝗼𝗗𝗟 𝟮𝟬𝟮𝟱), co-located with TPDL 2025 in Tampere, Finland, on 𝗦𝗲𝗽𝘁𝗲𝗺𝗯𝗲𝗿 𝟮𝟯, 𝟮𝟬𝟮𝟱.

EcoDL 2025 aims to bring together researchers and practitioners at the intersection of ecology, artificial intelligence, and digital libraries. The workshop will explore innovative approaches to:
• 𝗔𝗜-𝗱𝗿𝗶𝘃𝗲𝗻 𝗸𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲 𝘀𝘆𝗻𝘁𝗵𝗲𝘀𝗶𝘀 in ecological research
• 𝗦𝗲𝗺𝗮𝗻𝘁𝗶𝗰 𝘀𝗲𝗮𝗿𝗰𝗵 and information retrieval for environmental data
• 𝗙𝗔𝗜𝗥 𝗱𝗮𝘁𝗮 𝗽𝗿𝗶𝗻𝗰𝗶𝗽𝗹𝗲𝘀 and open science infrastructures
• 𝗠𝗮𝗰𝗵𝗶𝗻𝗲 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗮𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 in biodiversity and conservation
• 𝗔𝗜-𝗽𝗼𝘄𝗲𝗿𝗲𝗱 𝗱𝗲𝗰𝗶𝘀𝗶𝗼𝗻 𝘀𝘂𝗽𝗽𝗼𝗿𝘁 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 for ecosystem management

We welcome submissions of:
📄 𝗟𝗼𝗻𝗴 𝗣𝗮𝗽𝗲𝗿𝘀 (up to 15 pages)
📝 𝗦𝗵𝗼𝗿𝘁 𝗣𝗮𝗽𝗲𝗿𝘀 (up to 10 pages)
🧾 𝗔𝗯𝘀𝘁𝗿𝗮𝗰𝘁𝘀 (up to 2 pages)

Accepted long and short papers will be published in the 𝗦𝗽𝗿𝗶𝗻𝗴𝗲𝗿 𝗖𝗼𝗺𝗺𝘂𝗻𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 𝗶𝗻 𝗖𝗼𝗺𝗽𝘂𝘁𝗲𝗿 𝗮𝗻𝗱 𝗜𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻 𝗦𝗰𝗶𝗲𝗻𝗰𝗲 (𝗖𝗖𝗜𝗦) proceedings.

📅 𝗜𝗺𝗽𝗼𝗿𝘁𝗮𝗻𝘁 𝗗𝗮𝘁𝗲𝘀:
• 𝗦𝘂𝗯𝗺𝗶𝘀𝘀𝗶𝗼𝗻 𝗗𝗲𝗮𝗱𝗹𝗶𝗻𝗲: May 16, 2025 June 12, 2025

🔗 For full details and submission guidelines, visit: https://sites.google.com/view/ecodl2025/

And kindly help us 𝘀𝗽𝗿𝗲𝗮𝗱 𝘁𝗵𝗲 𝘄𝗼𝗿𝗱 within your professional networks! 🙌

👀 Alan Stenhouse

Jon Garfunkel (jongarfunkel@gmail.com)

2025-05-05 09:09:15

Checking back in here, I'm only an amateur in the ecology field, serving on my town Conservation Board. I'd found this group some time back when I'd heard about @Sara Beery’s work in AI detection efforts. I'm also reading up on the UC Davis "Google Weed View" project.
https://caes.ucdavis.edu/news/google-weed-view-professor-trains-computer-spot-invasive-weed
Is any of this work... productized yet? I'd love to be able to propose to our town, we can pick an area (i.e., the town), pick a set of species (here in the Northeast US: Ailanthus, Japanese Knotweed, Porcelain Berry) -- and do an inventory of these. I'm aware that some consultants are offering bespoke services for such, but I think we can only afford an off-the-shelf product for this assuming it exists)

College of Agricultural and Environmental Sciences

Google Weed View? Professor Trains Computer to Spot Invasive Weed

To manage johnsongrass, a noxious weed that crowds out cotton and sickens horses, farmers have tried herbicides, burning and hand-pulling. Now, researchers at University of California, Davis, have developed a more high-tech weapon against the invasive weed: artificial intelligence and machine learning.

Original URL: https://caes.ucdavis.edu/news/google-weed-view-professor-trains-computer-spot-invasive-weed

Jon Garfunkel (jongarfunkel@gmail.com)

2025-05-05 11:38:40

and related question - consider the input image being a wide shot or panoramic or sphere -- is there a particular app that can identify or even count multiple species at once?

Fadel Seydou (fadel.seydou@gmail.com)

2025-05-07 10:28:36

📣New Dataset Release for Wildlife Detection 🐘
Hi everyone,
We’ve just published an expert-annotated upgrade to the SAVMAP UAV wildlife imagery dataset from Namibia — now optimized for ML workflows!
Annotated by @Paul Allin (PhD candidate, Stellenbosch University) and packaged by myself (Independent ML engineer), this version corrects the many false positives in the original crowd-sourced annotations.

✅ COCO-style bounding boxes
✅ Hugging Face-compatible
✅ Ideal for object detection, conservation tech, and remote sensing research

More annotated data coming soon — stay tuned!

Check it out: link
#WildlifeConservation #MachineLearning #ComputerVision #AIforGood #RemoteSensing #OpenData #UAV #HuggingFace

Zenodo

Near real-time ultrahigh-resolution imaging from unmanned aerial vehicles for sustainable land use management and biodiversity conservation in semi-arid savanna under regional and global change (SAVMAP)

To prevent aggravation of existing poverty in semi-arid savannas, a comprehensive concept for the sustainable adaptive management and use of these ecosystems under unprecedented conditions is needed. SAVMAP is an innovative, trans-, and inter-disciplinary initiative whose goal is to develop a valuable monitoring tool for both sustainable land-use management and rare species conservation (black rhinoceros) in semi-arid savanna in Namibia. SAVMAP uses near real-time ultrahigh-resolution photographic imaging (NURI) facilitated by unmanned aerial vehicles (UAVs) designed at EPFL.

Original URL: https://zenodo.org/records/1204408

huggingface.co

fadel841/savmap · Datasets at Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Original URL: https://huggingface.co/datasets/fadel841/savmap

🙌 Blair Costelloe, Timm Haucke, Sara Beery, Jenna Kline, Elizabeth Campolongo, Yseult Hb, Christoph Praschl, Alexander Merdian-Tarko, Ethan White, Rebecca Wilks, Eugene Galaxy

❤️ Sara Beery

🎉 Jenna Kline, Dan Morris, Niki Amini-Naieni, Jon Van Oast, Lasha Otarashvili

Niki Amini-Naieni (niki@adainnovations.ai)

2025-05-07 11:40:10

Why do they all have categories: 1? Do you have the name of the animal annotated in each image?

Fadel Seydou (fadel.seydou@gmail.com)

2025-05-07 11:43:10

There is only one class "wildlife". We have not identified species. Please refer to the original publication https://zenodo.org/records/1204408 for more details. Our upcoming annotated data will distinguish the different species

👍 Niki Amini-Naieni

Dan Morris (agentmorris@gmail.com)

2025-05-07 18:46:32

The most recent version on Zenodo is dated 2015... is the updated version (i.e., the reduction in false positives that you describe in this thread) only on Hugging Face, or is "version 2.0" on Zenodo identical to what's on HF?

Fadel Seydou (fadel.seydou@gmail.com)

2025-05-08 06:01:28

@Dan Morris The updated version mentioned here is only on HF. It's not identical to "version 2" in the sense that the images have been cropped (2000**2000) and only a portion of the negative samples (i.e. empty images) were kept. The dataset comprises 3545 negative samples and 379 positive samples.

Dan Morris (agentmorris@gmail.com)

2025-05-08 09:21:38

All clear, thanks.

Devis Tuia (devis.tuia@epfl.ch)

2025-05-09 08:30:23

Great, Fadel, it is always a pleasure to see SAVMAP alive and kicking since 2014, when it was started also with us in the loop!
Btw @Giacomo May will be in Kuzikus next month to acquire new data in the framework of the Wildrone project.

Fadel Seydou (fadel.seydou@gmail.com)

2025-05-10 05:56:03

Hello Prof Tuia @Devis Tuia, it's been a while.
Oooh nice! That's interesting news. We would be very interested to have a chat with him

Eugene Galaxy (eugene@animaldetect.com)

2025-07-10 03:14:25

@Hugo check it out. Animal recall rate in our drone imagery projects would be higher, no?

Jan Huus (jhuus1@gmail.com)

2025-05-09 07:58:01

New paper! We developed HawkEars, a CNN training framework and classifier for a regional community of acoustic animals - 328 bird species and 13 amphibian species in Canada. We used an embedding-based search to efficiently generate strong labels from weakly-labelled open-access training datasets. Our regional model outperformed existing global models, BirdNET, and Perch, by substantial margins, setting benchmarks for state-of-the-art classification performance. We argue regional classifiers may lead to higher model performance because a smaller species pool allows for more resources allocated per species to training and tuning and reduces the risk of class overlap, and our strong labeling method ensures high-quality training data. HawkEars will be of interest to practitioners studying acoustic wildlife in Canada and the northern United States, but the open-source code can also be used for developing classifiers for other regions.

Want more info?

Read more about HawkEars and why it’s a great tool for acoustic monitoring in the recently published paper here

Find open-access code and models for HawkEars on Github here

Watch this space for future information on integration of HawkEars into WildTrax and online tutorials for running HawkEars!

WildTrax

Pen (@nacly@floss.social)

WildTrax Manage, store, process, and share environmental sensor data.

Original URL: https://wildtrax.ca/

🙌 Justin Kay, Dan Morris, Elizabeth Campolongo, Georgia Atkinson, Sara Beery, Brett Ford, Pat Lyon, Alexander Merdian-Tarko, Mike Gallimore, Thomas Radinger

🎉 Jon Van Oast

❤️ Pat Lyon, Catherine Villeneuve, Leonardo Viotti

🙌:skin_tone_3: Alan Stenhouse

Jan Huus (jhuus1@gmail.com)

2025-05-09 08:29:07

Ah thanks Devis. Should I delete this one and re-post there?

Devis Tuia (devis.tuia@epfl.ch)

2025-05-09 08:30:50

naah, just for next time

Jan Huus (jhuus1@gmail.com)

2025-05-09 08:31:02

Sounds good - thanks again!

Sara Beery (sara.beery@gmail.com)

2025-05-09 09:29:17

We've been really impressed with HawkEars! Exciting to see the paper out!

👍 Jan Huus

Alexander Merdian-Tarko (alexander.merdian-tarko@posteo.de)

2025-05-09 10:33:42

I recently developed TeraiNet, a species classification model for the Terai region in Nepal :flag_np:, to support Bengal tiger 🐅 conservation through the use of ML in analyzing vast amounts of camera trap images.

🙏 For me this has been the first time building a species classifier. And it was a really nice experience - especially because I could rely on support and tools from this community here:

@Peter van Lunteren kindly supported me throughout the process and intergrated the model in AddaxAI

@Dan Morris and folks created MegaDetector, which I rely on for data preparation

@Zach Aandahl and @Barry Brook created MEWC, which I used as an orientation for building a species classifier

🧰 It would be great to enable others to easily built their own species classifiers for their respective use cases. So I tried to build TeraiNet as kind of a reproducible template, using open data (e. g. LILA BC), open tools (e. g. MegaDetector and MEWC) and open compute resources (e. g. Google Colab and Kaggle). Ideally, switching out the desired species, and changing a couple other things down the line, I hope that a new species classifier should be quickly built. I'll give it a try myself soon, and if anyone here is interested too or has any questions, feel free to reach out!

⁉️ There remain a couple challenges I faced during development:

How does one make the best of the images available on LILA BC when using them to create train/val/test sets?

How do I go about species in my region of interest for which there are no images openly available? Can I find "substitute" species?

How do I train an effective model with limited storage and compute resources?

:ladybug: PS: TeraiNet is under active development and I have a couple things still on my to do list. So feel free to contribute, file issues, report bugs, give feedback and suggest improvements.

🙏 Lily Xu, Avishek Dangol

🎉 Hammed Akande, Dan Morris, Carly Batist, Jon Van Oast, Cara Appel, Shir Bar, Sara Beery, Elizabeth Campolongo, Alan Stenhouse

🙌 Owen Martin, Thomas Radinger

❤️ Peter van Lunteren

Avishek Dangol (dangolavishek@gmail.com)

2025-05-13 05:30:17

As a Nepalese citizen, I appreciate the effort. I am currently studying AI, so a bit of beginner in the field. But, would love to contribute to the project in any way possible!!

Alexander Merdian-Tarko (alexander.merdian-tarko@posteo.de)

2025-05-13 14:53:45

That's great @Avishek Dangol! I'll send you a dm.

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2025-05-12 03:15:22

FYI I'm organising an online event with WILDLABS on how to 💰 sustainably fund open source technologies at 17:00 UTC (18:00 BST/19:00 CEST) this Thursday, 15 May 2025:
https://wildlabs.net/event/sustainable-financing-open-source-conservation-tech-open-source-solutions-funding-and-finance
I think the overall discussion about sustainable funding may be of interest, so please sign up and join us! ❤️

Some posts to share:
https://floss.social/@nacly/114460108796260675
https://bsky.app/profile/wildlabs.bsky.social/post/3logn42vuh22k
https://www.linkedin.com/feed/update/urn:li:share:7325176998979362816/

FLOSS.social

Attached: 1 image Wondering how to make #opensource tech financially sustainable? I'm co-organising an online #WILDLABS community meeting on 15 May 2025 to discuss the challenges and opportunities for #funding open source #conservation tech. Peter van Lunteren of <a href="http://AddaxDataScience.com">AddaxDataScience.com</a>, Susan Allen of <a href="http://FieldKit.org">FieldKit.org</a>, and Alasdair Davies of <a href="http://Arribada.org">Arribada.org</a> will discuss sustainable pathways to fund open source work, with Q&A and discussion. Register here: <https://wildlabs.net/event/sustainable-financing-open-source-conservation-tech-open-source-solutions-funding-and-finance>

Original URL: https://floss.social/@nacly/114460108796260675

bluesky

WILDLABS Community (https://bsky.app/profile/wildlabs.bsky.social)

Curious how to make open source tech financially sustainable? Join the WILDLABS Open Source Solutions and Funding & Finance community meeting on May 15 at 5pm UTC to discuss financial challenges + opportunities, featuring talks from <a href="https://bsky.app/profile/fieldkit.bsky.social">@fieldkit.bsky.social</a> and Addax AI! <https://wildlabs.net/event/sustainable-financing-open-source-conservation-tech>

Original URL: https://bsky.app/profile/wildlabs.bsky.social/post/3logn42vuh22k

linkedin.com

Sustainable financing for open source conservation tech - Open Source… | WILDLABS Community

Wondering how to make #opensource tech financially sustainable? Join the WILDLABS Open Source Solutions and Funding & Finance community meeting on 15 May to discuss the challenges and opportunities for financing open source #conservationtech. Guest speakers Peter van Lunteren of Addax Data Science, Susan Allen of FieldKit, and Alasdair Davies of Arribada Initiative will discuss their experiences and perspectives on sustainable financial pathways to support the open source work, followed by community Q&A and discussion. Learn more and register here: <https://lnkd.in/gPksVmnx> Thank you to our Funding & Finance and Open Source Solutions Group Leaders for organising and hosting this event: Frank van der Most, PhD, Pen-Yuan Hsing, Aurélie Pollak, Brianna Johns, Henry Rees, and Amanda Matthes.

Original URL: https://www.linkedin.com/feed/update/urn:li:share:7325176998979362816/

❤️ Sara Beery, Elizabeth Campolongo, Don Cosseboom, Vanesa Reyes, Talia Speaker, Alan Stenhouse

🙏 Lily Xu

Sara Beery (sara.beery@gmail.com)

2025-05-12 15:55:45

The call for applications to CV4Ecology 2026 is out!!

When: Jan 12-30, 2026
Where: Smithsonian Conservation Biology Institute
Apply by June 6th! Info here: https://cv4ecology.caltech.edu/call_for_applications.html

Help spread the word :)

https://www.linkedin.com/posts/sara-beery-1ba89166_cv4ecology2026-activity-7327773006[…]m=member_desktop&rcm=ACoAAA4DIN4BIW_3ubfcfe3GqvBT-NI--ZBxMzQ

https://bsky.app/profile/sarameghanbeery.bsky.social/post/3loyo7lnna22l

https://x.com/sarameghanbeery/status/1922016981152043171

linkedin.com

#cv4ecology2026 | Sara Beery

The call for applications has just been released for #CV4Ecology2026!! This three-week intensive program trains ecologists and conservation practitioners to develop their own AI tools for their own data. When: Jan 12-30, 2026 Where: Smithsonian Conservation Biology Institute Read more and apply (by June 6th) at <https://lnkd.in/eE9T5vAB>

Original URL: https://www.linkedin.com/posts/sara-beery-1ba89166_cv4ecology2026-activity-7327773006926159873-jT4K?utm_source=share&utm_medium=member_desktop&rcm=ACoAAA4DIN4BIW_3ubfcfe3GqvBT-NI--ZBxMzQ

bluesky

Sara Beery (https://bsky.app/profile/sarameghanbeery.bsky.social)

The call for applications has just been released for #CV4Ecology2026!! This three-week intensive program trains ecologists and conservation practitioners to develop their own AI tools for their own data. When: Jan 12-30, 2026 Where: SCBI <a href="https://bsky.app/profile/smconservation.bsky.social">@smconservation.bsky.social</a>

Original URL: https://bsky.app/profile/sarameghanbeery.bsky.social/post/3loyo7lnna22l

X (formerly Twitter)

Impact of transfer learning methods and dataset characteristics on generalization in birdsong classification

Very excited to announce the call for applications for the next @cv4ecology workshop!! Apply by June 6th!

Original URL: https://x.com/sarameghanbeery/status/1922016981152043171

🙌 Justin Kay, Jason Holmberg (Wild Me), Shir Bar, Nico Lang, Matthias Zuerl, Robin Zbinden, Alexander Merdian-Tarko, Puneet Kollipara, Lily Xu, Peter van Lunteren, Maria Jose Guerrero, Andrew Schulz, Talia Speaker

🎉 Jon Van Oast, Dan Morris, Mitchell Rogers, Jason Holmberg (Wild Me), Taiki Sakai - NOAA Affiliate, Lily Xu

❤️ murilo, Jonathan Sauder, Wenxin Yang, Jess Tam, Jason Holmberg (Wild Me), Cara Appel, Maria Napolitani, Yseult Hb, Holly Houliston, Alan Stenhouse

🐢 Burak Ekim

Eric Orenstein (eorenstein@mbari.org)

2025-06-12 15:18:28

The deadline for the 2026 CV4Ecology workshop has been extended until June 21! Spread the word! Tell your friends! Get those applications over the finish line!

https://cv4ecology.caltech.edu/call_for_applications.html

🎉 Jon Van Oast, Sara Beery, Holly Houliston, Maria Napolitani

Burooj Ghani (buroojghani@gmail.com)

2025-05-13 08:05:35

🐦 New paper alert! 🐦

Our latest study explores how transfer learning boosts bird sound recognition across challenges. We compare CNNs vs Transformers, fine-tuning vs distillation, training w/ and w/o background labels in crowdsourced animal sound datasets.

Key takeaway? Label your background species too - it matters.

https://www.nature.com/articles/s41598-025-00996-2.epdf?sharing_token=jhgH6vxm9O2kFZ[…]GMvMAXkSSea8IlfLShA7BoTcxj3AMsE1scnIx67j7JvJhyQBWbjjyKfugo%3D

nature.com

Scientific Reports - Impact of transfer learning methods and dataset characteristics on generalization in birdsong classification

Original URL: https://www.nature.com/articles/s41598-025-00996-2.epdf?sharing_token=jhgH6vxm9O2kFZfhOeorXtRgN0jAjWel9jnR3ZoTv0M371N5wrdf7CLPW_phsBsNHCAhs5vhDG9P7cPEH2PCuBTWjtNnQ2na9Ud_wdIRX4IVig4rVGMvMAXkSSea8IlfLShA7BoTcxj3AMsE1scnIx67j7JvJhyQBWbjjyKfugo%3D

👏 Viktor Domazetoski, Brian Geuther, Juan Sebastián Cañas, Maddie Cusimano, Abigail Oppong

🙌 Jan Huus, Dante Wasmuht, Elizabeth Campolongo, Benjamin Hoffman, Owen Martin, Alexander Merdian-Tarko, Anthony Miyaguchi, Maria Jose Guerrero

🙌:skin_tone_2: Cara Appel

🙌:skin_tone_3: Alan Stenhouse

Brian Geuther (Brian.Geuther@gmail.com)

2025-05-13 08:50:47

I think Fig 7 is great. I think that the central idea of "data is king" is largely forgotten in our current era of benchmarks. While it's great to have models that can do more with less... particularly in fields that don't yet have an appropriate benchmark, investing more in annotations will often outperform spending more time tinkering with models/training routines. Unfortunately, it's not a fun conclusion for the people that would prefer working on that tinkering.

👍 Elizabeth Campolongo

Burooj Ghani (buroojghani@gmail.com)

2025-05-13 08:54:28

Right, also, if a certain vocalisation is not present in the training set, no amount of "tinkering" can enable the model to recognise it. Thanks for your applaud.

Somil (somilshah1000@gmail.com)

2025-05-14 19:31:51

Hi everyone!

Sharing something from our lab that might be of interest here: the Moodeng AI Challenge, hosted by the MIT Media Lab and the Board of the Zoological Park Organization of Thailand 🦛

Taking place this July in Thailand, the challenge is inspired by ✨Moodeng✨ the pygmy hippo and explores AI-powered human-animal interaction, conservation, and behavior analysis.

It’s a great fit for folks working across AI, conservation, animal behavior, HCI, and welfare tech.

Dr. Rebecca Kleinberger from the Interact Animal Lab at Northeastern University will be serving as one of the judges.

🔗 moodeng.media.mit.edu
You can follow our lab for updates: Instagram | LinkedIn | X
📣 LinkedIn post

linkedin.com

#moodeng #mitmedialab #hci #aci #thailandzoo #submissionsopen | INTERACT Animal Lab

🚀 Exciting News! We’re thrilled to share that Dr. Rébecca Kleinberger will be serving as a judge at the MoodEng AI Challenge, organised by the MIT Media Lab! 🎉 Organized by Dr. Pat Pataranutaporn, along with Patrick Chwalek, Dr. Napat Jatusripitak, Harpreet Sareen, and Dr. Jade Donavanik, this challenge takes place this July in Thailand, in collaboration with the Board of the Zoological Park Organization of Thailand. Inspired by Moodeng, the pygmy hippopotamus, the challenge brings together AI researchers, conservationists, and technologists to explore how artificial intelligence can enhance human-animal interaction, behavioral understanding, and wildlife care. With tracks focusing on decoding animal communication, fostering empathy, and augmenting zookeeper capabilities, this event is a key moment in Affective Computing, AI Ethics, and Animal-Computer Interaction. 📩 Submissions are now open! 🔗 Learn more & get involved: <a href="http://moodeng.media.mit.edu">moodeng.media.mit.edu</a> #Moodeng #MITMediaLab #HCI #ACI #ThailandZoo #SubmissionsOpen

Original URL: https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.linkedin.com%2Fposts%2Finteract-animal-lab_moodeng-mitmedialab-hci-activity-7326736870422319104-chZj&data=05%7C02%7Cshah.som%40northeastern.edu%7C45a53e0dcfc543ce74f808dd92d6ff59%7Ca8eec281aaa34daeac9b9a398b9215e7%7C0%7C0%7C638828174318464653%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=LoKhtAhBRuh2N1anl%2BX0llPNJ%2Bd6ltyhDew1SV%2FrS4M%3D&reserved=0

💚 Lily Xu, Tim Zhou, Klára Janoušková, Rebecca Kleinberger, PhD

🦛 Mitchell Rogers, Rebecca Kleinberger, PhD, Alan Stenhouse

Youssef Bayouli (youssef.projetnemo@gmail.com)

2025-05-15 04:23:18

Hi everyone 👋
I’m currently (and a bit desperately!) looking for a summer internship or end-of-studies internship (Starting from February 2026) in the areas of AI, robotics, or tech applied to wildlife conservation. I’m pursuing an MSc in Advanced Engineering for Robotic Systems and AI.
Alongside my studies, Few years ago , I completed an externship with National Geographic and The Nature Conservancy, focusing on marine life conservation, which deepened my drive to work at the intersection of tech and environmental impact.
If you know of any opportunities or anyone I should talk to I'd be incredibly grateful.
Thanks so much for your time and support!

Vanesa Reyes (vanesa.reyes@wildlabs.net)

2025-05-15 09:58:58

Hi all! If you are planning to attend the Living Data Conference, either in person or virtually, we are organizing a session on bioacoustics, including AI related analysis. Calls for abstract is open until May 20. More info here: https://wildlabs.net/event/submit-living-data-abstracts-20-may-listen-future-mobilizing-bioacoustic-data-meet

❤️ Jon Van Oast, Mélisande Teng, Sara Beery, Elena Sierra, Filippo Varini, Alan Stenhouse

🙌 Laura Chrobak, Elena Sierra

Filippo Varini (fppvrn@gmail.com)

2025-05-17 04:01:10

Good morning everyone!
Does anyone know studies that successfully applied the Megadetector (and any downstream model) to drive policy/conservation decisions in South America?

Dan Morris (agentmorris@gmail.com)

2025-05-17 11:33:25

I'll be a little pessimistic about finding that direct a link, for three reasons: (1) policy documents likely have dozens or hundreds of inputs, so it becomes very hard to say that this one thing that helped with this one step in this one project did or didn't have a direct impact on policy (this one is not specific to MD, this is true for any AI tool used for data review), (2) policy happens on the scale of years to decades, and it's only around 2023 that we started seeing MD cited in papers that are about environmental monitoring, as opposed to being about machine learning, and (3) in a good way, MD has now been diffused into enough other systems that most people using MD don't even know that they're using MD, they just know they're using AI, and in those cases the degree to which they even use AI is hard to measure. E.g. this would be true of everyone using Wildlife Insights: they are offered results that partially come from MD, but there's no reason for them to know or care that MD was part of the AI, and the degree to which they use the AI results varies in ways that are hard to track (just getting AI results doesn't mean you used AI results).

So, no one is more enthusiastic about AI for conservation than I am 🙂, but the lines to impact are long and blurry (with the notable exception IMO of anti-poaching and invasive species management applications, where the line to impact can be quite direct and on the scale of hours).

That said, I do track papers (including any public-sector reports I can find) that (a) used AI to process camera trap data and (b) were actually about environmental monitoring, not just about AI. If you look at the papers here:

https://agentmorris.github.io/camera-trap-ml-survey/#papers-with-summaries

...that are tagged "ecology paper" and "MegaDetector", you'll find some super-direct examples of MD being used for science, with no ambiguity about whether MD was actually used. And some day, much of that science will be cited in a policy document somewhere, but it's a long and winding road to get there!

🦉 Alan Stenhouse

Dan Morris (agentmorris@gmail.com)

2025-05-17 11:39:26

And BTW, I would love to be wrong about everything I just said, and I hope someone shows me the direct link to policy that's missing from my list of papers. 🙂 But not having that nice clean link doesn't mean we're not doing our job as AI folks, it just means that the path from science to policy is complicated.

👍:skin_tone_3: Alan Stenhouse

Filippo Varini (fppvrn@gmail.com)

2025-05-19 13:07:22

Thanks Dan for your reply! And thanks so much for having a public list of papers with summaries, that is very helpful!

stefano puliti (stefano.puliti@nibio.no)

2025-05-19 06:25:02

📢 open postdoc opportunity in Norway :flag_no: https://www.linkedin.com/posts/stefano-puliti-854b7424_forestai-aifornature-computervision-activity-7330167792626806784-B35Y?utm_source=share&utm_medium=member_ios&rcm=ACoAAAUn2sABvkN71_GkuJHYSyqwV11nnLUTrBU

linkedin.com

#forestai #aifornature #computervision #panopticsegmentation #deeplearning… | Stefano Puliti

🚨 NEW POSTDOC ALERT: Are you a computer vision ninja with a soft spot for forest and the environment? 🌲🧠🤖 Ready to push the boundaries of forest computer vision? At SmartForest4.0, hosted by NIBIO Norwegian Institute of Bioeconomy Research, we’re on the hunt for a Postdoc in Computer Vision and AI to join our 🚀 #ForestAI team! 🌲 We are building the next-generation of forest vision models – not just tree ID, but full-on ecosystem understanding. We&#39;ve poured serious effort into creating a unique, high-quality image and video dataset from forest across Europe (yes, we’re talking one-of-a-kind stuff) and now we need you to turn them into a powerful vision suite to enable a holistic understanding of forest ecosystems 🍃 What you bring: ✅ PhD in Computer Science, Machine Learning, or a related field ✅ Solid experience in image/video analysis ✅ Strong publication record & excellent English skills ✅ Bonus: you love forests, robots, and table tennis What we offer: 💻 Access to state-of-the-art compute (hello A100s 🔥) 🌲 A one-of-its-kind dataset across 🤝 A young, and fun team working at the AI–robotics–forest–ecology intersection 🏓 Epic ping pong breaks 📍 Full-time, 3-year position based in Ås, Norway 📆 Apply by June 8, 2025 👉 Apply here: <https://lnkd.in/dHTv_X-K> #AIforNature #ComputerVision #PanopticSegmentation #DeepLearning #ForestTech #SmartForests #MachineLearning

Original URL: https://www.linkedin.com/posts/stefano-puliti-854b7424_forestai-aifornature-computervision-activity-7330167792626806784-B35Y?utm_source=share&utm_medium=member_ios&rcm=ACoAAAUn2sABvkN71_GkuJHYSyqwV11nnLUTrBU

🌳 Sara Beery, Nico Lang, Jason Holmberg (Wild Me), Binbin Xiang, Nora Gourmelon, Malte Pedersen, Clemens Mosig, Yuru Jia, Alan Stenhouse

👍 Oisin Mac Aodha, Jason Holmberg (Wild Me), Binbin Xiang

Lukas Picek (lukaspicek@gmail.com)

2025-05-20 06:35:02

Hello there👋,
Anyone knows about or is involved in any ICCV workshops focused on ecology or conservation?
Please tell us! 🙂

Malte Pedersen (mape@create.aau.dk)

2025-05-20 07:01:50

David Nakath and I are organizing a workshop on underwater imagery and marine environmental monitoring/conservation. (And I put your name on the tentative program committee list 5 minutes ago, so you may receive an email from us in the near future 😂)

😅 Lukas Picek

👍 Lukas Picek, Justin Kay, Bowen Zhang

Devis Tuia (devis.tuia@epfl.ch)

2025-05-20 07:28:46

I heard of a cv4ecology being organised, but I don’t have any evidence of names to provide

🤞 Lukas Picek

👍 Lukas Picek, Justin Kay, Sara Beery

Elena Sierra (esierra@stanford.edu)

2025-05-20 13:29:42

https://sustain-eo-ai.github.io/
“Sustainability with Earth Observation & AI”
More broadly sustainability but lists biodiversity as a topic

🤩 Devis Tuia, Bowen Zhang, Robin Zbinden

👍 Bowen Zhang

Sara Beery (sara.beery@gmail.com)

2025-05-20 14:43:20

Yes, theres a CV4Ecology workshop again! Mostly same folks as last time (me, @Eddie Vendrow, @Julia Chae, @Mohamed Elhoseiny + team)

🙌 Elena Sierra, Mohamed Elhoseiny, Bowen Zhang, Dan Morris

👍 Bowen Zhang

Malte Pedersen (mape@create.aau.dk)

2025-05-23 04:12:01

Here is the call for papers for the ICCV 2025 Joint Workshop on Marine Vision

👍 Lukas Picek

🌊 Lukas Picek

Devis Tuia (devis.tuia@epfl.ch)

2025-05-23 07:31:54

thanks for setting the deadline after the main conference decisions 🙏 !

🙏 Malte Pedersen

Malte Pedersen (mape@create.aau.dk)

2025-05-24 13:31:05

@Devis Tuia luckily the 'non-negotiable' deadline for decisions was extended by two weeks. Otherwise it would not have been possible!

👍 Devis Tuia, Loh Hong Sen

😆 Loh Hong Sen

Lily Xu (lily.x@columbia.edu)

2025-05-22 19:24:34

Applications open for an upcoming workshop on "Shaping the Future of AI for Conservation" at Oxford, Sept 22-26, 2025!

Funded by Schmidt Sciences, with need-based travel support.

https://app.onlinesurveys.jisc.ac.uk/s/oxford/shaping-the-future-of-ai-for-conservation-application-form

🎉 Jon Van Oast, Timm Haucke, Dan Morris, Elena Sierra, Oisin Mac Aodha, Malte Pedersen, Ghjulia Sialelli, Carlos Rodríguez - Pardo, Holly Houliston, Carly Batist, Robin Zbinden, Justin Kay, Anna Vallery, Sara Beery, Jason Holmberg (Wild Me), Andrew Schulz, Juan Sebastián Cañas, Abigail Oppong, Alan Stenhouse

👀 Carlos Rodríguez - Pardo, Jason Holmberg (Wild Me)

🙌 Carlos Rodríguez - Pardo, Jason Holmberg (Wild Me)

Andrew Lensen (andrew.lensen@vuw.ac.nz)

2025-05-26 01:44:18

🦜PhD Opportunity!! 🦜

Due to unforeseen circumstances, we are re-advertising for a computer vision PhD position to re-identify kākā in beautiful Wellington, New Zealand. Original advert below --- ignore the closing date, and feel free to just email me directly at andrew.lensen@vuw.ac.nz 🙂

https://www.wgtn.ac.nz/scholarships/current/phd-scholarship-recognising-taonga-with-ai-facial-recognition-for-kaka-conservation-management

🐦 Mitchell Rogers, Mamoon, Yseult Hb, Oisin Mac Aodha, Amee Assad, Brian Geuther, Don Cosseboom

:flag_nz: Mitchell Rogers, Andrew Lensen, Alexander Merdian-Tarko, Jess Tam, Brian Geuther, Don Cosseboom, Alan Stenhouse

🤩 Mitchell Rogers, Don Cosseboom, Paridhi Jain

🦜 Mitchell Rogers, Andrew Lensen, Yseult Hb, Shir Bar, Jaycee Choi, Brian Geuther, Anna Vallery, Don Cosseboom, Seven Song

Andrew Lensen (andrew.lensen@vuw.ac.nz)

2025-05-26 02:04:33

Thread Reply:

📸 Mitchell Rogers

❤️ Viktor Domazetoski, Elena Sierra, Shir Bar, Mark Goldwater, Jaycee Choi, Wenxin Yang, Jess Tam, Loh Hong Sen

😍 Anna Vallery

😀 Alan Stenhouse

Lily Xu (lily.x@columbia.edu)

2025-05-28 11:52:26

Hi! @Dave Thau, @Sara Beery, and I are organizing a webinar series on AI for Biodiversity for the United Nations International Telecommunication Union (ITU) AI for Good, with talks launching in June and continuing through October.

Tomorrow afternoon, I'll be hosting Sara for a pre-recorded session on "Multimodal AI and biodiversity". Since we'll be pre-recorded, I'd love any questions from you all, or topics you'd like to see discussed! Feel free to drop a comment or DM 🙂

🙌 Oisin Mac Aodha, Justin Kay, Mark Goldwater, Irina Tolkova, Nico Lang, Anna Vallery, Jason Holmberg (Wild Me), Shir Bar, Laura Chrobak, murilo, Malte Pedersen, Navodita Mathur, Carlos Rodríguez - Pardo, Aleksis Pirinen, Yseult Hb, Thijs van der Plas, Jaycee Choi, Robin Zbinden, Eric Greenlee, Dylan Van Bramer (she/her), Viktor Domazetoski, Johannes Dollinger, Laura Madrid

🎉 Jon Van Oast, Jason Holmberg (Wild Me), murilo, Carlos Rodríguez - Pardo

😎 Jon Van Oast, Jason Holmberg (Wild Me), Carlos Rodríguez - Pardo, Alexander Merdian-Tarko

😍 Bistra Dilkina, Carlos Rodríguez - Pardo, Jennifer

✅ Carlos Rodríguez - Pardo

🙌:skin_tone_3: Alan Stenhouse

Yseult Hb (yseulthb@gmail.com)

2025-06-26 14:02:32

Really sorry I missed that one 😔
Since it was pre-recorded, is there any chance it will be available for being watched later?

Lily Xu (lily.x@columbia.edu)

2025-06-30 09:55:45

Yes @Yseult Hb, the series is available here! https://aiforgood.itu.int/eventcat/discovery-ai-for-biodiversity/

More talks to be released (including live talks) throughout the summer and fall

🙌 Yseult Hb

🙏 Yseult Hb

Sara Beery (sara.beery@gmail.com)

2025-05-28 16:16:23

https://christopherren.substack.com/p/geospatial-foundational-disappointments

christopherren.substack.com

Geospatial Foundational Disappointments

TLDR; After 10²¹ FLOPs and 500 B patches, IBM’s TerraMind beats a supervised U‑Net by just +2 mIoU on PANGAEA; losing on 5/9 tasks, most other GFMs do worse.

Original URL: https://christopherren.substack.com/p/geospatial-foundational-disappointments

😟 Mitchell Rogers

👍 Aleksis Pirinen, Wenxin Yang, Emilio Luz-Ricca

😮 Chris Lange, Emilio Luz-Ricca

Christopher Ren (christopher.x.ren@gmail.com)

2025-05-29 14:20:10

Thanks for reposting this Sara! And thanks for letting me know this slack existed @Molly Blank 🙂. If anyone wants to chat about the implications of this, or more parsimonious and frankly convenient ways to use earth observation data please do reach out!

💖 Declan

🙌 Molly Blank, Aleksis Pirinen, Wenxin Yang

Dan Morris (agentmorris@gmail.com)

2025-05-28 19:06:25

tl;dr: if you are (like me) just starting on your AI-assisted/agentic coding journey, and you (like me) would benefit from social encouragement to try new things, I'm going to basically spend the PDT morning (8amPT-1pmPT) on Tuesday (6/3) doing random AI-code-skill-building things, and it would be great to see that totally arbitrary time become a "vibe coding party". If folks want to join me, we can use AI tools to work on each other's conservation-related repos and/or we can just make silly games, whatever forces us to experiment with new tools, and to show each other different AI-assisted workflows. DM or email (agentmorris@gmail.com) if you're interested and I'll send you a meeting link.

Now the long version...

I'm curious where everyone in our community is at in your journeys toward agentic coding... this is something we haven't talked a lot (at all?) about on this Slack, so I have no idea whether I'm just way behind the curve (at least for someone who works at an AI company). But in the last couple of weeks, I've finally "gotten it" (better late than never?). So, I went a little bit bananas last week and tried every tool that exists, and I made a super-long list (very much still a WIP) about everything I learned and tried (I know, shocking turn, Dan made an obscenely long list!):

http://lila.science/ai-coding

I've been experimenting with a mix of total vibe coding (usually using the prompt "make me a game in which wombats do battle with jackals") and real work (e.g. long-deferred maintenance tasks on the MegaDetector repo), and in short, my world has been completely turned upside down. Five minutes with the free trial of every web-based vibe coding platform I tried was enough to make a game that wouldn't be way out of place on a modern app store (with zero code). And both online and local agentic tools have totally changed the way I think about building real software. I still don't really "get" IDE-integrated tools like Cursor/Windsurf/Cline/Copilot, and I'd really like to see a power user show me how you use IDE integration for anything more than autocomplete, but I'm not skeptical, it's just that other form factors (Web-based agents and CLI tools) have just clicked better for me in terms of deeper engagement with AI and more agentic workflows. But everyone seems to love Cursor, so I want to understand it!

If I've come to any pithy conclusion from this deep dive, it's that we should all be setting aside time to build this new skill, i.e. I'm not sure that just using AI for autocomplete or asking for code snippets in a chat interface (which has also been amazing!) puts one on a path to keeping up with how much coding is changing, and that's even before we get to deeper questions about what this means for how we build ML models in 2026. In fact I think AI-based autocompletion is a local maximum that may actually hurt more than it helps in terms of both getting code written and thinking of new problems for the new world.

Maybe this is all old news to everyone, and I'm... let's say... not the youngest person on this Slack, so it's very plausible that I'm stuck in my circa-1998 concept of coding, but if you're also just starting this journey, and you're not already setting aside time for this, consider joining me for a vibe coding party. And vibe coding party aside, I'd love to hear about how folks are using agentic coding tools, which I loosely define as using AI for anything beyond autocomplete and code snippets.

Sidebar, this new podcast is a very hands-on discussion of exactly these things (basically asking power AI coders exactly what they do):

https://podcasts.apple.com/us/podcast/how-i-ai/id1809663079

I thought this episode was particularly enlightening:

https://podcasts.apple.com/us/podcast/vibe-coding-a-3d-multiplayer-game-in-15-minutes-with/id1809663079?i=1000706343916

...although this remains my favorite AI podcast, because it's hella funny, but it's not really about code:

https://www.aiforhumans.show/

🙌 Sara Beery, Chris Cosma, Mitchell Rogers, Mamoon, Malte Pedersen, Jonathan Roberts, Youssef Bayouli, Carly Batist, Holly Houliston, Lily Xu, Tessa Rhinehart, Alexander Merdian-Tarko, Filippo Varini, Tiziana Gelmi Candusso, Paridhi Jain, Eugene Galaxy

🙌:skin_tone_5: Prabath Gunawardane

🙌:skin_tone_3: Alan Stenhouse

🙌:skin_tone_4: Oc

Jon Van Oast (jon@wildme.org)

2025-05-28 20:04:52

i am currently abstaining from even forming an opinion on the topic (as i am stretched way too thin), but i can add to this that a friend has created https://vibes.diy/ (as an offshoot of their project https://use-fireproof.com/). again: no real time to dig into this very much, so i offer it as simply another option to consider. no endorsement. no qualms if ignored. 🙂

vibes.diy

Vibes DIY

Vibe coding made easy

Original URL: https://vibes.diy/

use-fireproof.com

The embedded database for AI Code Generation | Vibe coding just got easier

Generate complete, single-file applications with built-in data persistence and real-time sync.

Original URL: https://use-fireproof.com/

👍 Dan Morris

Dan Morris (agentmorris@gmail.com)

2025-05-28 20:24:36

Oooh, I hadn't tried that one. That seems to have a different model, where it's all-the-way free using whatever models happen to be free today via OpenRouter, and you fork the whole repo if you want to bring your own API keys, and run your own copy of the whole site locally. Fascinating. Off to add that to the list!

BTW it did just fine with my wombats-vs-jackals prompt.

🎉 Jon Van Oast

Dan Morris (agentmorris@gmail.com)

2025-05-28 20:26:05

OMG I went back to my other tab and it had also generated images, I totally misunderstood that this is what the progress bars were doing. My mind is blown in new ways every day.

🤩 Mitchell Rogers, Alan Stenhouse

Jon Van Oast (jon@wildme.org)

2025-05-28 20:26:47

😮

Jon Van Oast (jon@wildme.org)

2025-05-28 20:28:09

glad you are having fun with it. the folks working on it are great/smart friends. i have mostly been avoiding it due to time constraints, but have heard good things (as these vibe things go?) ... they have a discord and decent active community if this particular option seems worth pursuing.

👍 Dan Morris

Dan Morris (agentmorris@gmail.com)

2025-05-28 20:33:13

This is all from the initial prompt: "make a game in which wombats do battle with jackals". Note lots of subtle wombat references in the character names.

😲 Mitchell Rogers

😂 Holly Houliston, Alan Stenhouse

Paul (morescopm@gmail.com)

2025-05-29 12:12:18

Incredible. Count me in.

Ben Weinstein (benweinstein2010@gmail.com)

2025-05-29 14:24:04

@Dan Morris in terms of cursor I find one of the game changers in generating tests and documentation. The speed and breadth of it is amazing and has the added benefit of interpreting the code from a somewhat novice user perspective, I often find it highlights how workflows can be clearer by looking at the tests it generates and seeing if they look contrived at all. Updating documentation is really impressive, especially alongside pull request.

👍 Dan Morris

👍:skin_tone_3: Alan Stenhouse

Molly Blank (mblank@naturalstate.org)

2025-05-29 14:34:33

Cursor has actually been a more finicky recently (bugs with the interface, overzealous iteration loops in the chat) with their agent updates, although it came with command line integration for things like package installation fixes. I would consider myself a syntax-challenged coder so these tools have rocked my world, too. I am only now trying to adapt more sophisticated codebases to more nuanced questions but have had great success with spinning up data science and basic analysis pipeline prototypes.

Carl Boettiger (cboettig@berkeley.edu)

2025-05-29 14:41:38

Also curious about this, especially as it pertains to education. I used jupyter-ai integration in jupyterhub for my undergrad course last fall and continue.dev vscode plugin in a smaller grad class this spring. While I've found these tools useful in my own work, getting the students to adopt these integrations has been more of a struggle (though the undergrads are happy to paste questions into chatgpt, despite that fact that it does far worse without the appropriate contextual information the IDE-integrated tools provide). But this is possibly my fault for not teaching / demo-ing the use of the integrated tools.

Carly Batist (carly.h.batist@gmail.com)

2025-05-30 08:55:46

I've been playing around with Google's Firebase but haven't fully tried Cursor/Windsurf yet. Curious how Windsurf will change/develop now that it's been acquired by OpenAI. I'd love to join the session!

Prabath Gunawardane (prabathg@gmail.com)

2025-06-01 12:52:08

I'd love to join on Tuesday! 🙂

I've been using Cursor with a paid subscription for about 6 months , mostly to work on our Wildepod codebase (medium sized repo, ~ 20k lines). It has been great for certain tasks : writing test / debug scripts, doing fuzzy searches by functionality, high level questions about code flow and of course for writing small snippets of code. However it has missed the mark in implementing more complex features (which happen to touch several files) often generating code that would seem to work but actually is quite buggy. Just yesterday I was trying to get it to help with containerizing our dev environment and it kept missing cruicial dependencies and reverting back to the same mistakes that I corrected a couple prompts previously. Definitely have reached a local maximum there (at least in the way I use Cursor), and also given that it's been unstable lately, I'm wondering if it's worth keeping the subscription.

Anyways aside from that I've also been wondering about all the new tools that keep coming out every month and would love to see how people are using them. I have a couple of ideas for wildlife conservation related mini-games and this may be a perfect time to try and experiment with that! 😄

Dan Morris (agentmorris@gmail.com)

2025-06-01 16:06:08

@Prabath Gunawardane, invite sent.

And yes, I think when I started this thread I thought Cursor was a local maximum because it doesn't encourage us to think in more agentic ways, now I think it's much simpler than that: Cursor is a local maximum because it's $20/month, and that just doesn't buy you enough context to have high-level discussions with the model. You are technically using the latest and greatest models, but developers who are really seeing what agentic AI looks like are routinely spending $2000/month. I don't have a business degree, but... $2000 is a lot more than $20, so Cursor has to make a tradeoff somewhere to offer you a flat fee of $20/month, and at the end of the day the tradeoff is context.

Of course, $2000/month/developer is a great investment organizationally in many cases, but it's not sustainable for, e.g., hobby projects or unfunded open-source work. But once you have a discussion with an AI model that can remember all the important things about your entire code base, including which files it should open for the details it forgot, and everything you talked about over a couple of days, wow, it's a totally different thing, and it at least encourages you to use AI in different ways even when you have to go back to smaller context. IMO it's such a different thing that if you can, consider spending - even just once - $20 on a few hours of interaction with a system where you're using the full context of the model, e.g. using Aider (a CLI tool where you bring your own API keys), Cline (a Cursor-like VS Code extension where you bring your own API keys), or Claude Code (Anthropic's CLI for Claude models). Then work backwards from there to figure out the right way to get the most out of AI for a specific budget.

Personally, I'm excited to try Aider on Tuesday! I'll email everyone in advance to suggest that everyone define some goals before we meet, and one of mine will be trying some more "long-context-available" tools, beyond the ones I've tried already.

👍 Molly Blank

🎉 Jon Van Oast

👍:skin_tone_3: Alan Stenhouse

Prabath Gunawardane (prabathg@gmail.com)

2025-06-01 17:32:25

Thanks Dan! And good insight on the Cursor business model & the tradeoff on context lengths.
I tried Aider a bit just before I started using Cursor and thought it might get prohibitively expensive to use daily. However as you said, giving the model full context will probably lead to vastly improved interactions... I'm looking forward to try that / other new tools on Tuesday!

Prabath Gunawardane (prabathg@gmail.com)

2025-06-03 16:08:17

Thanks everyone! That was a lot of fun 😄. I added a screenshot of the elephant game and the initial prompt to Dan's Google doc.

✅ Paul

Aleksis Pirinen (aleksis.pirinen@ri.se)

2025-05-29 06:59:39

🚨 PhD student position at RISE, Lund, Sweden! 🚨

Are you passionate about machine learning and eager to apply it to environmentally motivated challenges? We're hiring a doctoral student for the SSF-funded project "ML-Earth: Robust and data-efficient machine learning for Earth observation", supervised by myself (RISE Research Institutes of Sweden) and Prof. Yifang Ban (KTH Royal Institute of Technology).

You'll not only contribute to advancing state-of-the-art ML for Earth observation data — you'll also work hands-on with real-world use cases in collaboration with diverse project stakeholders: The Swedish Environmental Protection Agency, Geological Survey of Sweden, Kungsbacka municipality, and Metria.

🌍 The project also includes international collaborations and offers a unique opportunity to work at the interface of ML research, environmental science, and public-sector impact.

Feel free to share — and apply if this sounds like you!

https://career.ri.se/en-GB/jobs/5989297-phd-student-in-machine-learning

#PhD #DoctoralPosition #MachineLearning #EarthObservation #environment

RISE Research Institutes of Sweden AB

PhD Student in Machine Learning - RISE Research Institutes of Sweden AB

Do you want to use your expertise in AI to help build a greener future?

Original URL: https://career.ri.se/en-GB/jobs/5989297-phd-student-in-machine-learning

📣 Oisin Mac Aodha, Sara Beery, Jennifer, Nico Lang, Malte Pedersen, Jonathan Sauder

murilo (murilogustineli@gatech.edu)

2025-05-29 18:18:48

Hi everyone! Here’s the info for the Bird Walk during CVPR in a couple of weeks. Feel free to share on socials and spread the word!

Join us for a Bird Walk during CVPR!
The AI for Conservation community is hosting a morning bird walk on Friday, June 13th, from 7:00 AM to 9:00 AM at the scenic Shelby Bottoms Nature Center & Greenway (1900 Davidson St, Nashville, TN 37206). 📍Google maps location.

Just a 10-minute drive from the conference venue, this beautiful park offers lush trails, diverse bird species, and a quiet observation deck, perfect for a morning escape into nature.

We’ll be meeting directly at the location. If you'd like to coordinate rides, feel free to connect in the AI for Conservation Slack. Need an invite to join? Email us at aiforconservation@gmail.com.

All are welcome, no birding experience required!

🐦 Mitchell Rogers, Shir Bar, Jenna Kline, Tuan-Anh VU, Cara Appel, Oisin Mac Aodha, Aleksis Pirinen, Jaycee Choi, Alexander Merdian-Tarko, Ritika Allada, Hemal Naik, Kevin Barnard, Alan Stenhouse

❤️ Jon Van Oast, Tuan-Anh VU, Aleksis Pirinen, Riley Knoedler, Tessa Rhinehart

🦆 Laura Chrobak

Mitchell Rogers (mitchell.rogers@auckland.ac.nz)

2025-05-29 18:21:17

Will anyone else be leaving from the conference venue?

👍 Oisin Mac Aodha

Sara Beery (sara.beery@gmail.com)

2025-06-03 11:27:38

This isn't "organized by FGVC" as much as "organized collectively by the AI for Conservation community" 🙂

👍 murilo

murilo (murilogustineli@gatech.edu)

2025-06-03 11:54:08

thanks @Sara Beery I updated the post

😁 Sara Beery

Sara Beery (sara.beery@gmail.com)

2025-06-03 12:01:54

Not that FGVC isn't related! But so is Earth Vision, CV4Animals, etc. etc. 🙂

Anthony Hoogs (anthony.hoogs@kitware.com)

2025-06-04 12:51:39

Is there interest in getting together for a happy hour at CVPR? The AI for Conservation meetup at CVPR 2024 was very popular.

Sara Beery (sara.beery@gmail.com)

2025-06-04 13:35:07

I don't know what others schedules look like, but I decided to only attend weds-fri this year and I think the workshop days can get quite busy? That's why I hadn't brought it up, but would be so supportive of a community driven event ❤️

Anthony Hoogs (anthony.hoogs@kitware.com)

2025-06-04 18:37:02

I'd suggest an event on Wednesday or Thursday evening, since the conference does not have anything scheduled for those nights. Maybe Thursday, right after the CV4Animals workshop? What do you think?

Sara Beery (sara.beery@gmail.com)

2025-06-04 18:42:11

Wednesday and Thursday are the workshop days and there tend to be a lot of dinners etc, but I think it would be nice to have a more open and inclusive event! That sounds great, maybe just pick a bar and folks could drop in as available?

👍 Mitchell Rogers

Anthony Hoogs (anthony.hoogs@kitware.com)

2025-06-04 18:43:27

Sounds like a plan. I'll look into bars near the conference center.

❤️ Sara Beery

Anthony Hoogs (anthony.hoogs@kitware.com)

2025-06-05 15:28:57

The Broadway Brewhouse looks reasonable, many beers and a full menu https://maps.app.goo.gl/jvWMPXLG8bc48DYe8 https://www.broadwaybrewhousedowntown.com/. I'd recommend Wednesday 6pm, after FGVC. If this sounds good to everyone I'll post on the main thread.

google.com

Broadway Brewhouse Downtown · 317 Broadway, Nashville, TN 37201

★★★★☆ · Bar & grill

Original URL: https://maps.app.goo.gl/jvWMPXLG8bc48DYe8

Sara Beery (sara.beery@gmail.com)

2025-06-05 15:57:49

I think that sounds reasonable! I know the FGVC organizers will be at a dinner together, so wouldn't be there at 6. I would come by after dinner if people are still around!

Anthony Hoogs (anthony.hoogs@kitware.com)

2025-06-05 18:27:20

We could do Thursday if the FGVC dinner involves a lot of people who would otherwise join. A minor disadvantage of Thursday is it's the night before the bird walk, early at 7am, but maybe that doesn't matter much.

Rebekka (rebekka.stutz@sensingclues.org)

2025-05-30 02:31:06

Hey everyone! 👋 We're hosting a free webinar on our Sturgeon Agent-Based Model 🐟 – join us to explore how it supports conservation efforts!

🗓️ Thursday, 12 June 2025
⏰ 13:00 - 14:00 CEST
🔗 https://www.sensingclues.org/news/webinar-sturgeon-agent-based-model

Hope to see you there!

Sensing Clues

Webinar "The road towards predicting sturgeon migration" — Sensing Clues

Join us on 12 June 2025 for a webinar on predicting Beluga sturgeon migration. Discover how Sensing Clues Conservation Technologies, GIS, and cross-disciplinary research are shaping a practical tool to support conservation in the Danube Delta.

Original URL: https://www.sensingclues.org/news/webinar-sturgeon-agent-based-model

👍 Paul, Rebekka

Sara Beery (sara.beery@gmail.com)

2025-05-30 09:24:08

Sorry about the spam from that Chloe account everyone. I was hoping they would stop after being asked to, but they didn't so I've deactivated the account

👍 Georgia Atkinson, Malte Pedersen, Oisin Mac Aodha, Mark Goldwater, murilo, Devis Tuia, Alexander Merdian-Tarko, Toryn Schafer

🙏 Malte Pedersen, Paul, Shir Bar, Avishek Dangol, Carly Batist, Navodita Mathur, Mamoon, Kate Rakelly, Sara Si-Moussi, Polina Stucke, Morgan Langley

Paul Bodesheim (paul.bodesheim@uni-jena.de)

2025-05-30 10:26:39

📣 CamTrapAI Workshop 2025 in Seattle, USA hosted by Ai2 as hybrid event with online participation option on September 9-10!
📄 We just released the Call for Papers for extended abstracts (4 pages) and full papers (6 pages), specifically encouraging you to submit early findings and preliminary results.
🎯 Paper submission Deadline: July 1st (AOE)
🏅 Authors of selected best papers will be invited to submit extended versions to a Special Issue of the IET Computer Vision Journal.
✨ More information on the workshop website: https://camtrapai.github.io/workshop-2025.html

camtrapai.github.io

Camera Traps, AI, and Ecology - International Workshop Archive

Camera Traps, AI, and Ecology - International Workshop Series

Original URL: https://camtrapai.github.io/workshop-2025.html

🙌 Timm Haucke, Ted Schmitt, Dante Wasmuht, Peter Bull, Alexander Merdian-Tarko, Mohamed Elhoseiny, Aleksis Pirinen

🎉 Jon Van Oast

Ted Schmitt (teds@allenai.org)

2025-05-30 11:06:56

We are delighted to be able to host in our brand new location.

🎉 Jon Van Oast, Jason Holmberg (Wild Me), Mark Otterlee, Sara Beery, Dan Morris, Constanza Andrea Molina Catricheo, Lily Xu

Lily Xu (lily.x@columbia.edu)

2025-06-01 06:50:10

$10,000 grant opportunity for young (MS/PhD) female researchers in wildlife and environmental conservation: https://www.belikecam.org/scholarship/

❤️ Sara Beery, Dylan Van Bramer (she/her), Holly Houliston, Aleksis Pirinen

Carly Batist (carly.h.batist@gmail.com)

2025-06-02 11:08:36

Conservation AI Internship (paid) at Conservation International!
View the posting and apply here. Please share widely!

Where: Any US CI office (Arlington, VA/DC area; Seattle, WA; Santa Barbara, CA)
When: 3-month position starting ASAP (w/ option to extend); position open until filled
Who: undergrad or grad students enrolled in an educational institution or < 3 months post graduation (int'l students eligible if legally able to work in the US)
What: This internship position will support Conservation International's Moore Center for Science & Solutions in exploring and implementing AI solutions that address these fundamental conservation challenges. We are looking for someone familiar with modern AI technologies (generative AI, AI agents, LLMs, and foundation models) to prototype solutions, bridge communication between scientific staff and technical developers, and explore innovative approaches to conservation challenges.
How: apply here!

phh.tbe.taleo.net

Hiring Conservation AI Intern, - Arlington, VA, United States

View job details and apply now

Original URL: https://phh.tbe.taleo.net/phh04/ats/careers/v2/viewRequisition?org=CONSERVATION&cws=39&rid=2499&source=LinkedIn

👀 Youssef Bayouli, Mark Goldwater, Hannah Murray

🎉 Jon Van Oast, Talia Speaker, Elaine Steinberg, Lily Xu

💚 Lily Xu, Alan Stenhouse

Lukas Picek (lukaspicek@gmail.com)

2025-06-03 08:35:32

Hi all,
We’re excited to share that we will be hosting half-day tutorial on Animal Re-Identification at CVPR! 📢
When: June 12, 2025, 8:00–12:00 🕘 📍
Where: Room 202 C, CVPR (hybrid format available!)

🔍 What You’ll Learn:
• Intro to Re-ID: Discover the core principles behind identifying individual animals using computer vision—no invasive tags required! 🍃📷
• Real-World Impact: See how Re-ID helps track population trends, study animal behavior, and support conservation efforts worldwide. 🌿🦏
• Technical Deep Dive: Explore state-of-the-art algorithms, feature extraction, and data challenges unique to wildlife datasets. 🤖🐾
• Ethical Considerations: Understand privacy, data bias, and best practices for non-invasive monitoring. 🌱⚖️
• Interactive Q&A: Engage with experts 🎤

🎓 Organizers & Speakers: Lukáš Picek (INRIA) • Sara Beery (MIT) • Lasha Otarashvili (Wild Me) • Lukáš Adam (UWB) • Jason Holmberg (Wild Me) • Timm Haucke (MIT) • Jiří Matas (CTU) • Ryan Farrell (BYU) • Charles Stewart (RPI)

🔗 Full Details & Program: https://sites.google.com/view/animal-reid/ 👈

🐘 Oisin Mac Aodha, Sara Beery, Justin Kay, Elizabeth Campolongo, Thijs van der Plas, Timm Haucke, Dante Wasmuht, Subhransu Maji, murilo, Shir Bar, Alan Stenhouse, Aleksis Pirinen

🎉 Sara Beery, Nico Lang, Justin Kay, Cameron Trotter, Rupa Kurinchi-Vendhan, Jon Van Oast, Thijs van der Plas, Timm Haucke, Subhransu Maji, murilo, Malte Pedersen, Elena Sierra, Shir Bar, Kirk Larsen, Kevin Barnard, Aleksis Pirinen

👀 murilo, Laura Chrobak, Ed Miller

🙌 Kostas Papafitsoros

Ed Miller (ed@hypraptive.com)

2025-06-09 18:14:56

@Lukas Picek - awesome! Will this be recorded?

Lukas Picek (lukaspicek@gmail.com)

2025-06-10 09:56:21

Hi @Ed Miller, Yes, everything will be recorded but due to the CVPR guidelines we will share it in around 3 months.

👍 Thor Veen, Ed Miller, Antón Álvarez

Antón Álvarez (aalvarez@wwf.es)

2025-07-02 08:14:30

@Lukas Picek Wow, I totally missed that! I didn't realise it was a hybrid. Could you please share the video in the #animal_re-id channel too? Things get buried so quickly in #general. Thanks so much!

Lukas Picek (lukaspicek@gmail.com)

2025-07-02 08:38:25

Hi @Antón Álvarez, if you bought the CVPR registration, you should be able to watch the tutorial already on the CVPR website. If not, you have to wait 2 more months since we cannot share the recordings yet. I can ping you once it is available. 🙂

Antón Álvarez (aalvarez@wwf.es)

2025-07-02 08:46:15

Hi @Lukas Picek, I can wait 2 months...my organisation is not going to pay me the CVPR registration. It would be perfect if you could ping me when it's available, Thanks 🙂

👍 Lukas Picek

Vincent Lostanlen (vincent.lostanlen@gmail.com)

2025-06-03 11:21:59

Are you a researcher in signal processing for bioacoustics / ecoacoustics / urban / industrial / medical acoustics?

Come give a 45-minute talk in Paris!
on Thursday, Nov 13th, 2025
on the topic of your choice
streamed and recorded by Ircam’s professional A/V crew

Please share:
gdr-iasis.cnrs.fr/reunions/dom ...

Write to Mathieu Lagrange at mathieu.lagrange (@) ls2n.fr and Vincent Lostanlen at vincent.lostanlen (@) ls2n.fr before September 1st, providing a title, a summary of the presentation, and a brief biography. Presentations are preferably in English and last 45 minutes, followed by 15 minutes of questions.

GdR IASIS

Domaines émergents en traitement du signal audionumérique : éco-acoustique, environnements urbains, industriels et médicaux - GdR IASIS

(English version below) Éthologie, écologie, géographie urbaine, génie industriel et génie biomédical sont autant de disciplines qui valorisent le son comme information. Longtemps restreinte aux seuls signaux de parole et musique, le traitement du signal audio trouve dans ces disciplines des applications probantes pour l’instrumentation. En retour, il se nourrit... Lire la suite

Durée de lecture estimée

3 minutes

Written by

Mathieu Lagrange, Vincent Lostanlen

Original URL: https://gdr-iasis.cnrs.fr/reunions/domaines-emergents-en-traitement-du-signal-audionumerique-eco-acoustique-environnements-urbains-industriels-et-medicaux-2/

🎉 Jon Van Oast, Sara Beery, Ilyass Moummad

Vincent Lostanlen (vincent.lostanlen@gmail.com)

2025-06-04 02:59:41

sorry @Devis Tuia, will do and remember that in the future

Devis Tuia (devis.tuia@epfl.ch)

2025-06-04 03:00:11

no problem, just trying to keep the slack tidy 😉

🫡 Vincent Lostanlen

Dan Stowell (dan.stowell@naturalis.nl)

2025-06-03 11:31:55

The evaluation data for BioDCASE has been released! https://biodcase.github.io/challenge2025/ New acoustic data from forests and oceans, I'd love to know if your team will submit a solution to BioDCASE?

🙌 Benjamin Hoffman, Aleksis Pirinen

👍 Holger Klinck, Alexander Merdian-Tarko

🪶 Irina Tolkova

Dan Morris (agentmorris@gmail.com)

2025-06-03 17:36:20

Post-game recap on the "vibe coding party" discussed earlier on this chat... overall, a success! We had maybe ~10 people on average in the call (it fluctuated over what turned out to be a very long call), with a wide range of coding experience.

Before my summary, apologies to the ~3 people I left hanging in the waiting room for silly logistic reasons. I owe all of you fancy coffee the next time we're in the same city.

Now, notes for anyone who wants to organize something like this again:

The only "prep" was a google doc in which everyone was encouraged to write down some specific goals in terms of AI tools we wanted to try and/or things we wanted to try building with AI tools. I think this was constructive, and folks who wrote something in the doc mostly stuck to what they said they would do. IMO this was a local maximum in terms of ROI on prep time.

We spent almost an hour just going around the room and talking about our experiences so far with AI-assisted coding tools, with a brief discussion of where everyone is at wrt climate considerations of AI.

Then it got real quiet for a couple hours while everyone turned their cameras off and worked quietly, with the gentle social pressure of wanting to re-convene later and talk about what we learned.

Then we re-convened and shared some stuff we tried, and what we learned. People tried at least Windsurf, Cursor, Aider, Jules, and Perplexity Labs (which apparently is a thing).

Then most importantly, at the end we spent 10 minutes making games starting from a shared conservation-themed game concept about avoiding human-elephant conflict. That was fun, but also useful IMO, since most of us had spent the previous couple of hours working on much more mundane tasks that didn't necessarily encourage high-level prompting, and it's good to practice high-level prompting.

Gentle social pressure FTW, I would do this again if someone else wants to organize. Sharing screenshots of the games I made at the end (literally four games in two minutes, what a world); I encourage others to do the same.

🙌 Mitchell Rogers, Justin Kay, Felipe Parodi, Elly Knight, Bernie Boscoe, Lily Xu, Rachel Mason, Rachael Laidlaw, Jonathan Roberts, Carly Batist, Kalindi Fonda, Alexander Merdian-Tarko, Jose Ruiz-Munoz, Tessa Rhinehart, Sam Lapp, Talia Speaker

👀 Irina Tolkova

🙌:skin_tone_2: Cara Appel

🙌:skin_tone_5: Prabath Gunawardane

👏 Silvia Zuffi

🙌:skin_tone_3: Alan Stenhouse

Prabath Gunawardane (prabathg@gmail.com)

2025-06-03 18:14:26

It was a lot of fun! Thanks Dan & everyone else for organizing and participating!
I really enjoyed the collaborative / learning together aspect of this, maybe especially since I do most of my work independently now. I definitely won't mind organizing something like this again 🙂 .

It was amazing to see how much rapid development & prototyping we can do starting with simple / vague prompts. Especially impressed with tools that made working demos on the first attempt (Loveable & Vibes). I'm definitely impressed with Aider as well (even though it didn't get things right in one go) and will continue using it probably in combination with Cursor. It was great to see Perplexity in action too!

My screenshot from the Elephant game (I will try to get it to a playable state and share the link & code eventually 😄 ):

🐘 Dan Morris, Carl Boettiger, Yumi Arimitsu

Yumi Arimitsu (yarimitsu@gmail.com)

2025-06-04 12:51:17

The vibe coding party was great! I ended up with something useful at the end of the day: a functional website that still needs lots of work. Cursor helped me setup my DNS, get things working properly on my machine, setup git pages, and build the skeleton code for the webpage in about 1 hour (it would be faster if I knew what I was doing of course). I gave it links to publications and asked it to summarize abstracts and write the citations, which it did well ~80% of the time (there were plenty of hallucinations in this task though).

🎉 Dan Morris, Kalindi Fonda, Prabath Gunawardane

Ben Weinstein (benweinstein2010@gmail.com)

2025-06-09 13:35:51

Cursor 1.0 came out this week. They have a new concept. "background agents" that perform large long running tasks that you interact with. I paid 50$ for it and playing around now. My imagination is a bot that can check out issues on our github repos and submit PRs for use to review. They we could have a tag for 'these issues look like something a bot could do'. I am also experimenting with giving it a full project to own. I will report back.

❤️ Sara Beery, Nathan Fox, Prabath Gunawardane, Lily Xu, Chris Lange

😎 Jon Van Oast, Amee Assad, Sam Lapp, Dan Morris, Alan Stenhouse

Ben Weinstein (benweinstein2010@gmail.com)

2025-06-09 13:38:21

also this thread has a real, end of the world, vibe to it, so here is my cat pato who is working alongside me.

😻 Prabath Gunawardane, Elizabeth Campolongo, Dan Morris, Kalindi Fonda

Ben Weinstein (benweinstein2010@gmail.com)

2025-06-09 14:02:57

I was able to generate a background bot agent to achieve my task within ten minutes, which is great, but also gave me some weird existential crisis when it used my github username -> https://github.com/weecology/DeepForest/pull/1065

😂 Elizabeth Campolongo, Alan Stenhouse

Sam Lapp (sam.lapp@pitt.edu)

2025-06-10 21:33:54

@Dan Morris though I couldn’t join in real time I tried my own session. I had a pretty frustrating afternoon trying to use cursor to improve upon an existing GUI project. Do you feel like it’s really good at making games compared to other things? And/or letting the agentic AI work from scratch rather than improve upon an existing tools?

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2025-06-10 23:29:24

@Sam Lapp If you go from scratch, you need to be quite careful about not letting the models run off. It's going to be a mess for the first few commits, but once you get to a point where it's making small edits, things have gone quite well for me. I've found it's quite good practice to be very intentional about what you're asking it to do. If you ask for too much, it's likely to get lost.

Are you able to share more about what you wanted the model to do, and what it didn't?

I can't comment on GUI work because getting these models to make visual changes is the subject of many memes. But they can absolutely do more than games, especially if you let it access docs + web search.

The new cursor rule structure also helps, because you can get it to load e.g. design rules, commit rules as you need (most of the time it figures which mdc to grab). You really want to provide as many automatic constraints as possible (like your own tests, rules that it will check, asking it to plan out what it will do without writing code, etc).

👍 Sam Lapp

Dan Morris (agentmorris@gmail.com)

2025-06-11 09:21:57

@Sam Lapp I still don't really "get" Cursor, it just hasn't clicked for me. Part of that is just that the form factor doesn't fit my workflow, but part of it is the fact that Cursor places really tight limits on context (if $20/month for thousands of dollars of tokens sounds too good to be true, it is...). So, take advice from other comments (like Josh's) on how to get the most out of Cursor, but, my thoughts on other things you might want to try... although first, YMMV, I haven't necessarily done these things, they just sound like good ideas, I'm exploring all of these along with everyone else here. But, some other things to try...

Rather than using Cursor to do both the high-level thinking (which may exceed Cursor's context limitations) and the execution (which by all accounts Cursor is good at), use a chat interface (likely a free-tier chat interface is fine) to read your existing code and write a design doc for the changes you want, which Cursor (or you!) can then interpret. The Gemini interface allows you pull a GitHub repo in directly; if you prefer another (non-Gemini) model or your project isn't on GitHub, you can use something like repomix to squish your repo into an xml file that you can upload to any chat interface.

Rather than using Cursor, switch to plain-vanilla VS Code and install Gemini Code Assist. The experience should be similar to Cursor, but according to the documentation, even the free tier of GCA allows a 1M-token context window (so, e.g., using Gemini 2.5 Pro through GCA may be significantly different than using Gemini 2.5 Pro through Cursor). Reminder that I work for Google, so if you want to be suspicious of my recommendation to try a Google tool, by all means, be suspicious, I want nothing more than to hear that this was the best option. :) That said...

Splurge on $20 of Anthropic credits and try Claude Code. It may totally blow your mind. Whether the cost is sustainable is a separate question, but it's such a different experience than Cursor that it may expand your horizons.

👍:skin_tone_3: Alan Stenhouse

Josh Veitch-Michaelis (j.veitchmichaelis@gmail.com)

2025-06-11 10:50:37

Yeah I don't think there's anything particularly special about Cursor, you just need to find what working style is best for you. My experience (with Cursor) is about 2 solid days of hacking around.

At least the pricing is basically uniform for all the tools, so if $20-25 is affordable for you, try a different one every month. I would not pay for a year unless you're sure you want to commit to a working style. Things are changing too fast.

Kalindi Fonda (kalindi.fonda@gmail.com)

2025-06-12 02:54:31

@Ben Weinstein oh wow to the fact that it used your username.... All the other ones I've used so far use the bot-ty ones and I actually really appreciate that.

Sam Lapp (sam.lapp@pitt.edu)

2025-06-12 12:15:17

Great great, thanks all! Will plan to try GCA and Claude Code for my next trial runs.

Dan Morris (agentmorris@gmail.com)

2025-06-13 23:31:27

I had another vibe coding party today, this time with grad school friends. We used the same general approach that we used for the AI4C vibe coding party last week: everyone wrote some goals down in a doc, then we bantered about nonsense for a while, then we all went quiet and worked on stuff, then we came back for show and tell. This is a great way to force myself to set aside time to learn this new skill set, so, let the vibe partying continue!

My project tonight was a multiplayer game inspired by a random dragon combat simulator I thought was the bee's knees in 1990:

https://github.com/agentmorris/dragon-game

I won't say it was the best game ever created, but it totally worked, including multiplayer online play, and I didn't look at the code at all until after I had pushed it to GitHub. I'm still not entirely sure what language it's written in. The only thing I added was the screenshot at the top of the README. What a new world we've entered!

Dan Morris (agentmorris@gmail.com)

2025-06-13 23:34:07

On the other hand, it was also a tedious amount of back and forth over a couple hours, so while we're in an amazing place with these new tools, there's definitely a new set of skills to learn. In particular the slowest part of the system was me clicking stuff in a browser to tell the AI what was printing out in the console, and I feel like if I really knew what I was doing, I shouldn't be necessary for that part, and I could have spent the same amount of time making the content interesting. Lots to learn!

👍:skin_tone_3: Alan Stenhouse

stefano puliti (stefano.puliti@nibio.no)

2025-06-05 11:31:35

Hi! As part of a national Norwegian initiative to recruit scientists based in the US, we’re planning to hire a postdoc for a 3-year position.

If you already hold a PhD, are currently living in the US, and are experienced in deep learning—with a particular interest in forests—we’d love to hear from you! 😊

Let me know if you’d like a more formal or more casual version as well.

👍 Aleksis Pirinen

Anthony Hoogs (anthony.hoogs@kitware.com)

2025-06-06 13:39:25

If you're attending CVPR in Nashville next week, there will be an AI for Conservation happy hour on Thursday, June 12, 6pm-9+. The location will be the Broadway Brewhouse Downtown, in the heart of the city about two blocks from the convention center. The Brewhouse has dozens of beers on tap and a full dinner menu too. We don't have a room reserved, but we'll congregate near the front bar - wear your CVPR badge so we can find each other. No need to RSVP. Please spread the word and bring all your friends. Everyone is welcome!

🙌 murilo, Nathan Jacobs, Justin Kay, Malte Pedersen, Timm Haucke, Oisin Mac Aodha, Don Cosseboom, Mitchell Rogers, Shir Bar, Klára Janoušková, Chinmay Talegaonkar, Sara Beery, Laura Chrobak, Anthony Hoogs, Jaycee Choi, Aleksis Pirinen

🎉 Jenna Kline, Lukas Picek, S M Rayeed, Timm Haucke, Don Cosseboom, Sara Beery, Klára Janoušková, Lasha Otarashvili, Kevin Barnard, Riley Knoedler, Alice Heiman

🍺 Alan Stenhouse

Sara Beery (sara.beery@gmail.com)

2025-06-06 15:08:27

I'll plan to stop by after dinner!

Laura Chrobak (lchrobak@mbari.org)

2025-06-09 13:00:51

Can't wait!

Ben Weinstein (benweinstein2010@gmail.com)

2025-06-08 16:28:09

Is anyone using starlink roam to upload data, 500$, 50gb link upload per month, 1000$ a year subscription. Thinking about this ongoing challenge of bringing compute to remote locations. Very reasonable cost compared to bringing GPUs in the field.

❤️ Sara Beery

🤔 Jose Ruiz-Munoz

Alex Rood (alex.rood@wildlabs.net)

2025-06-09 13:42:48

If anyone is looking to get involved with building a stronger conservation tech community, applications for the WILDLABS group leadership program are open! We're looking for volunteers who can spend 1-2 hours a week participating in conversations on the platform, planning community events, connecting individuals, and more. It's a great way to build your network, make an impact in your field, and get leadership experience! 🙂

There are opportunities available for our AI subgroup in addition to our camera trap, acoustic, and animal movement groups (and so many more!) You can learn more and apply here: https://wildlabs.net/article/applications-open-become-2025-2026-wildlabs-group-leader

🎉 Jon Van Oast, Sara Beery, Youssef Bayouli, Dan Morris, Lily Xu

Youssef Bayouli (youssef.projetnemo@gmail.com)

2025-06-09 15:16:32

Hi all, 🤗

I just published a discussion thread on WILDLABS proposing a new community group on Edge AI and Edge Computing for Conservation, as part of my application for the 2025–2026 Group Leader Programme.
If you're working on, or interested in, things like low-power ML, real-time field data processing, or transmission solutions for remote monitoring—I'd love your feedback and support!
Here’s the post:
🔗 https://wildlabs.net/discussion/exploring-wild-edge-proposal-new-wildlabs-group
Feel free to react or leave a comment if this group sounds like something you'd like to be part of.
Thanks so much!

❤️ Owen Martin, Timm Haucke, Arthur Caillau, Anna Vallery

🎉 Jenna Kline, Navodita Mathur, Dan Morris, Timm Haucke, Aakash Gupta, Céline Angonin, Viktor Domazetoski

Sara Beery (sara.beery@gmail.com)

2025-06-10 13:57:12

For those attending CVPR, lets start a thread about relevant talks/workshops/etc 🙂

🙌 murilo, Bowen Zhang, Mitchell Rogers, Ana Maria Quintero

Sara Beery (sara.beery@gmail.com)

2025-06-10 13:59:07

Workshops/Tutorials:

Weds:
FGVC12 https://sites.google.com/view/fgvc12
EarthVision https://www.grss-ieee.org/events/earthvision-2025/

Thurs:
CV4Animals https://www.cv4animals.com/
Animal Re-Identification https://sites.google.com/view/animal-reid

sites.google.com

FGVC12 Workshop

Announcements Poster session: ExHall D poster boards #373 - #403 at time 4pm - 6pm. Poster board assignment: [here] Room: 104 E (Level 1) [Floor plan] [CVPR workshop page] FGVC12 is scheduled on June 11. Speakers announced! We are excited to hear from the best in class including: Randall

Original URL: https://sites.google.com/view/fgvc12

GRSS-IEEE

EARTHVISION 2025 - GRSS-IEEE

EARTHVISION 2025 June 11th, Nashville, TN, USA in conjuction with the Computer Vision and Pattern Recognition (CVPR) 2025 Conference Aims and Scope / Call for Papers Important Dates Program Challenge […]

Est. reading time

17 minutes

Original URL: https://www.grss-ieee.org/events/earthvision-2025/

cv4animals.com

512 Korean Veterans Blvd · Nashville, Tennessee

About

Original URL: https://www.cv4animals.com/

sites.google.com

Animal Re-ID

Overview

Original URL: https://sites.google.com/view/animal-reid

👍 Bowen Zhang, Navodita Mathur

❤️ Maria Napolitani

Owen Martin (owma6084@colorado.edu)

2025-06-11 02:05:45

Will any of these be available to non-attendees?

Devis Tuia (devis.tuia@epfl.ch)

2025-06-11 08:08:51

All papers of Earthvision are here: https://openaccess.thecvf.com/CVPR2025_workshops/EarthVision

Devis Tuia (devis.tuia@epfl.ch)

2025-06-11 10:37:40

I run a quick search on the main conf program and found these papers that could be interesting to check during your visit:

❤️ Sara Beery

Devis Tuia (devis.tuia@epfl.ch)

2025-06-11 10:39:30

Animal reconstruction / pose / motion estimation:

https://openaccess.thecvf.com/content/CVPR2025/html/Wang_AniMo_Species-Aware_Model_for_Text-Driven_Animal_Motion_Generation_CVPR_2025_paper.html

https://openaccess.thecvf.com/content/CVPR2025/html/Rao_Probabilistic_Prompt_Distribution_Learning_for_Animal_Pose_Estimation_CVPR_2025_paper.html

https://openaccess.thecvf.com/content/CVPR2025/html/Lyu_AniMer_Animal_Pose_and_Shape_Estimation_Using_Family_Aware_Transformer_CVPR_2025_paper.html

https://openaccess.thecvf.com/content/CVPR2025/html/Wang_AniMo_Species-Aware_Model_for_Text-Driven_Animal_Motion_Generation_CVPR_2025_paper.html

Datasets:

https://openaccess.thecvf.com/content/CVPR2025/html/Brookes_The_PanAf-FGBG_Dataset_Unde[…]the_Impact_of_Backgrounds_in_Wildlife_CVPR_2025_paper.html

https://openaccess.thecvf.com/content/CVPR2025/html/Gabeff_MammAlps_A_Multi-view_Video_[…]or_Monitoring_Dataset_of_Wild_Mammals_CVPR_2025_paper.html

❤️ Owen Martin

Anthony Hoogs (anthony.hoogs@kitware.com)

2025-06-12 01:03:21

And a reminder about our CVPR Happy Hour: Thursday, June 12, 6pm-9+, at the Broadway Brewhouse Downtown. The Broadway strip is very lively, not to be missed while you're in town. Bring your friends and join us for a beer, drink, dinner and the buzz of downtown Nashville.

🎉 Sara Beery, Tejaswi Kasarla

Sara Beery (sara.beery@gmail.com)

2025-06-12 09:53:27

For the bird walk tomorrow, if folks want to share rides we could meet at 6:30/6:40 by the conference center: https://maps.app.goo.gl/njCoLrRuYKShKWyF7

google.com

Find local businesses, view maps and get driving directions in Google Maps.

Original URL: https://maps.app.goo.gl/njCoLrRuYKShKWyF7

👍 Oisin Mac Aodha, Mitchell Rogers, Aaron Serianni, Jaycee Choi, Lasha Otarashvili, Tejaswi Kasarla, Burak Ekim

Tejaswi Kasarla (t.kasarla@uva.nl)

2025-06-13 07:06:39

Is there a meeting point at the Nature Center? I’ll head there directly.

Sara Beery (sara.beery@gmail.com)

2025-06-13 07:33:30

Yes, the meeting point on the original post! Near the entrance

👍 Tejaswi Kasarla

Jing Liu (lakex24@gmail.com)

2025-06-13 08:03:36

We (2 people) just got on uber, will be 10min late and will catch up if you already left meetup point

Tejaswi Kasarla (t.kasarla@uva.nl)

2025-06-13 08:04:10

About 4 minutes away from the meeting point.

Sara Beery (sara.beery@gmail.com)

2025-06-12 11:23:33

Very annoying, apparently slack just updated and now invite links can only be active for 30 days, which means my standard "email aiforconservation to join" will take some more management. Any vibe-coders want to come up with something that could automate the request to join?

😢 Mitchell Rogers, Tiziana Gelmi Candusso

Sara Beery (sara.beery@gmail.com)

2025-06-12 11:24:09

I've updated the link for now in the auto-reply

🎉 Jon Van Oast

Sara Beery (sara.beery@gmail.com)

2025-06-13 08:07:24

Bird walk folks, we're here right now! https://maps.app.goo.gl/NLwRvJz3H4y1SpnRA

google.com

Nashville, TN 37206 · Nashville, Tennessee

Find local businesses, view maps and get driving directions in Google Maps.

Original URL: https://maps.app.goo.gl/NLwRvJz3H4y1SpnRA

Sara Beery (sara.beery@gmail.com)

2025-06-13 08:09:23

But walking towards this point: https://maps.app.goo.gl/SYA47hZnstoitvD3A

google.com

Shelby Bottoms Greenway · Nashville, Tennessee

Find local businesses, view maps and get driving directions in Google Maps.

Original URL: https://maps.app.goo.gl/SYA47hZnstoitvD3A

John Martinsson (john.martinsson@ri.se)

2025-06-13 09:28:09

I want to share a webpage about a recent bioacoustic monitoring setup that we have established at the Auk Lab at Stora Karlsö in the Baltic Sea. The Auk Lab has been monitoring and collecting data about Common Guillemots at Stora Karlsö for over a decade, and this year I was fortunate enough to be part of a bioacoustic monitoring installation on the island. These seabirds are important indicators of the overall health of the ecosystems in the Baltic Sea. Hopefully we will learn more about the Guillemots from their vocalizations.

There will be many interesting audio-visual and other multimodal modelling challenges here (the curent weather conditions are logged every second et cetera), which will make for fun audio modelling problems. Can we jointly identify audio-visual behavior cues to learn more about what a certain vocalization might mean? Can wind and rain noise reduction methods be informed by the logged weather data? It is going to be fun trying to figure it out!

auklab-audio.github.io

Auklab Audio Research

News and updates about the continuous acoustic monitoring at the Auk Lab

Original URL: https://auklab-audio.github.io/

🐦 Elizabeth Campolongo, Alexander Merdian-Tarko, Joe Leggett, Alan Stenhouse, Aleksis Pirinen

Sara Beery (sara.beery@gmail.com)

2025-06-13 10:48:52

So fun to see so many of you this morning!

🐦 Oisin Mac Aodha, Timm Haucke, Kevin Barnard, Aaron Serianni, Rupa Kurinchi-Vendhan, Yseult Hb, Matthias Zuerl, Don Cosseboom, Cara Appel, Valentin Gabeff, Alice Heiman, Burak Ekim, Neerja Thakkar, Shir Bar, Ted Schmitt, Mohamed Elhoseiny, Jess Tam, Carly Batist, Fagner Cunha, Ritika Allada, Jorrit van Gils, Xiaoxuan (Shirley) Ma, Maria Napolitani, Alan Stenhouse, Aleksis Pirinen, Marion Richardot

👀 Levi Cai, Don Cosseboom, Marion Richardot

Dan Morris (agentmorris@gmail.com)

2025-06-13 12:05:10

Help me live vicariously... favorite bird sighting?

Sara Beery (sara.beery@gmail.com)

2025-06-13 12:05:55

There was a pretty nice set of hawks! Coopers and red-shouldered

🪶 Dan Morris

🦤 Mitchell Rogers

Levi Cai (levi@levicai.com)

2025-06-13 15:40:18

Hm...how hard is it to organize fish/marine-organism watching instead... 😛

🐠 Shir Bar, Kalindi Fonda, Malte Pedersen

Sara Beery (sara.beery@gmail.com)

2025-06-13 15:49:36

Obviously much harder in Nashville

Kevin Barnard (kbarnard@mbari.org)

2025-06-13 17:01:31

@Levi Cai we kinda did that two years ago in Vancouver 😉
https://wildlabs.net/event/fathomnet-workshop-cvpr-2023

🐟 Levi Cai

❤️ Sara Beery, Levi Cai

Levi Cai (levi@levicai.com)

2025-06-13 17:06:02

MIW was pretty sweet too! Snorkeling next time? 😛 I guess there's ICCV!

👍 Malte Pedersen

gvanhorn (grv22@cornell.edu)

2025-06-16 11:32:08

@Dan Morris here is your vicarious experience: audio vouchers for the species detected by Sound ID, and an animation of our track along with the location of the first confident detections. These are prototype outputs from an internal version of Merlin.

👀 Thijs van der Plas

👍 Thor Veen

Dan Morris (agentmorris@gmail.com)

2025-06-16 11:36:43

@gvanhorn That's super cool, is the whole pipeline for making that en route to being public, or should I think of "internal version" more like "Secret Grant Code" (tm)?

😂 Sara Beery

gvanhorn (grv22@cornell.edu)

2025-06-16 11:41:16

Still in the ideation / prototyping phase; so no concrete public plans. Mainly two goals with these tools:

Deep dive into what an ARU processing pipeline looks like where the goal is to output detected species lists and "good" vouchers

Rethinking what "My recordings" screen looks like in Merlin and how to facilitate sharing Sound ID sessions.

Dan Morris (agentmorris@gmail.com)

2025-06-16 11:48:09

Awesome, let me know if/when I can beta-test. Also this should be built in to Strava. I'm busy listening to podcasts while I'm running, but it would be fun to go to a run page on Strava and see/hear all the birds I didn't notice, or get buzzed during my run when Merlin is confident that it hears a bird I haven't seen before. Totally willing to risk burning my battery out and having to run the last few minutes - gasp - without podcasts.

💡 Malte Pedersen

Simple Poll

2025-06-13 15:24:02

SlackAction(SlackButton(Open))
SlackAction(SlackButton(Add option))
Created by @Patrick Beukema with /poll

Patrick Beukema (patrickb@allenai.org)

2025-06-13 15:24:51

Hi all,

I confirmed with Sara already that this was ok to post. Just looking for some general feedback on some work we are doing now on GPS.

Appreciate any feedback or comments!

David C. Schedl (david.schedl@fh-hagenberg.at)

2025-06-14 11:18:05

We built a tool for annotating drone imagery/videos to retrieve geospatiol locations and tracks of animals. Not sure if that falls within the applications you had in mind. But reach out please if so. :) [edited as typed on tablet, sorry]

🙏 Patrick Beukema

Mohamed Elhoseiny (mohamed.elhoseiny@gmail.com)

2025-06-15 16:38:24

https://x.com/baldassarrefe/status/1933975376377892974?s=46&t=6oRdhrrtKxRsNDVj8lI3Kw

X (formerly Twitter)

Federico Baldassarre (@BaldassarreFe) on X

DINOv2 meets text at #CVPR 2025! Why choose between high-quality DINO features and CLIP-style vision-language alignment? Pick both with dino.txt 🦖📖 We align frozen DINOv2 features with text captions, obtaining both image-level and patch-level alignment at a minimal cost. [1/N]

Original URL: https://x.com/baldassarrefe/status/1933975376377892974?s=46&t=6oRdhrrtKxRsNDVj8lI3Kw

👏 Cameron Trotter

Elizabeth Campolongo (e.campolongo479@gmail.com)

2025-06-16 16:32:48

Beetles, butterflies, birds, plants, fish, …. and beyond!
How can we apply recent ML and AI innovations, in particular in computer vision, to most effectively and accurately extract functional traits from images?

Bring your data and questions about functional traits,

Bring your own ML tools,

Bring your enthusiasm, skills, and expertise,...

And join Imageomics for FuncaPalooza 2025!

Applications due June 23rd (see QR code for details).

❤️ Alessandra Vidal Meza, Sara Si-Moussi, Yseult Hb, Chris Lange, Sara Beery, Maria Napolitani, Wei-Lun (Harry) Chao, Holly Houliston, Alan Stenhouse

🎉 Jenna Kline, Leonardo Buitrago, Sam Stevens, Scot Nielson

Maria Napolitani (napolitanima@gmail.com)

2025-06-18 17:20:23

Hi Elizabeth! I just submitted my application for FuncaPalooza 2025. I’m super excited about the event! It would be amazing to work on my fish image dataset there and learn from everyone involved. Thanks for sharing it here! this workshop feels like the perfect space to grow both technically and scientifically and it would be a great experience outside Brazil! 😊

❤️ Elizabeth Campolongo

Levi Cai (levi@levicai.com)

2025-06-18 14:54:42

I dunno where this belongs, but one of my biologist collaborators is interested in annotation tools that allow simultaneous audio and visual annotation (and lets you see them both in the same interface, synced), ideally a local things so she can just annotate multiple 30GB videos this way (both with frame-level and hopefully polygon annotations). Is this a thing? Or do people handroll multiple things and just write code?

Oisin Mac Aodha (macaodha@caltech.edu)

2025-06-18 16:11:56

We recently developed this annotation tool, but it is only for audio:
https://github.com/mbsantiago/whombat/

Levi Cai (levi@levicai.com)

2025-06-18 16:19:26

Very cool! Unfortunately the video-audio simultaneous visualization is quite desired (especially with annotating the video frames in response to things happening in the audio, if that makes sense)

👍 Oisin Mac Aodha

Benjamin Hoffman (benjaminsshoffman@gmail.com)

2025-06-18 16:34:34

We used Premiere Pro for something like this. It required a lot of workarounds, but ultimately got the job done.

Levi Cai (levi@levicai.com)

2025-06-18 16:43:14

Yeah, we were also talking about using Premiere, but exporting labels and data seems rather funky. Do you guys have anything written up about that? No worries if not, I understand it's likely quite bespoke.

(Aside: seems like there's space for multi-modal annotation tools...hehe....maybe AI tools will get there quickly...)

Benjamin Hoffman (benjaminsshoffman@gmail.com)

2025-06-18 16:47:16

unfortunately not, and i don’t think the code for dealing with exports is in a condition to share atm

👍 Levi Cai

Owen Martin (owma6084@colorado.edu)

2025-06-18 17:26:26

BORIS is an option perhaps?

Owen Martin (owma6084@colorado.edu)

2025-06-18 17:26:44

https://www.boris.unito.it/

Levi Cai (levi@levicai.com)

2025-06-18 17:53:52

That was the main one in consideration, but they were not able to get their files to load at all it seems. Thank you!

Arthur Caillau (arthur@caillau.me)

2025-06-19 12:38:35

labelstudio ? You can run it locally and annotate audio and video data.

Levi Cai (levi@levicai.com)

2025-06-20 11:55:14

I will look into it! We had started going down the CVAT path since it lets us at least use big files and just doing the acoustics separately unfortunately. I had seen Labelstudio, but didn't realize you could annotate audio, thanks for the info!!

Julia Chae (chaenayo@mit.edu)

2025-06-20 00:03:54

We are thrilled to announce that CV for Ecology Workshop is returning for its second year at #ICCV2025 in Honolulu, Hawaii! If your work combines computer vision and ecology, submit a paper and join us!

Deadlines: July 4 (Proceedings) / July 22 (Non-Archival)

🎉 Timm Haucke, gyeongsu, Mitchell Rogers, Justin Kay, Thijs van der Plas, Holly Houliston, Cameron Trotter, Sara Beery, Elena Sierra, Rupa Kurinchi-Vendhan, Andy Huynh, Dan Morris, Owen Martin, Gustavo Perez, David Russell, Levi Cai, Aleksis Pirinen, Kevin Barnard, Yuyan Chen, Shir Bar, Jonathan Sauder, Robin Zbinden, Nico Lang

❤️ Timm Haucke, murilo, Jonathan Sauder, Elena Sierra, Andy Huynh, Rupa Kurinchi-Vendhan, Nina van Tiel, Yseult Hb, Alan Stenhouse

👍 Oisin Mac Aodha, Andy Huynh, Sam Stevens

Levi Cai (levi@levicai.com)

2025-06-20 11:50:42

Very cool! Is there a link to the workshop?

❤️ Julia Chae

Julia Chae (chaenayo@mit.edu)

2025-06-20 17:53:06

We are preparing an update to last year’s https://cv4e.netlify.app ASAP with a new submission link! I will post an update here when ready. Apologies for the delay

cv4e.netlify.app

CV4E

Workshop on computer vision for ecology at ECCV 2024

Original URL: https://cv4e.netlify.app

❤️ Andy Huynh

Julia Chae (chaenayo@mit.edu)

2025-06-24 01:08:42

Our CV4E Workshop Submission link is live! Looking forward to cool submissions from the community 🙂 🎉

For more information, please visit our website: https://cv4e.netlify.app/submit/

openreview.net

ICCV 2025 Workshop CV4E

Welcome to the OpenReview homepage for ICCV 2025 Workshop CV4E

Original URL: https://openreview.net/group?id=thecvf.com/ICCV/2025/Workshop/CV4E

CV4E

Submit a paper or abstract | CV4E

Original URL: https://cv4e.netlify.app/submit/

🎉 Yuyan Chen, Robin Zbinden, Hemal Naik, Shir Bar, Jason Holmberg (Wild Me), Elena Sierra, Mélisande Teng, Timm Haucke, Andy Huynh, Otto Brookes, Aleksis Pirinen, Justin Kay

Julia Chae (chaenayo@mit.edu)

2025-07-02 12:27:58

Only two days until the archival deadline! 🚨 🚨 I want to highlight that our submission requirements match ICCV standards (4 pages for short-form and 8 pages for long-form). Thank you to the members of our community for pointing out the discrepancy on our website!

If there are any questions, please reach out to myself or other members of our committee!

Julia Chae (chaenayo@mit.edu)

2025-07-03 18:57:15

Due to popular demand, we have extended the deadline to the 7th for our Proceedings track!

https://x.com/cv4e_iccv/status/1940907333145804850?s=46&t=lC6jjlU2f593Nlj8jo9ILw

X (formerly Twitter)

CV4E Workshop @ ICCV (@CV4E_ICCV) on X

🚨 Good news! The submission deadline for the CV4E workshop at #ICCV2025 has been extended from July 4 to July 7! #CV4E #ICCV

Original URL: https://x.com/cv4e_iccv/status/1940907333145804850?s=46&t=lC6jjlU2f593Nlj8jo9ILw

🙌 Felipe Parodi, Andy Huynh, Navodita Mathur, Robin Zbinden, Sara Beery

Anthony Hoogs (anthony.hoogs@kitware.com)

2025-06-20 18:25:10

Similarly, at ICCV 2025 there will be the Joint Workshop on Marine Vision. This workshop combines two previous workshop series in computer vision related to the oceans, both above the surface and below. If you are preparing a paper in this area, please submit! Due date is June 30.

🙌 Rupa Kurinchi-Vendhan, Navodita Mathur, Shir Bar, Julia Chae, Andy Huynh, Malte Pedersen, Sam Stevens, Robin Zbinden, Hirokatsu Kataoka (AIST / Oxford VGG)

Sonny Burniston (sonnyburniston@yahoo.co.uk)

2025-06-24 15:19:32

Hi Everyone,
Does anyone know of any software for doing distortion correction on underwater GoPro footage?
Also wondering if anyone had any recommendations of settings to minimise distortion? I've seen online that Wide is recommended over Linear as Linear has a tendency of overcorrecting.
Interested to hear from anyone who has experience using GoPro footage with models and handling distortions,
Thanks in advance! 😊

Sara Beery (sara.beery@gmail.com)

2025-06-25 07:17:41

@Jonathan Sauder @Rupa Kurinchi-Vendhan

Jonathan Sauder (sauder@mit.edu)

2025-06-25 08:33:06

There's no out of the box solution for all GoPro footage, because the distortion changes depending on whether you use a case, and even (a little) depending on the water conditions at hand.

If you have imagery (videos or a bunch of images) that is suitable for structure-from-motion photogrammetry (i.e. the main movement of the camera is translation and not rotation, and you can clearly see rigid substrate features throughout your footage), your best bet is to take some images (or extract some frames from your video at 3-5 FPS), and throw them into COLMAP or GLOMAP and run the reconstruction with the RADIAL camera mode, giving you a very good estimate of the distortion parameters. You can then undistort with COLMAP itself (https://colmap.github.io/cli.html), or use Python+OpenCV.

In general the linear setting will have the least distortion. The wide view will already be distorted even when you're not underwater. At the same time, I think the linear setting is essentially only a digital crop of the wide view, so you might as well take videos in wide and then crop them yourself. COLMAP should be able to give you accurate distortion parameters for undistortion for either video setting.

Hope this helps!

Sonny Burniston (sonnyburniston@yahoo.co.uk)

2025-06-25 09:17:49

Hi @Jonathan Sauder, thanks so much! This is super useful!
So my use case is running pose estimation models on footage of sharks and rays.
I’m trying to use camera settings that will minimise any distortion so I don’t need to do much post correction on the footage prior to training/ inference.
I did see this which seems to suggest wide mode underwater has very minimal distortion?
https://www.goproforums.com/threads/lens-distortion-underwater.89767/

Rupa Kurinchi-Vendhan (rkurinchivendhan@gmail.com)

2025-06-25 10:12:14

Jonathan's covered it! The wide-angle settings definitely have more distortions--using the linear setting is your best bet. This works better if the sharks/rays you are monitoring are closer to the camera. Something that I've tried before was a pipeline for directly removing fisheye-effects (https://github.com/duducosmos/defisheye) but this requires some guessing of camera parameters. to get outputs that look good.

Cody Kupferschmidt (kupfersc@uoguelph.ca)

2025-06-25 10:24:24

I did some work on this back in 2015. The GoPro lenses (at the time) were cheap, and to get reasonable results I had to calculate the co-efficients for each lens. OpenCV has some tools that help you do this. Another option is to perform your analysis in the wide-angle space.

Sonny Burniston (sonnyburniston@yahoo.co.uk)

2025-06-25 10:42:12

Thanks all for the input! Here is what I managed to get modelling with openCV. Green dots are the corner positions that OpenCV actually detects in the photo whilst the red dots are where those same chessboard corners should land according to the camera model I calibrated.
I get the red dots by taking the known 3-D coordinates of each checker-square corner, transforming them with the board’s pose for that image, and then projecting them through the calibrated lens ( K + distortion ).
The pixel gap between each red-green pair is the reprojection error which in this case (one of the worst images) is 6.362 px.
This is for wide mode FYI

Sonny Burniston (sonnyburniston@yahoo.co.uk)

2025-06-25 10:43:02

This is for Linear - here one of the worst errors with 2.831 px

Sonny Burniston (sonnyburniston@yahoo.co.uk)

2025-06-25 10:43:42

RMS reprojection error : 1.0483 px for Linear
RMS reprojection error : 2.5203 px for Wide

Sonny Burniston (sonnyburniston@yahoo.co.uk)

2025-06-25 10:44:19

Any one has any further thoughts/ feedback on this it would be much appreciated! 😊
(also feel free to DM me if you like)

Jonathan Sauder (sauder@mit.edu)

2025-06-25 10:50:53

As far as I know, current wisdom seems to be that successful SfM (with something like COLMAP) in a rigid scene often gives better calibration than OpenCV calibration, which can be quite sensitive, for example to the motion blur that's visible in your images!

Cody Kupferschmidt (kupfersc@uoguelph.ca)

2025-06-25 10:52:16

Note that your error is going to very pretty significantly depending on where in the image the checkerboard is placed.

Sonny Burniston (sonnyburniston@yahoo.co.uk)

2025-06-25 10:54:45

Yeah I'm aware i'm not fully capturing the distortion at the edges as the checkerboard is pretty central in my clips

Sonny Burniston (sonnyburniston@yahoo.co.uk)

2025-06-25 10:55:14

So the recommendation would be to go with Linear and then run through COLMAP?

Jonathan Sauder (sauder@mit.edu)

2025-06-25 10:58:02

Linear or wide shouldn't matter too much! After undistortion with correctly estimated distortion params, either should give you a pinhole image. But yes, my recommendation is to make a small reference SfM scene with COLMAP

Sonny Burniston (sonnyburniston@yahoo.co.uk)

2025-06-25 11:01:14

Thanks! I'm not too familiar with COLMAP. Is it relatively easy to use?

Jonathan Sauder (sauder@mit.edu)

2025-06-25 11:10:56

I'd say so, but I may be biased cause I'm very familiar with it! But the installation has always worked out of the box for me and the GUI is quite intuitive, happy to answer questions about what kind of imagery to put in and what settings to choose on the feature extraction & matching

Sonny Burniston (sonnyburniston@yahoo.co.uk)

2025-06-25 11:12:07

That would be great, I may DM you if that is okay?
So is it a case of uploading images and it will generate a correction map that you can then upload future images to and run it through?

stefano puliti (stefano.puliti@nibio.no)

2025-06-26 06:14:39

👋 for those interested in forest 3D vision, we just released the #ForestFormer3D pre-print 🚀, presenting the our latest panoptic model that is trained in an end-to-end fashion to perform panoptic segmentation (ground, leaf wood) and instance segmentation (individual 🌳) in forest point cloud scenes.

This model outcompetes all previous baselines on open forest point cloud benchmarks 📈

https://www.linkedin.com/feed/update/urn:li:activity:7343920037180628992/

linkedin.com

:evergreen_tree::rocket: Big News from the #SmartForest frontier! | SmartForest4.0

🌲🚀 Big News from the #SmartForest frontier! 📣The ForestryAI@NIBIO team just dropped a game-changer! 💥 💣 Introducing #ForestFormer3D – our new state-of-the-art transformer model for forest panoptic segmentation! 🌳It tackles semantic (ground, wood, leaf) and instance-level (individual trees!) segmentation in 3D forests like never before. 📃 pre-print: <https://lnkd.in/dQBNG2Fs> project page: <https://lnkd.in/dbWjQSJb> code and data will follow later! 🔍Highlights: ✨ FOR-instance V2 – a next-level benchmark for complex, layered, and overlapping tree canopies 🌳🌳🌳 🧠 Transformer architecture – more complex architecture &#61; improved performance 🎯 Instance + semantic-aware prompting – intelligent prompts &#61; better results 🧩 other innovations – score-based block merging + one-to-many association 🏆 Beats the best – outperforms #ForAINet, #SegmentAnyTree, and #TreeLearn 📊 🌲 With such advances, #SingleTree forest management is getting closer and closer to becoming operational reality. ✨ A world of opportunities lies ahead. 👏 Huge congrats to binbin xiang for leading this groundbreaking study, and shoutout to our awesome team at NIBIO (Maciej Wielgosz, Stefano Puliti, Rasmus Astrup), and Czech collaborators Kamil Kral, Martin Krucek, and Azim Missarov. #AI #Forestry #DeepLearning #ForestTech #PanopticSegmentation #NIBIO #SOTA

Original URL: https://www.linkedin.com/feed/update/urn:li:activity:7343920037180628992/

❤️ Owen Martin, Andy Viet Huynh, Sara Beery, Binbin Xiang, Ghazi Randhawa

👍 Aleksis Pirinen, Andy Viet Huynh, Runyu Guan, Tanya Berger-Wolf, Ayush Prasad

🌳 Sara Beery

Ben Weinstein (benweinstein2010@gmail.com)

2025-06-26 10:02:08

Super cool! Any results/intuition on lower density LiDAR clouds?

👍 David Russell, stefano puliti

stefano puliti (stefano.puliti@nibio.no)

2025-06-26 11:35:25

We didn’t check as is very hard to get our hands on reliably labeled dataset. No doubt it outcompetes CHM segmentations but not sure by how much :) perhaps an interesting research question?

👍 Ben Weinstein

Ben Weinstein (benweinstein2010@gmail.com)

2025-06-26 12:00:21

We have some draped RGB annotations over the cloud, if it’s ever useful to check. I’ll have a look at it once the code is released. As you said, not exactly pristine annotation structure.

Ben Weinstein (benweinstein2010@gmail.com)

2025-06-26 12:01:30

Not a massive priority for us, so if any students here want a go, happy to share.

Paul Bodesheim (paul.bodesheim@uni-jena.de)

2025-07-02 06:29:31

📣 We extended the deadline for submissions to the CamTrapAI Workshop to July 9th (AOE)
✨ Check the workshop website for submission instructions and further information: https://camtrapai.github.io/workshop-2025.html

camtrapai.github.io

Camera Traps, AI, and Ecology - International Workshop Archive

Camera Traps, AI, and Ecology - International Workshop Series

Original URL: https://camtrapai.github.io/workshop-2025.html

🎉 Sara Beery, Andy Huynh

Dan Morris (agentmorris@gmail.com)

2025-07-02 12:01:36

I still find that I'm only making progress on my "AI coding journey" when I set aside time to sit in a (virtual) room with other people to specifically focus on this. So, are folks up for doing another "vibe coding party"? More specifically: I'm going to set aside 9am-12pm PT on Monday July 14th to build stuff with AI, if anyone wants to join me, ping me by email (agentmorris@gmail.com).

FWIW the last one was both fun and productive. So, barring other suggestions, we will adopt the same structure as last time: I'll send out an empty doc in advance for everyone to write down goals (tools to explore and/or stuff to build), then we'll spend maybe 20 minutes talking about our recent AI coding experiences, then we'll each go off and code quietly and check in with the room as necessary, then we'll re-group near the end for show and tell and a "speed round" where we all totally vibe-code from a common idea/prompt.

This is not mutually exclusive with any other times that folks want to do this, there is IMO no limit to how much time we should all be investing in building these new skills, so if someone wants to organize one literally 30 minutes after this one, or on a weekend, or whatever, (likely) count me in.

✔️ Jon Van Oast

🙌 Justin Kay, Josh Seltzer, Carly Batist, Shir Bar, Amee Assad, Negar Sadrzadeh

💃 Yumi Arimitsu

🙌:skin_tone_5: Prabath Gunawardane

Carly Batist (carly.h.batist@gmail.com)

2025-07-02 13:45:47

Ahhh I'll be on vacation that week but definitely want to set these up more regularly!

Prabath Gunawardane (prabathg@gmail.com)

2025-07-02 16:57:20

I really enjoyed last time, and would love to join again! Will drop you an email.

Yumi Arimitsu (yarimitsu@gmail.com)

2025-07-03 17:44:52

Thanks for organizing @Dan Morris. I would love to join. I've been on my own AI coding journey since that last one. I've messed with Cursor, Cline, and Claude (in and out of VS Code) to build a number of things, and have been watching Reddit pretty closely for tips and tricks (hahaha and whatever other crazy stuff happens there). I'm ready to Party!

Kalindi Fonda (kalindi.fonda@gmail.com)

2025-07-07 04:26:51

Oh no! Will be offline for the whole duration, but would love to attend any future ones.

Christine Muthee (xtinemuthee@gmail.com)

2025-09-25 10:12:37

Ready to join you anytime..

Tanya Berger-Wolf (berger-wolf.1@osu.edu)

2025-07-02 21:31:42

Join the Department of AI and Society, University at Buffalo: Multiple Faculty Openings (Open Rank)
The Department of AI and Society (AIS) is the newest department in the University at Buffalo (UB) jointly supported by the School of Engineering and Applied Science (SEAS) and the College of Arts and Sciences (CAS). The department of AIS has been initially supported by New York State with the goal of creating a future world where AI systems are built by society, for society.
As part of our founding initiative, we are recruiting transdisciplinary scholars for multiple positions as Assistant Professor, Associate Professor, or Full Professor. We expect to hire up to tenfaculty members over the next two years.
Review of applications is ongoing, with appointments beginning in Fall 2025; applications will be considered until all positions are filled.
If you know anyone who might be interested in such a position, please point them to the following job posting:
https://www.ubjobs.buffalo.edu/postings/57734
For questions about the position, please contact
- Atri Rudra (atri@buffalo.edu), chair of the department,
- Jeff Good (jcgood@buffalo.edu), associate chair.
Warm regards,
Enki Yoo

ubjobs.buffalo.edu

Assistant, Associate or Full Professor, AI & Society

The Department of AI and Society (AIS) at the University at Buffalo (UB) invites candidates to apply for multiple positions as Assistant Professor, Associate Professor, or Full Professor. The new AIS department was created with the goal of creating AI systems that are designed to be built by society, for society. We seek to upend the dominant approach where AI systems are built with accuracy as the primary metric and then “fine tuned” for societal considerations. A successful candidate is expected to contribute to our initial attempts to make this vision a reality. To that end, we are looking for candidates who can operate effectively in a team environment with a diverse community of students and faculty, and who share and want to help shape our vision for a new department in the field of AI and Society.Review of applications will begin immediately with start dates in Fall 2025 and beyond; applications will be considered until all positions are filled.The AIS department seeks to hire interdisciplinary scholars whose research agenda connects the study of AI with humanistic and/or social scientific line(s) of study and are open to hire scholars with training in a wide range of home disciplines, including (but not limited to) Social Work, History, Computer Science, Engineering, Public Health, Digital Media and Art, Education, Psychology, Sociology, Geography, Languages and Literatures, etc. We expect a successful applicant to be a transdisciplinary researcher who incorporates insights across AI and the arts, humanities and/or one or more of the social sciences.We envision successful candidates might include scholars who:Use methodologies, theories, and/or concepts from the arts, humanities and/or social sciences to develop new kinds of AI (e.g. new models or new uses of existing models informed by humanistic or social scientific theory);Use AI to engage social science questions (e.g. computational social science) or humanistic lines of inquiry (e.g. digital humanities concerned with social change and justice);Develop community-driven (e.g. participatory-design-based) AI solutions to address societal problems (e.g. HCI, Human Factors, and other design scholars specializing in community work);Pursue critical work on AI through a social scientific or humanistic lens (e.g. critical algorithm studies or critical artistic work).These lines of inquiry are by no means exhaustive, but reflect our belief as a department that 1) we cannot have AI systems that are built for society without undertaking meaningful analyses of the strengths and weaknesses of existing AI systems, 2) that AI can, if used effectively, be an important tool in the social sciences, humanities, and arts, and 3) that attempts to build “AI for social good” that do not involve impacted communities from the start of the design process are doomed to fail. Given that AIS is a new department, initially, all faculty members are expected to fully engage and participate in the building of a new department including but not limited toDeveloping a vision for the department that encompasses its role in enhancing training opportunities and engaging communities around issues related to AI and society,Developing department policies and procedures,Creating undergraduate and graduate programs, including developing and implementing new courses, in the <a href="http://department.In">department.In</a> the steady state, successful candidates will be expected to teach courses at the graduate and undergraduate levels, mentor graduate students, advise students at all levels, and maintain an active research program. The successful Associate Professor or Full Professor candidates should also have a record of scholarly accomplishments, teaching experience, and a record of sustained external funding through grants, fellowships, or foundations. We also expect any successful candidate to work with faculty and experts from multiple fields within the AIS department and/or the UB community in general.About the DepartmentThe Department of AI and Society (AIS) is the newest department at UB and reports to both School of Engineering and Applied Science (SEAS) as well as the College of Arts and Sciences (CAS). The department has been initially supported by New York State with the goal of being the epicenter of new foundational education and research at the intersection of AI and society. The University at Buffalo (UB) is a flagship of the State University of New York (SUNY) system, a member of the prestigious American Association of Universities, is the largest and most comprehensive university in the SUNY system, with about 21,000 undergraduates 11,000 graduate students, and 2500 full-time faculty. The College of Arts and Sciences is the home for the liberal arts at UB, with 30 departments, 16 academic programs, and 23 centers and institutes covering the full range of the Arts and Humanities, Social Sciences, the Natural Sciences, and Mathematics. The School of Engineering and Applied Sciences has about 8,000 students across 10 academic departments. Funding of $400M was recently announced to launch EmpireAI, located at UB and supporting the state-wide consortium. The prestigious NSF National AI Institute for Exceptional Education continues to innovate in the application of AI to address important problems across disciplines. Buffalo is a city with a rapidly growing economy, eclectic neighborhoods, world-class art galleries and museums, a vibrant theater and music community, the Lake Erie waterfront, a city-wide system of parks designed by renowned landscape architect Frederick Law Olmsted, and major and minor league sports teams (yes, we are Buffalo Bills fans!). In short – Buffalo is a well-kept secret. It’s not just snow (although yes, we do get snow), but a city where you can afford to raise a family while still enjoying a vibrant small-city life. And, of course, the awe-inspiring Niagara Falls is just 20 minutes away. The department will be located on the UB North Campus in suburban Amherst, an area that combines outstanding public schools and services with a surprisingly low cost of living. Learn more:Our benefits, where we prioritize your well-being and success to enhance every aspect of your life.Being a part of the University at Buffalo community.University at Buffalo is an affirmative action/equal opportunity employer and, in keeping with our commitment, welcomes all to apply including veterans and individuals with disabilities.

Original URL: https://www.ubjobs.buffalo.edu/postings/57734

Tanya Berger-Wolf (berger-wolf.1@osu.edu)

2025-07-07 07:34:37

Renaissance philanthropies BiTS moonshot cohort programs are live. UK deadline July 25, Germany Aug 1, the rest will be in the fall:
https://bits.renaissancephilanthropy.org/

bits.renaissancephilanthropy.org

Big if True Science Accelerator

A talent accelerator for scientists to design and execute coordinated research programs to unlock transformative technologies for humanity's future.

Original URL: https://bits.renaissancephilanthropy.org/

👀 Kalindi Fonda

Atul Ingle (ingle@uwalumni.com)

2025-07-08 01:05:56

I'm looking for an ROC curve for megadetector. What true positive/false alarm rate can I expect when using the default detection thresholds? Eventually I'd like to have a "slider control" that sweeps over points along the ROC curve. Does this make any sense? Has anyone tried this before?

👀 Eugene Galaxy

Sebastien Ouellet (sebouel@gmail.com)

2025-07-08 02:38:38

It makes sense, I use ROC/PR curves every time I look at classification tasks to either adjust the threshold given the impact of false positives or to give users some additional control, and for Megadetector I would calculate them myself using a hand-labeled subset. Papers using Megadetector may publish their own curves but dataset characteristics will differ enough that we should compute them on a case by case basis

:gratitude_thank_you: Atul Ingle

Benjamin Kellenberger (benjamin.kellenberger@yale.edu)

2025-07-08 03:26:55

ROC curves are not a sensible option for object detection. Their calculation involves the concept of true negatives, of which there are theoretically up to infinity in such a context. This is why object detection performance is usually measured with precision-recall curves instead.

Since MegaDetector is these days built on Ultralytics/YOLO, it should be relatively straightforward to obtain those directly.

As for real-world performance, I dare say this depends on the dataset complexity and environment that you are using. Also, not sure if P-R curves, mAP or other metrics have ever been published for MD. It's probably best to compute it yourselves on a held-out test set.

👍 Jonathan Sauder, Atul Ingle, Justin Kay, Bernie Boscoe

Atul Ingle (ingle@uwalumni.com)

2025-07-08 11:18:43

My collaborators are primarily interested in removing all the empty images, not so much about the locations of detected bounding boxes.

So I'm thinking of using megadetector just as a 2-class classifier for their images ("empty vs interesting"), and not care about the location of the bounding boxes.

Perhaps an ROC curve would still make sense in that setting?

Dan Morris (agentmorris@gmail.com)

2025-07-08 14:14:51

Atul's intuition is correct: an ROC curve is at least technically appropriate for the typical MegaDetector use case where it's being used as an image classifier. Box location doesn't matter in most scenarios, and in fact a "true positive" where MegaDetector totally missed the animal and put a high-confidence box on a tree is exactly the same for this analysis as a "true positive" where MD correctly boxed the animal.

A couple of caveats, though:

Before making an ROC curve, I would encourage your collaborators to answer the question: what percentage of your animals are you comfortable missing? The portion of the ROC curve with recall lower than that value is irrelevant, to the point where the curve is probably distracting, and if you want to systematize the selection of a threshold, I would just pick the maximum threshold that meets the recall requirement.

More specifically, I would encourage your collaborators to answer the question: what percentage of your small (in terms of pixels, not in terms of body size) animals are you comfortable missing? This may lead to a lower threshold, and is a reminder that the ROC curve will be dominated by common species, which are often the easy species. If you are doing a spotted skunk survey, an ROC curve dominated by performance on deer isn't going to help you choose a threshold.

As Beni mentioned, this is extremely specific to individual datasets, primarily based on the size and visibility of animals, and whether the camera is in a "typical" camera orientation. So there isn't really an "ROC curve for MegaDetector".

With all that said, if I haven't successfully talked you out of making an ROC curve 🙂, I do this pretty often using this module:

https://megadetector.readthedocs.io/en/latest/postprocessing.html#module-megadetector.postprocessing.postprocess_batch_results

99% of the time I'm using that in scenarios where ground truth isn't available, in which case it makes pages like this:

https://lila.science/public/snapshot_safari_public/snapshot-safari_KAR_mdv5a.0.0_results.filtered_rde_0.150_0.850_15_0.200

...but it can also produce P/R curves (at the image level) when GT is available, like this:

https://lila.science/public/snapshot_safari_public/snapshot-safari_KAR_mdv5a.0.0_results.filtered_rde_0.150_0.850_15_0.200_gt

Note in that case that most of the "false negatives" are actually labeling issues. 🙂 This is an example of why the module doesn't just make a P/R curve, it also samples images to help you visualize what the correct/incorrect samples look like, which can be more informative than the curve itself.

Eugene Galaxy (eugene@animaldetect.com)

2025-07-10 03:43:31

Hey @Atul Ingle. We are using MegaDetector as an empty-image filtering step in our Animal Detect platform. It might not be technically what you are after, ROC/curve, but perhaps it can be visually practical to observe the detection rate when moving the threshold slider. Check the short video demo I just captured. The images with animals/humans/vehicles are in the left column, empty ones on the right.
If this is useful to your team, I would be more than happy to chat over a call and see if we can find some work together! 🙂 The platform is 100% developer/controlled by me and @Hugo so there is no upper-limit to what we can implement and adapt there.

Cheers!

Atul Ingle (ingle@uwalumni.com)

2025-07-10 11:14:01

This is all very helpful! Eugene, this slider bar type tool looks really cool and is exactly what prompted me to think about the ROC curve. My collaborators were unsure where to set the slider at. It's more intuitive to have a slider that controls, say, detection probability, and then they must live with whatever false alarm rate that operating point would correspond to --- systematize the selection of the threshold value, as Dan puts it!

Atul Ingle (ingle@uwalumni.com)

2025-07-10 11:17:37

Dan has convinced me that drawing the entire ROC curve is neither necessary nor very useful here (other than, maybe, satisfying a purely academic itch that I have!)

Esther Rolf (esther_rolf@berkeley.edu)

2025-07-08 14:51:36

📣 We have an opening for a postdoctoral fellow in AI/ML for Environmental Data Science at CU Boulder! Hosted by me and Claire Monteleoni in Computer Science, with option to have an additional mentor from the Environmental Data Science Innovation and Impact Lab (ESIIL). Reach out if you're interested and please forward to anyone you know who might be! We start reviewing applications July 15, but ping me if that deadline is an issue.

jobs.colorado.edu

Machine Learning and Environment Postdoctoral Fellow

Machine Learning and Environment Postdoctoral Fellow at created 11-Oct-2024

Original URL: https://jobs.colorado.edu/jobs/JobDetail/Machine-Learning-and-Environment-Postdoctoral-Fellow/59901

💚 Burak Ekim, Robin Zbinden, Carlos Rodríguez - Pardo, Lily Xu, Owen Martin, Bernie Boscoe, Eugene Galaxy, Jonathan Sauder, Shir Bar, Ghjulia Sialelli, Sara Beery, Aleksis Pirinen

Dan Morris (agentmorris@gmail.com)

2025-07-10 09:53:17

New classifier available for MegaDetector crops... but this time for vehicle crops (I love pictures of animals, but other things appear in camera trap images too!):

https://github.com/agentmorris/camera-trap-vehicle-classifier

Specifically, it classifies vehicle crops into four categories: car/truck, bicycle, motorbike, and quad. This classifier was developed with a WWF group studying the impact of recreation on wildlife in mixed-use areas. Thanks to Robin Naidoo from WWF for a very efficient collaboration to get this classifier up and running.

I'm hoping this can plant a seed for an expanded set of vehicle classifiers, so if you have camera trap data in which you have labeled vehicles into finer-grained categories than just "vehicle", let me know.

Other fun facts:

My initial instinct was to use off-the-shelf ImageNet-trained classifiers, since ImageNet includes a variety of vehicle categories, and they weren't terrible, but they weren't great; my best guess is that the degree of motion blur in camera trap images was beyond what's included in ImageNet.

Apropos of other threads on this channel, this project was 100x faster because of AI. Actually more like infinity times faster: I've done the data wrangling part of this project a zillion times, but the image classification part is outside of my normal toolset and I probably wouldn't have taken this project on if AI didn't basically write all the AI code. AI also made it much easier to try a variety of ImageNet classifiers before going down the path of training a custom model.

🙌 murilo, Atul Ingle, Jennifer, Shir Bar, Mitchell Rogers, Jaycee Choi, Negar Sadrzadeh, Piotr Tynecki

:the_horns: Paul, Hugo, Malte Pedersen, Sara Beery, Piotr Tynecki

🎉 Eugene Galaxy, Yumi Arimitsu, Tiziana Gelmi Candusso, Piotr Tynecki

😎 Matt Hron, Rachael Laidlaw, Brooke Goodman, Piotr Tynecki

🔥 Hugo, Burak Ekim, Josh Seltzer, Tiziana Gelmi Candusso, Piotr Tynecki

🚲 Jon Van Oast, Piotr Tynecki

🙌:skin_tone_5: Prabath Gunawardane

Dan Morris (agentmorris@gmail.com)

2025-07-14 23:25:50

Another fun vibe coding party today! We had a diverse set of backgrounds again - first-time vibe coders, people edging toward fluency in AI assisted coding, our first visit from someone who works for a vibe coding platform company - but everyone was connected to AI4C in some way.

Watching @Josh Seltzer's Copilot workflow helped me recognize some of the silly ergonomic reasons I wasn't getting along well with in-IDE AI tools: I'm super-dependent on having an IPython console, and it just takes up so much space that every time I added another giant widget to my screen while I was trying in-IDE AI tools, I had a bad experience, but the secret was just to float one or the other out to a different monitor. Voila, now I can have AI in my IDE.

And for my vibe coding project, I made "The LILA Game"...

https://dmorris.net/lila-game/
https://github.com/agentmorris/lila-game

It's running on the same tiny VM that hosts a bunch of other things, so I don't really know what will happen if folks start visiting it, but the idea is that AI helped me populate a proper DB with all the camera trap images from LILA, then the game selects a taxon (usually a species) at random, and shows you N random sequences containing that taxon, and you have to guess what it is. There's fun autocomplete, and - unless I hit the limit of the Gemini free tier before anyone gets to try it - it tells you AI-generated fun facts about each animal after each round, and you can ask for AI-generated hints. Lots more I'd like to do, but I learned a bunch of DB stuff, a bunch of using-LLMs-from-code stuff, and just practiced more general prompting skills. Getting there!

This LILA game would have taken me precisely seventy-gazillion hours a year ago.

Others who attended today, feel free to share screenshots and/or lessons learned on this thread!

🙌 Mitchell Rogers, Atul Ingle, Andy Huynh, Shir Bar, Irina Tolkova, Lily Xu, Jaycee Choi, Josh Seltzer, Matt Hron, Carly Batist

🎉 Eugene Galaxy, Sara Beery, Jennifer, Aamir Ahmad, Alan Stenhouse

😎 Jon Van Oast

🙌:skin_tone_5: Prabath Gunawardane

Yumi Arimitsu (yarimitsu@gmail.com)

2025-07-15 17:19:29

Thanks @Dan Morris for hosting another fun vibe coding party. I made some stuff and was glad to see what others are working on. It's like a workshop for building shiny things. 🎉 I was working on Chomp Watch, which of course still needs lots of work now, but I found it kind of eerie that this project summary seems more along the lines of what I was thinking than what I told it to make. It also made up a whole research team, listing Dr. Sarah Marine, Captain Joe Fisher, and Maria Outreach as key contacts! 😂 Here's a Chomp Watch V1 screenshot, and for Dan, an 'animal pic' of that magnificent Black-footed Albatross from last weekend in southeast Alaska.

🦆 Dan Morris, Prabath Gunawardane

Prabath Gunawardane (prabathg@gmail.com)

2025-07-16 20:43:29

Got sidetracked from posting this update, but I had a great time at the vibe coding session. Thanks for organizing Dan! It was very informative to hear everyone's experience with Claude Code, Gemini CLI, Github copilot, Cursor & Lovable.

From my side, I was a bit ambitious with trying to get Claude Code to develop an 'Amoeba Sim game' which would allow players to add custom species with scripted behavior into a 2d game world. This had separate front-end & backend code, database connectivity, a custom script compiler (for custom behavior coding by players), and real-time connectivity via web sockets. Given all the moving parts I'm pretty impressed with the progress Claude Code made, and in particular the tooling and testing scripts it came up with.

Even though the code is very much in draft state, I uploaded it to github to create a checkpoint for sharing and tinkering with later : https://github.com/prabathgu/ai-amoebas
Screenshot:

🦠 Dan Morris, Josh Seltzer, Yumi Arimitsu

Aamir Ahmad (aamir.ahmad@ifr.uni-stuttgart.de)

2025-07-15 07:06:43

Hi folks,

TLDR: I would like to share one of my group's latest "product" -- anyone caring about animal re-identification (for animals with patterns) might like it. The method is good for drone-deployment or any edge-device deployment. Find it here: https://github.com/robot-perception-group/Real-time_Animal_reID

Some more details:

The work is called RAPID -- Real-time Animal reID on edge devices. It is a lightweight and high-speed individual re-identification algorithm that distinguishes uniquely patterned animals within the same species faster and with lower computational demand than existing open-source methods. RAPID runs only on a CPU and can ID animal species with an accuracy (~95%) at-par with most state-of-the-art deep learning-based methods, but at blazing speeds (40-60 hz) on a laptop using a CPU only without using a GPU! It light-weight so that you can deploy it on an edge-device with a small CPU like Texas-Instrumets tda4vm and obtain near-real-time (10 hz) identification.

You can ping me for more info.

Code: https://github.com/robot-perception-group/Real-time_Animal_reID
Pre-print: https://www.biorxiv.org/content/10.1101/2025.07.07.663143v1
Video: https://youtu.be/xGas3IlLrVo

bioRxiv

RAPID: Real-time Animal Pattern re-Identification on edge Devices

Automatic re-identification of animals has significant potential for addressing pressing challenges of our times via population monitoring, individual health assessment or detailed behavior analysis. Although numerous computer vision based solutions have been proposed and many achieve high accuracy, those still remain unsuitable for real-time analysis and deployment on edge devices. Here, we introduce RAPID, an open-source1 algorithm for re-identifying patterned animal images at a rate exceeding 40-60 queries per second on standard PC or Laptop and over 10 queries per second on an inexpensive off-the-shelf edge device. RAPID operates efficiently in computationally restricted environments, relying solely on CPU, leaving GPU resources available for other tasks, all while maintaining state-ofthe-art accuracy. Our approach leverages SIFT descriptors, which continue to demonstrate competitive robustness and accuracy against recent traditional and deep-learning based methods. To counterpoise SIFT’s primary limitations its high-dimensional feature vectors and the associated computational costs - we incorporate recent advancements in vector similarity search and construct a database of feature vectors rather than database images, further accelerating the search process. We evaluate our method, RAPID, on public re-identification datasets; additionally propose a RAPID-based tool, FalseTagFinder, for cleaning benchmark dataset labels and as a demonstration, provide corrections for the StripeSpotter dataset. ### Competing Interest Statement The authors have declared no competing interest. Ministry of Culture and Innovation of Hungary from the National Research, Development and Innovation Fund, Financed under the KDP-2023 Funding scheme, c2264929 project WildCap-Health funded through <a href="http://TUE.AI">TUE.AI</a> Agile Research Funds 2024

Original URL: https://www.biorxiv.org/content/10.1101/2025.07.07.663143v1

YouTube

András Zábó (https://www.youtube.com/@zaband11)

RAPID: Real-time Animal reID on edge devices

Original URL: https://youtu.be/xGas3IlLrVo

❤️ Sara Beery, Timm Haucke, Josh Seltzer, Navodita Mathur, Chaim

🙌 Dan Morris, Jenna Kline

🐅 Dan Morris

Matthias Zuerl (matthias.zuerl@fau.de)

2025-07-15 07:16:52

nice!

🙏 Aamir Ahmad

Blair Costelloe (blaircostelloe@gmail.com)

2025-07-15 07:24:31

Super cool!!

🙏 Aamir Ahmad

Carly Batist (carly.h.batist@gmail.com)

2025-07-18 05:12:08

Very cool! Would recommend chatting with the folks at WildMe/WildBook about this and opps for collab’ing!

👍 Aamir Ahmad

Ghjulia Sialelli (ghjulia.sialelli@inf.ethz.ch)

2025-07-15 07:56:19

Hi everyone! Using this platform to quickly advertise an upcoming event, the AI + Environment Summit 2025. This full-day event will be part of the Zürich AI Festival, but with a focus on exploring how AI research and innovation can contribute to sustainable solutions. There will be inspiring talks (including from @Sara Beery!) and panel discussions, as well as a poster session (so don't hesitate to send in some of your work, the deadline is Aug. 1st). We will be providing amazing coffee and great food all day long ☕

🔗 : https://ai-environment-summit.com/
📆 : October 1st, 2025
📍 : Innovation Park Zürich (Switzerland)

❤️ Sara Beery, Samantha Biegel, Robin Zbinden, Josh Seltzer, Sarthak Arora, Viktor Domazetoski, Andy Huynh, Owen Martin, Jonathan Sauder, Navodita Mathur, Lukas Drees, Michael Wethington, Leonie Baier, Jason Holmberg (Wild Me), Lily Xu, Carly Batist, Valerie, Olivier Dietrich, Alan Stenhouse, Johannes Dollinger

☕ Lily Xu

Lily Xu (lily.x@columbia.edu)

2025-07-17 08:54:50

Vouching for the quality of the coffee! 💚 And the quality of the entire event — thank you Ghjulia and team for running this event for the third time.

❤️ Ghjulia Sialelli, Johannes Dollinger

David Russell (davidrussell327@gmail.com)

2025-07-16 12:14:17

Hey everyone! We've started holding office hours for the Open Forest Observatory to help people use our open source tools and data for drone-based forest mapping. Or to just chat about this field. They are held on the first Wednesday of the month at 4PM PST and the third Wednesday of the month at 9AM PST. Both occur at this zoom link and one is happening now.

openforestobservatory.org

Open Forest Observatory

Drone-derived forest maps, mapping tools, and field reference data

Original URL: https://openforestobservatory.org/

🎉 Timm Haucke, Amritha Pallavoor, Shir Bar, Ian McGregor

❤️ Timm Haucke, Declan, Lily Xu, Alexander Merdian-Tarko, Carly Batist, Alan Stenhouse

Dan Sheldon (sheldon@cs.umass.edu)

2025-07-16 16:09:22

Hi folks, FYI, we are running a special track for social impact at AAAI. It covers a broad range of topics including climate and environmental sustainability:

We are pleased to announce the continuation of the special track on Artificial Intelligence for Social Impact (AISI) at the AAAI 2026 conference. This track recognizes that high-quality research conducted in social impact domains often leads to papers that differ from traditional AAAI submissions and invites authors to submit papers that delve deeper into the challenges and opportunities of using AI to solve substantive problems in social impact domains and showcase the social benefits of artificial intelligence.

AISI uses special review criteria and keywords, which can be found in the call for papers.

Call for papers and detailed timeline: link
Submission deadline: July 25 (abstract), August 1 (paper)

High-quality reviews are critical for making the track what it is. If you would be willing to contribute reviews, and you have experience with AI for social impact research and academic reviewing, please complete this reviewer interest form.

Thanks,
Andrew Perrault (The Ohio State University)
Dan Sheldon (University of Massachusetts Amherst)
Program Co-Chairs, AAAI 2026 Special Track on AI for Social Impact

AAAI

Call for the Special Track on AI for Social Impact - AAAI

Est. reading time

6 minutes

Original URL: https://aaai.org/conference/aaai/aaai-26/aisi-call/

❤️ Gustavo Perez, Jason Holmberg (Wild Me), Alexander Merdian-Tarko, Sara Beery

Dan Morris (agentmorris@gmail.com)

2025-07-16 17:05:42

New dataset on LILA, thanks to @Cara Appel:

https://lila.science/datasets/nkhotakota-camera-traps/

This is >300k images from Malawi, with 46 labeled taxa.

As always, when releasing new datasets, I like to remind everyone that individual datasets are a figment of our collective imagination; this dataset is already mapped into the common taxonomy used across LILA and is included in the Giant CSV File:

https://lila.science/taxonomy-mapping-for-camera-trap-data-sets/

Thanks to Cara and her collaborators for contributing this data, and congratulations on having the associated paper accepted!

😎 Peter Bull, Justin Kay, Timm Haucke, Sam Lapp, Jason Holmberg (Wild Me), Sara Beery

🙌 Kaitlyn Gaynor, Peter Bull, Justin Kay, Timm Haucke, Shir Bar, Jason Holmberg (Wild Me), Aamir Ahmad, Eugene Galaxy, Alexander Merdian-Tarko, Elizabeth Campolongo

🎉 Jon Van Oast, Peter Bull, Justin Kay, Timm Haucke, Matt Weldy, Atul Ingle, Jason Holmberg (Wild Me), Jess Tam, Hayley Smith, Carly Batist, Jennifer

Alice Wilby-Ward (alicewilbyward@gmail.com)

2025-07-17 11:08:46

Heya, my name is Alice and I am an undergraduate student studying marine biology and was introduced to this group by @Sergei Nozdrenkov. Separate from university, I am currently exploring the potential of Open Ai and Gemini AI to automate UK seahorse spot pattern recognition to reduce the manual effort and time it takes, along with removing the need to tag seahorse individuals with FLOY tags in situ. I have managed to get Gemini to identify the seahorse individual using a bounding box as visualisation through my entire image dataset, along with a coding block that can compare pairs of images of the same and different seahorse individuals by using a veto-based analysis prompt. This is still very much a work in progress, and I have encountered a bit of a roadblock, as no matter how precise the prompt is for the AI to identify the seahorse individuals' spot pattern markings using bounding box visualisation, it isn't correctly identifying or accurately outputting the image with the Regions of Interest (ROI's) encircled. I am now going to go down the route of using Wildbook AI and try and adapt it to this project, but I still think incorporating AI into this project would be cool. If anyone has any suggestions on where to take this project next or has any recommendations of what tools to try out, it would be very much appreciated! Thank you :)

Dan Morris (agentmorris@gmail.com)

2025-07-17 16:57:11

My experience has also been that right now, frontier models still struggle with precise localization tasks in complicated images. They're great at things like parsing user interface screenshots, but natural images remain difficult. I have no doubt they will get there, because things improve faster than anyone can keep up, but right now I think you're right that this is tough for frontier models.

That said, I'm always interested in experimenting with prompts, so, would you be able to share a couple of sample images here?

Also in addition to Wildbook, you may want to checkout the wildlife-tools package and the associated MegaDescriptor model for animal re-ID (I haven't used either, but they sit in the space you're looking for):

https://github.com/WildlifeDatasets/wildlife-tools
https://huggingface.co/BVRA/MegaDescriptor-L-384

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2025-07-17 21:51:04

MiewID v3 developed by the Wildbook team includes leafy and weedy seadragon individuals in the training data in addition to 63 other species. No guarantee, but it might have some zero shot potential, and we also have Hotspotter in Wildbook for just basic SIFT matching. Happy to talk to you about running some tests with both.

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2025-07-17 21:51:04

https://huggingface.co/conservationxlabs/miewid-msv3

huggingface.co

conservationxlabs/miewid-msv3 · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Original URL: https://huggingface.co/conservationxlabs/miewid-msv3

Alice Wilby-Ward (alicewilbyward@gmail.com)

2025-07-18 06:37:18

@Dan Morris Yes, I definitely agree with you. AI model will just continue becoming smarter and smarter with the strengthened ability for more highly precise specialised tasks.

Alice Wilby-Ward (alicewilbyward@gmail.com)

2025-07-18 06:40:43

Here is an example of the most accurate prompt (including precision spot pattern bounding box prompts [which were unsuccessful due to the inaccuracy]). I have also been working on a seahorse comparison model where Gemini is prompted to compare a pair of seahorse images (different or same individual) and output the points of interest and a confidence percentage of how sure it is that the photo is of the same individual.

Alice Wilby-Ward (alicewilbyward@gmail.com)

2025-07-18 06:41:22

Thread Reply:

Alice Wilby-Ward (alicewilbyward@gmail.com)

2025-07-18 06:47:59

@Jason Holmberg (Wild Me) This looks like a fantastic possibility. I would be very happy to have a chat about the potential use of this. I saw that WildMe uses SIFT descriptors and Hotspotter and think this could be the perfect solution too. Considering Weedy Sea dragons are in the same family and have much more complex dermal appendages this could definitely work for the less elaborate UK seahorse species.

Atul Ingle (ingle@uwalumni.com)

2025-07-18 21:21:45

@Alice Wilby-Ward Do you have any insights on designing the best text prompt?
I can think of a few different ways...

Create a laundry list style description of what you were trying to segment and just shove a really long prompt into the model

Give smaller prompts, one at a time, to guide it towards the right answer

Alice Wilby-Ward (alicewilbyward@gmail.com)

2025-07-21 05:01:25

hey @Atul Ingle I found that the laundry list style approach worked best. Organised into different subsections (for your understanding and tracking). Smaller prompts one by one was quite long winded but still was effective if you chose the correct prompt. Both approaches just take a lot of tinkering to get right. Hope this helps!

👍 Atul Ingle

Jorge Lizarazo (jorge.lizarazo.b@gmail.com)

2025-07-20 19:57:10

Hi everyone,

I recently watched a talk by Aza Raskin where he discusses the idea of using generative models to explore communication with whales. While the conversation wasn’t deeply technical, it sparked a line of thought I’ve been quietly exploring: can we use generative AI to better understand, replicate, and eventually "translate" the communicative structures in animal vocalisations?
I work primarily in conservation and ecology, and I’ve applied several predictive models (CNNs, LSTMs, and some transformer-based architectures with transfer learning). However, I’ve never trained a generative model myself. Still, I’ve been sketching a conceptual pipeline that could link bioacoustics and behavioural ecology.
The idea is now something like start with a large, unlabeled dataset of an animal vocalisations (e.g. bird songs, whale calls, primate vocalisations). Use unsupervised learning (e.g. embeddings + UMAP + HDBSCAN or even kmens, don’t know) to group vocalisations by structure or spectral similarity.
Then, based on field knowledge or ethological observations, manually label some clusters with possible communicative functions (e.g. alarm, contact, courtship). Only if they make sense. Use these labelled clusters to fine-tune a generative model (like SpecGAN, AudioLDM, or even an autoregressive model like WaveNet or VALL-E) to create synthetic sounds conditioned on function.
But… then explore whether this can help us simulate or even engage in meaningful communicative loops, perhaps as a tool for playback experiments or for probing animal perception in a controlled field experiment or exsitu one.
This is still a very very very early-stage idea, more like a sketch hahaha. but I’m curious whether anyone in this community has tried something similar. Not necessarily for whales—any species or sound system would be relevant.

Levi Cai (levi@levicai.com)

2025-07-20 22:32:51

In short, yes, there are a couple communities looking into these now. Some examples I'm aware of:

https://www.projectceti.org/

https://www.earthspecies.org/ (I guess this is Aza's effort)

https://www.wilddolphinproject.org/our-research/dolphin-communication/

https://www.elephantvoices.org/elephant-ethogram.html

https://moodeng.media.mit.edu/#tracks

https://blog.google/technology/ai/dolphingemma/

Many of these groups (several are part of this slack :P) are working with academics to do both supervised and unsupervised/self-supervised strategies for finding language structure, roughly following some of the procedures you're interested in.

projectceti.org

Project CETI •-- Home

CETI is a nonprofit organization applying advanced machine learning and state-of-the-art robotics to listen to and translate the communication of sperm whales.

Original URL: https://www.projectceti.org/

elephantvoices.org

The Elephant Ethogram

Introduces the elephant ethogram: A Library of African Elephant Behavior

Original URL: https://www.elephantvoices.org/elephant-ethogram.html

Marzi (marziyeh.mohammadi@fau.de)

2025-07-21 03:37:25

Hi, cool idea. It depends on your data type. Generative models are all about patterns—if there is a meaningful pattern in those signals, you can detect and even classify them based on their similarity to create a dictionary. Apply a model and analyze the results to see if any meaningful patterns are detected. If so, explore more suitable models based on your results and analysis.

Cody Kupferschmidt (kupfersc@uoguelph.ca)

2025-07-22 12:36:09

I was just looking through the iNaturalist API terms of use and noticed that training AI models is prohibited. https://www.inaturalist.org/pages/terms

Are there any other data sources I could use to build a tool for lookup or prediction of local native plant species in Canada and the US?

iNaturalist

Original URL: https://www.inaturalist.org/pages/terms

Oisin Mac Aodha (macaodha@caltech.edu)

2025-07-22 12:43:31

Looks like it depends on your use case:
"Users may not use any iNaturalist data for training artificial intelligence, machine learning models, large language models, or similar networks, algorithms, or systems for commercial purposes."

💯 Declan, Nicole Lopez

Julia Chae (chaenayo@mit.edu)

2025-07-22 21:22:15

Thank you so much everyone for your support in CV4E ICCV Workshop's proceedings track! Our non-archival track is now OPEN and ACCEPTING SUBMISSIONS, including previously published submissions (within the previous 6 months). The updated deadline is August 15th, 2025. The submission link is at https://openreview.net/group?id=thecvf.com/ICCV/2025/Workshop/CV4E_Non-Archival#tab-recent-activity

openreview.net

ICCV 2025 Workshop CV4E Non-Archival

Welcome to the OpenReview homepage for ICCV 2025 Workshop CV4E Non-Archival

Original URL: https://openreview.net/group?id=thecvf.com/ICCV/2025/Workshop/CV4E_Non-Archival#tab-recent-activity

❤️ Navodita Mathur, Elena Sierra, Yuyan Chen, Timm Haucke, Andy Huynh, Andy Viet Huynh, Rupa Kurinchi-Vendhan, Oisin Mac Aodha, Robin Zbinden, S M Rayeed

🎉 Jon Van Oast, Andy Huynh, Andy Viet Huynh, Rupa Kurinchi-Vendhan, Robin Zbinden

Alex Logan (alex@cecil.earth)

2025-07-22 21:41:52

Hey everyone - I am Alex Logan a co-founder and ceo of cecil (https://docs.cecil.earth/) I was kindly invited by @Dan Morris

Cecil is my first personal journey into conversation and nature. I have a background in software and tech (but fortunately supported by an amazing science team at Cecil).

The first product we built at Cecil almost 4.5 years ago was a software platform for managing nature based projects in one place (e.g. afforestation). While building that product we found it extremely difficult to get access to datasets for supporting our customers with eligibility assessments, monitor land use change over time, and more. There was no good API documentation, it was hard to know what datasets to use in different regions, and the pricing / licensing was often extremely prohibitive.

So we pivoted to build a better experience for others building in this domain and make nature data accessible. We're still early on this journey but you can see our progress on our docs page: https://docs.cecil.earth/ and follow our newsletter here: https://newsletter.cecil.earth/

I am curious about what datasets teams are using successfully, licensing challenges, dataset evaluation, and understanding any challenges or gaps with datasets.

👋 Dan Morris, Timm Haucke, Jon Van Oast, Andy Huynh, Navodita Mathur, Alessandra Sellini, Carly Batist, Eugene Galaxy, Yumi Arimitsu, Asa DeHaan, Alexander Merdian-Tarko, Darshana Salvi

🎉 Jon Van Oast, Andy Huynh, Jennifer, Shir Bar, Leonie Baier, Alessandra Sellini, Joe Ferdinando

‼️ Tyler Marino, Asa DeHaan

Alice Heiman (aheiman@stanford.edu)

2025-07-24 00:42:03

Hi everyone! 👋

My name is Alice Heiman, and I’m a current student at Stanford studying Computer Science and Sustainability. I’m part of the planning committee for Stanford’s Annual Climate Week, now entering its second year following a successful launch last spring!

Our inaugural Climate Week brought together over 600 students across 40+ events featuring student groups, faculty, alumni, and industry leaders from 7+ schools across campus. We’re excited to build on that momentum this year from October 20–26, 2025, and would love to see some events on AI+Conservation! :smilingfacewith3hearts:

If this sounds interesting and you would like to partner/co-host an event, feel free to send me an email at aheiman@stanford.edu! I look forward to collaborating!

🌲 Andy Viet Huynh, Michelle L, murilo, Nicole Lopez

👏 Andy Viet Huynh, FANQI Z, Eugene Galaxy, murilo, Darshana Salvi, Sofía Miñano

☺️ Laura Chrobak

Ellie Bowler (eller@bas.ac.uk)

2025-07-28 06:29:56

Hi Anna, i work in a team at the British Antarctic Survey developing satellite monitoring for seabirds and some members might be interested in this! A few are already planning to attend the conference. If you're able to message me your contact details would it be ok to put you in touch?

Anna Vallery (annacvallery@gmail.com)

2025-07-28 11:03:25

Hi @Ellie Bowler, that would be amazing! My email is annacvallery@gmail.com. Looking forward to connecting with members of your team!

Ellie Bowler (eller@bas.ac.uk)

2025-07-28 11:19:34

Thanks!

Yumi Arimitsu (yarimitsu@gmail.com)

2025-07-28 13:58:10

This will be a great session Anna! We're using our time lapse image data to assess disturbance rates at colonies too. I don't know if I'll be able to make WSC this time, unfortunately... so much is still uncertain. But please keep me in mind to help if hybrid options become available. :)

❤️ Anna Vallery

Anna Vallery (annacvallery@gmail.com)

2025-07-28 14:04:10

@Yumi Arimitsu So cool! I will absolutely keep you in the loop!

Tom Walker (tom@cecil.earth)

2025-07-31 04:42:42

Hi everyone! 👋

My name is Tom Walker. I'm an ecosystem ecologist and biogeochemist at heart and spent a decade or so as an academic before finally throwing in the towel and heading out into the "real" (?!) world. I'm now working alongside @Alex Logan as head of science at Cecil.

I'm passionate about making nature and environmental data accessible, interoperable, and operational at scale. There are loads of components to this, including metadata, benchmarking, standardising protocols, integrating new tech, etc., so I'm really excited to support and learn from this community to ensure the nature data sector matures and scales in a rigorous, thoughtful, and conscientious way.

Thanks for the invitation to get stuck in!

🙌 Nuno Sa, Shir Bar, Sam Lapp, Leonie Baier, Anna Vallery, Carly Batist

👋 Lukas Picek, Sara Beery, Xiangteng He, Irina Tolkova, Alexander Merdian-Tarko, Meredith Palmer

Xiangteng He (xiangteng.he@ubc.ca)

2025-07-31 11:16:44

Hi everyone 👋 I'm Xiangteng He, a Postdoctoral Researcher at the University of British Columbia.
My work focuses on Fine-grained Analysis on Multimodal Data, Vision-Language Models and their application. Thrilled to join this community of 3100+ global experts! Would love to connect with others working on computer vision for wildlife monitoring / ML in biodiversity conservation. Feel free to DM me to brainstorm ideas or explore collaborations 🌍
https://hexiangteng.github.io/

Thank you, @Sara Beery, for building such a vibrant space at the intersection of AI and conservation.

👋 Youssef Bayouli, Oisin Mac Aodha, Jon Van Oast, Irina Tolkova, Mamoon, Sara Beery, Andy Huynh, Alexander Merdian-Tarko, Shir Bar, gyeongsu, Santiago, Tim Zhou, Jason Holmberg (Wild Me), Roberta Hunt, Hiroto

:flag_ca: Tim Zhou, Jason Holmberg (Wild Me)

Santiago (santiagovillarreal60@gmail.com)

2025-07-31 14:43:11

Hi everyone!
I’m Santiago Villarreal, a Master of Engineering student in ECE at the University of Manitoba, focusing on AI and computer vision.smart_cat
I’m passionate about nature and love applying technology to help protect our environment. Right now, I’m building a drone-based vision system for smarter wildfire response, and I’m excited to learn and contribute!

👋 Xiangteng He, Alexander Merdian-Tarko, Shir Bar, gyeongsu, Jason Holmberg (Wild Me), Sofía Miñano, Daniel Olhagaray

Otto Brookes (otto.brookes@bristol.ac.uk)

2025-08-01 05:54:12

2025 Workshop on Camera Traps, AI, and Ecology - Registration Now Open

We are pleased to announce that registration is now open for the 2025 Workshop on Camera Traps, AI, and Ecology, to be held September 9–10 as a hybrid event in Seattle, USA, hosted by AI2.

The workshop will feature keynote talks from:

Prof. Tanya Berger-Wolf – Director of the Translational Data Analytics Institute and Professor at The Ohio State University

Prof. Iain Couzin – Director of the Max Planck Institute of Animal Behavior and Professor at the University of Konstanz

Dr. Dan Morris – Senior Staff Research Scientist, Google AI for Nature

Prof. Ulrik Pagh Schultz Lundquist – Head of the Drone Center, University of Southern Denmark

Please note that in-person attendance is limited to 30 participants, and early registration is strongly recommended. Online attendance will not be limited.

🔗 For more information and to register, visit: camtrapai.github.io

📅 The full program will be announced shortly.

camtrapai.github.io

Camera Traps, AI, and Ecology - International Workshop Archive

Camera Traps, AI, and Ecology - International Workshop Series

Original URL: https://camtrapai.github.io/

🙌 Carly Batist, Dan Morris, Sara Beery, Hugo, Timm Haucke, Sofía Miñano, Talia Speaker, Paul Bodesheim

👍 Suman Ghosh

🙌:skin_tone_3: Alan Stenhouse

Dan Morris (agentmorris@gmail.com)

2025-08-02 22:03:40

A couple of weeks ago I posted about a model I worked on with a WWF group who wanted to add finer-grained classifications to the vehicles in their camera trap images, to more precisely assess the impact of recreation types on wildlife. They wanted to do the same for horses, specifically to assess how many horses on trails were packhorses vs. saddlehorses (a word I just learned), which also required adding a "free-ranging horse" category (i.e., neither of the above). So, new classifier available:

https://github.com/agentmorris/camera-trap-horse-classifier

We run this on the MegaDetector crops that SpeciesNet says are horses. A three-model cascade!!! Maybe next we can classify the shoe brands being worn by the riders on the saddlehorses, then we could get to a four-model cascade.

👍 Aamir Ahmad, Alexander Merdian-Tarko, Hugo, Bowen Zhang, Rebecca Kleinberger, PhD, Peter Bull, Jason Holmberg (Wild Me), Xiangteng He, Eugene Galaxy, Christoph Praschl, Talia Speaker

🎠 Atul Ingle, Paul, Mitchell Rogers, Bowen Zhang, Rebecca Kleinberger, PhD, Peter Bull, Jason Holmberg (Wild Me), Yumi Arimitsu, Tom Walker, Joao Alves

🏇 Malte Pedersen, Bowen Zhang, Peter Bull, Jason Holmberg (Wild Me), PG

👍:skin_tone_3: Alan Stenhouse

PG (premsgill@gmail.com)

2025-08-07 04:02:39

“animal: 98% packhorse: 100.0%” a small, but not zero, chance we’ve countered a non animal packhorse 👽

Eugene Galaxy (eugene@animaldetect.com)

2025-08-04 02:47:15

Hey everyone!

We made a free online tool to find animals on single images.
Link is here: https://www.animaldetect.com/wildlife-detector

It works very simple: drop an image -> get animal (or human/vehicles) detected.

No sign-up needed.

You can share images via URL 🔗 or Download ⏬ them.

Images you upload are not publicly shown to others, unless you share the link.

The tool uses same AI models as our Animal Detect platform.
-> MegaDetector for detection
-> SpeciesNet for classification

"What's the catch?"

There isn't one, really. While this tool is a good "demo" of what our main bulk image processing platform can do, we have enough spare resources to let this tool live online and free forever.

How the idea was born?
We noticed that a lot of our new users are creating a project (data deployment/session) and go through the whole flow just to test one image. While this works, it is TOO MUCH commitment for just one image. Animal Detect shines best at bulk data processing == large volumes. To get gigabytes of images in-and-out.
Hence we made a 1-image quick tool. Since it uses the same AI-stack -> you can expect the bulk processing on the platform to yield the same result.
Moreover, we got multiple requests for an API. Think of this tool as a "visual demo" of API that is coming soon.

P.S. Don't try to upload anything shady, we have NSFW filters! 😎
P.S2 Feedback is always very welcomed!

👍 Aamir Ahmad, Nuno Sa, Lasha Otarashvili, Dan Morris, Tiziana Gelmi Candusso, Jason Holmberg (Wild Me), Malte Pedersen, Alexander Merdian-Tarko

:the_horns: Hugo, Tiziana Gelmi Candusso, Jason Holmberg (Wild Me)

❤️ Victor Anton

👍:skin_tone_3: Alan Stenhouse

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:04:03

excellent, i'll give it a go

🙌 Eugene Galaxy

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:05:50

perfect hehe

❤️ Eugene Galaxy

Eugene Galaxy (eugene@animaldetect.com)

2025-08-04 04:05:53

Don't forget to specify country where an image was taken, it will help to classify specie more precisely.

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:06:41

completely didnt do that but it worked very well

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:06:56

just used some example from the internet

Eugene Galaxy (eugene@animaldetect.com)

2025-08-04 04:07:04

It's on USA by default which is best represented in the AI model. So good chances are - it did the trick! 😄

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:07:21

the source was azores

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:07:39

but its an easy one "domestic cat"

😺 Jon Van Oast

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:07:58

still as a tester this is very good

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:09:52

this is very interesting app to be honest

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:10:10

so if i want to use it for more data i would have to go into a paid plan?

Eugene Galaxy (eugene@animaldetect.com)

2025-08-04 04:10:58

In theory yes. However, each new account receives 1GB of data for free, so I would recommend to create an account and try it out!

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:11:48

very cool

Eugene Galaxy (eugene@animaldetect.com)

2025-08-04 04:12:03

Otherwise, if you are a part of a wildlife organisation and needs something more/custom/extra on top of the processing, just send me a message, we do collaborate with organisations!

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:12:13

i dont work directly on the ground but this could be of interest

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:12:17

to some people who do

💚 Eugene Galaxy

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:13:01

cool stuff 🙂 i'll share around

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:13:27

thank you!

Eugene Galaxy (eugene@animaldetect.com)

2025-08-04 04:15:33

Thank you kindly!

If you have any feedback/criticism/improvements -> just shoot them! Always happy to hear what people think. The major part of what exists on Animal Detect was crafted purely from what we hear in the community or directly from the users.

🙌 Nuno Sa

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:17:29

my main feedback right now is that its so simple to use and with an easy to understand interface 🙂 - the identification seemed spot on but that was an easy call

Eugene Galaxy (eugene@animaldetect.com)

2025-08-04 04:22:05

It is based on MegaDetector + SpeciesNet AI models, probably the single greatest contributions to wildlife conservation AI-tech, made by google/microsoft wildlife teams (@Dan Morris/@Ștefan Istrate/@Zhongqi Miao).

Our contribution is packing it into "user friendly" interfaces and providing it online for who do not have enough resources to run it locally. Also, on many images (bulk), we apply our own custom algorithms that boost detection accuracy of the mentioned models as well. Although one cannot see it in action in the one-image-tool. 😁

🙌 Nuno Sa

🙌:skin_tone_3: Alan Stenhouse

Abhi M (abhimandela@gmail.com)

2025-08-04 04:47:09

Hi @Eugene Galaxy, great work, thanks for sharing. I tried with some aerial images, dead animals and camera trap images from India, the results were impressive when the tool is able to accurately draw a bounding box around the animal in large pictures with camouflaged individuals and even blurry photos.

Room for improvement: sometimes, the results were vague (family of species told rather than the species itself) or wrong sometimes (Indian mongoose identified as a Shrew etc...) It may be important for this to be worked on, if you can, because many of the times the images being used to analyse are from camera traps which are not good examples of images taken by people / handheld cameras (targeted)

Thanks and best regards,
Abhi, WildSonic.AI

👀 Nuno Sa

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-08-04 04:57:00

you guys have both a great product 🙂

:smiling_face_with_3_hearts: Eugene Galaxy

Eugene Galaxy (eugene@animaldetect.com)

2025-08-04 05:07:14

Hey @Abhi M, thanks for the feedback! In the bulk-processing platform it is possible to change labels/add new ones both for one and multiple images at once. Would this be something? 🙂

Your project sounds (pun intended! 😄) cool!

Abhi M (abhimandela@gmail.com)

2025-08-04 05:30:06

@Eugene Galaxy, thank you, I will try the bulk platform and let you know.

🙌 Eugene Galaxy

Luke Meyers (lmeyers@SEATTLEU.EDU)

2025-08-04 10:13:49

@Anirudh Potlapally

Burooj Ghani (buroojghani@gmail.com)

2025-08-04 08:07:56

Workshop titled AI for non-human animal communication at NeurIPS 2025 - call for papers now open

I'm proud to announce that we will be organising a workshop on AI for non-human animal communication at NeurIPS 2025. This workshop will explore the intersection of AI and bioacoustics, aiming to address challenges in processing complex bioacoustic data and interpreting animal signals in order to advance our understanding of non-human animal communication.

The call for papers and proposals is currently open and the deadline is August 31, 2025 AOE. For more details, please check: https://aiforanimalcomms.org

👍 Eugene Galaxy, Ben Seleb, Sara Beery, Jason Holmberg (Wild Me), Boyu Zhang, Benjamin Hoffman, Xiangteng He, Nuno Sa, Sasha Syunkova, takatomi-k, Andy Huynh, Nicole Lopez, Adam de Zoete, Maria Jose Guerrero, Malte Pedersen, Ben McEwen, Leonie Baier, Govind Ramnarayan

Alex Logan (alex@cecil.earth)

2025-08-05 12:38:59

Dataset evaluation newsletter
Today our science team at Cecil released a newsletter on dataset evaluation. A bit of an uncomfortable topic, but an important one.

https://www.linkedin.com/posts/alex-logan-cecil_today-cecil-released-our-latest-newsl[…]m=member_desktop&rcm=ACoAAA8nEioBT4Fr4s1CHTRxV04K9Czi50XRyIo

🙌 Carly Batist, Eugene Galaxy

Bree Shi (bshi42@gatech.edu)

2025-08-05 18:32:22

Looking for PhD, Post-Doc, Faculty collaborators (some projects are not conservation tech, but most are) here is our blub (geared toward phd and doctoral students but anyone can feel free to fill out the form and we will contact you to see what kind of collaboration makes sense)

Bree Shi (bshi42@gatech.edu)

2025-08-05 18:32:30

If you're a Post-Doctoral Fellow, a PhD student, or a member of the research staff interested in applying your computational skills to support active research publications, please read on to learn about the Cross-disciplinary Leadership Education for Advanced Researchers (CLEAR) Program .
We are seeking new CLEAR experts to serve as project advisors, participating in structured reviews for ongoing student projects (~5 hours/week). In return for your constructive feedback and contributions, you will be credited as a co-author on the resulting publication. Your role would mirror that of a conference or journal reviewer guiding students toward higher-quality submissions.

Fall 2025 areas of project interest:

Conservation Technology

Legal system automation

Methods Development

Statistical modeling

Computer vision

Benefits of the CLEAR Program:

Publication co-authorship

Research Project Management Training ( Developed by collaboration with three Georgia Tech Department, Used in live project management course)

Access to Network of interdisciplinary Faculty, Post-Docs, and PhD Researchers.

About us:
Human-Augmented Analytics Group(HAAG) is a virtual research initiative that connects online master’s students in computer science with research labs that have analytical, or automation needs but lack in-house computational expertise. Students undertake these collaborations as part of a 3-semester project with the goal of developing and publishing a research tool. This model benefits students by offering them hands-on research experience and benefits labs by providing access to computational support. You can read about our collaborative model in "FAIR-CS: Framework for Interdisciplinary Research Collaborations in Online Computing Programs" which won Best Paper for Workforce Development at PEARC 2025.

The CLEAR Program was launched in Spring 2025 with 5 members of the inaugural cohort many of whom will be publishing their first HAAG papers this Fall! We have since grown to a cohort of 15 doctoral and post-doctoral CLEAR participants. We hope to see you in our Fall 2025 cohort.

Interested?
You can check out our open projects by completing this form:
https://gatech.co1.qualtrics.com/jfe/form/SV_4IoCtf7fGXEIRgi

How It Works (In-Brief) :
You’ll be matched with an interdisciplinary project led by one of our faculty affiliates who will share their goals for the project.
You will be trained in research project management and develop your own goals for the project.
You will meet with your students weekly and your faculty affiliate monthly to ensure adequate progress towards research goals.
In the two weeks before publication submission, we ask that you make yourself available for additional feedback rounds and virtual meetings as needed.
For more information on HAAG click here.

gatech.co1.qualtrics.com

Qualtrics Survey | Qualtrics Experience Management

The most powerful, simple and trusted way to gather experience data. Start your journey to experience management and try a free account today.

Original URL: https://gatech.co1.qualtrics.com/jfe/form/SV_4IoCtf7fGXEIRgi

Sholto West (sholto.west@milkywire.com)

2025-08-06 09:40:10

👋 Hi everyone!
Sholto West here, I am heading up an impact fund for MILKYWIRE!

We’re excited to share a fantastic opportunity: Klarna’s AI for Climate Resilience Program, hosted by Milkywire. They're offering grants of up to $300,000, plus access to mentorship, training, and a global peer‑learning community—supporting AI‑driven projects in climate‑vulnerable regions. Projects that strengthen food security, health systems, or coastal resilience are especially welcome (among others) that are empowering local communities in designing appropriate AI tools applicable to their contexts.

🗓️ Application deadline: August 31, 2025
Project start: January 2026 (12–18 month duration)
Eligibility highlights:

Must focus on real-world implementation in low‑ and middle‑income countries

Responsible, transparent AI that centers local communities

Not open to pure research, emergency relief, or GHG-only projects

👉 Interested? Apply or learn more via the official portal:
https://www.milkywire.com/ai-for-climate-resilience-program

Please feel free to share this within your networks, especially with teams working at the intersection of AI and climate adaptation and also reach out to me on LinkedIn if you would like more info!

milkywire.com

Apply to join the AI for Climate Resilience Program

Apply to the AI for Climate Resilience Program. Funded by Klarna, this grant supports AI-driven climate adaptation projects in vulnerable regions

Original URL: https://www.milkywire.com/ai-for-climate-resilience-program

😎 Jason Holmberg (Wild Me)

Sara Beery (sara.beery@gmail.com)

2025-08-06 16:55:31

Congrats to our community's very own (Dr.) @Levi Cai on his successful PhD defense today!!! He's a driving force in marine animal monitoring with autonomous vehicles, excited to see what's next ❤️

🎉 Don Cosseboom, Esther Rolf, Rupa Kurinchi-Vendhan, Declan, murilo, Timm Haucke, Mitchell Rogers, Jon Van Oast, Xiangteng He, Andy Huynh, Kevin Barnard, Matt Weldy, Sasha Syunkova, Justin Kay, Navodita Mathur, James Liu, Jason Holmberg (Wild Me), Arjun Subramonian (they/them), Mamoon, Dan Morris, Ragib Nihal, Shir Bar, Eugene Galaxy, Patrik Reiske, Leonie Baier, Gözde Cilingir, Hugo, Blair Costelloe, Jonathan Sauder, Nuno Sa, Malte Pedersen, Ethan White, Robin Zbinden, Fagner Cunha, Nicole Lopez, Santiago, Devi Ayyagari, Anna Vallery, Dante Wasmuht, Eric Cunningham, Alexander Merdian-Tarko, Yu Wang, Lauren Harrell, Joseph DelPreto, Namitha Suresh, Sergei Nozdrenkov, Elizabeth Campolongo, Aleksis Pirinen, Neha Hulkund, Ghazi Randhawa

🙌 Laura Chrobak, Jason Holmberg (Wild Me), Shir Bar, Jenna Kline, Lauren Harrell, Joseph DelPreto, Neha Hulkund

❤️ Levi Cai, Joseph DelPreto, Neha Hulkund

Laura Chrobak (lchrobak@mbari.org)

2025-08-06 19:29:50

Congrats @Levi Cai!!

Levi Cai (levi@levicai.com)

2025-08-08 12:30:35

Thank you so much @Sara Beery, this community is awesome 🙂!!

❤️ Sara Beery

Sara Beery (sara.beery@gmail.com)

2025-08-07 15:51:24

@tom denton this paper title is AMAZING

https://arxiv.org/abs/2508.04665

arXiv.org

Perch 2.0: The Bittern Lesson for Bioacoustics

Perch is a performant pre-trained model for bioacoustics. It was trained in supervised fashion, providing both off-the-shelf classification scores for thousands of vocalizing species as well as strong embeddings for transfer learning. In this new release, Perch 2.0, we expand from training exclusively on avian species to a large multi-taxa dataset. The model is trained with self-distillation using a prototype-learning classifier as well as a new source-prediction training criterion. Perch 2.0 obtains state-of-the-art performance on the BirdSet and BEANS benchmarks. It also outperforms specialized marine models on marine transfer learning tasks, despite having almost no marine training data. We present hypotheses as to why fine-grained species classification is a particularly robust pre-training task for bioacoustics.

Original URL: https://arxiv.org/abs/2508.04665

➕ Timm Haucke, Jason Holmberg (Wild Me), Anthony Miyaguchi

🎉 Jon Van Oast, Rupa Kurinchi-Vendhan, Jason Holmberg (Wild Me), Santiago, Timm Haucke, Mark Goldwater, Shir Bar, Levi Cai, Owen Martin, Dan Morris, Sam Lapp, Sasha Syunkova

❤️ Leonardo Viotti, Ben Weinstein, murilo, Viktor Domazetoski, Jason Holmberg (Wild Me), Timm Haucke, Owen Martin, Justin Kay, Nina van Tiel, Yseult Hb, Klára Janoušková, Ștefan Istrate, Ghjulia Sialelli, Ross Gardiner, Juan Sebastián Cañas, Sasha Syunkova

👏 Peter Bull, Jason Holmberg (Wild Me), Timm Haucke, Owen Martin, Yseult Hb, Burooj Ghani, Qasem Alharbi, Sasha Syunkova

🐦 murilo, Jason Holmberg (Wild Me), Timm Haucke, Suman Ghosh, Sam Lapp, Sasha Syunkova

😂 Mitchell Rogers, Mark Goldwater, Matt Weldy, Hayley Smith, Talia Speaker

✅ Jorge Lizarazo

👀 Jorge Lizarazo

🙌 Ben Williams

tom denton (sdenton4@gmail.com)

2025-08-07 16:36:17

They didn't want to let us do it. They said the ornithologists wouldn't know about Sutton, and the ML people clearly had no idea what a bittern is, and thought we were just bad at spelling. But we persevered, and added a figure. :)

😁 Sara Beery, Owen Martin, Klára Janoušková, Justin Kay, Burooj Ghani, Lauren Chronister, Dan Morris, Brooke Goodman

💪 Laura Chrobak, Owen Martin, Klára Janoušková, Lauren Chronister, Brooke Goodman

👏 Hayley Smith

Sara Beery (sara.beery@gmail.com)

2025-08-07 16:38:34

I'm pretty sure this counts as fan service 🤩🤩

Lauren Harrell (laurenaharrell@gmail.com)

2025-08-07 16:46:30

Yes I’m biased, but I still giggle each time I open the paper and see Figure 1. Australasian bittern, and the caption makes it even better.

Owen Martin (owma6084@colorado.edu)

2025-08-07 19:24:57

That abstract is mind blowing tbh

Owen Martin (owma6084@colorado.edu)

2025-08-07 19:26:36

No bats for perch! Curious to wonder if bat and bird vocabulary is similarly complex / information dense but occupying a different spectral niche

Burooj Ghani (buroojghani@gmail.com)

2025-08-08 08:12:23

Great work, @tom denton! Will you be presenting the work at https://aiforanimalcomms.org/? 🙂

tom denton (sdenton4@gmail.com)

2025-08-08 11:38:53

@Owen Martin - In the transfer learning paper (Ghani 2023) we found that pitch-shifted transfer learning for bats worked well, at least on that one data set. (And we still use that task as one of our validation tasks.) Just had a couple long conversations about bats - it seems they are still a challenging group from a data perspective (obtaining ground truth is extremely difficult, for starters) - and I'm interested in what can be done to help them out.

tom denton (sdenton4@gmail.com)

2025-08-08 11:39:25

@Burooj Ghani - "No one can say...." 🙂

tom denton (sdenton4@gmail.com)

2025-08-08 11:39:53

We hope to!

Burooj Ghani (buroojghani@gmail.com)

2025-08-08 11:45:01

“No one can say… but everyone can speculate.”

Dan Morris (agentmorris@gmail.com)

2025-08-08 12:13:39

@Peter Bull There is clearly a new challenger in the conservation technology pun department. The paper this thread is about, plus https://arxiv.org/abs/2505.03071. I think Peter and Tom should meet in a clash of titans at the next one of these: https://www.youtube.com/watch?v=OI4LaYBXAqc .

arXiv.org

The Search for Squawk: Agile Modeling in Bioacoustics

Passive acoustic monitoring (PAM) has shown great promise in helping ecologists understand the health of animal populations and ecosystems. However, extracting insights from millions of hours of audio recordings requires the development of specialized recognizers. This is typically a challenging task, necessitating large amounts of training data and machine learning expertise. In this work, we introduce a general, scalable and data-efficient system for developing recognizers for novel bioacoustic problems in under an hour. Our system consists of several key components that tackle problems in previous bioacoustic workflows: 1) highly generalizable acoustic embeddings pre-trained for birdsong classification minimize data hunger; 2) indexed audio search allows the efficient creation of classifier training datasets, and 3) precomputation of embeddings enables an efficient active learning loop, improving classifier quality iteratively with minimal wait time. Ecologists employed our system in three novel case studies: analyzing coral reef health through unidentified sounds; identifying juvenile Hawaiian bird calls to quantify breeding success and improve endangered species monitoring; and Christmas Island bird occupancy modeling. We augment the case studies with simulated experiments which explore the range of design decisions in a structured way and help establish best practices. Altogether these experiments showcase our system's scalability, efficiency, and generalizability, enabling scientists to quickly address new bioacoustic challenges.

Original URL: https://arxiv.org/abs/2505.03071

YouTube

sashimikid (https://www.youtube.com/@sashimikid)

2016 O.Henry Pun-Off World Championships semi-final round (Janani Krishna-Jha vs Jerzy Gwiazdowski )

Original URL: https://www.youtube.com/watch?v=OI4LaYBXAqc

😆 Peter Bull, Timm Haucke

:spock_hand: Lauren Harrell, Timm Haucke

Peter Bull (peter@drivendata.org)

2025-08-08 12:31:14

Honestly, hat's off. "The Bittern Lesson" is a high bar, and we accept the challenge.

Govind Ramnarayan (govind.ramnarayan@gmail.com)

2025-08-08 17:20:36

This paper looks cool! Just curious, is there a small model / dataset in this area of bioacoustics where hobbyists / individuals can try out different architectures or forms of data augmentation? Something like the equivalent of trying to judge if a technique helps on MNIST or CIFAR to identify whether it is interesting? For instance, I'm wondering if something like getting a description of the animal's calling sound from an LLM might help in identifying the best parts of the recording to use.

Anthony Miyaguchi (acmiyaguchi@gmail.com)

2025-08-13 18:42:15

@Govind Ramnarayan Try out one of the datasets on the birdclef competitions, those are pretty easy to get up and running for soundscape detection. It’s basically just a subset of xeno-canto, and then a bunch of unlabeled soundscapes with leaderboard style scoring. Otherwise birdset seems to be the thing to use

🙏 Govind Ramnarayan

tom denton (sdenton4@gmail.com)

2025-08-14 12:58:26

@Govind Ramnarayan Yep, BirdCLEF is a great place to get started.

For text descriptions of birdsong, you might look up David Robinson's work on BioLIngual (CLIP-like captioning model) and NatureLM. The difficulties with captioning are a) there's not a huge amount of birdsong captioning to learn from, b) human-language descriptions of birdsong are often crude and inexact, and c) (more avoidable) people often pick up general-purpose text-audio models, which are trained almost entirely on speech, and tend to fall over on bioacoustic input.

The biolingual paper does a credible job of getting around these challenges.
https://arxiv.org/abs/2308.04978

arXiv.org

Transferable Models for Bioacoustics with Human Language Supervision

Passive acoustic monitoring offers a scalable, non-invasive method for tracking global biodiversity and anthropogenic impacts on species. Although deep learning has become a vital tool for processing this data, current models are inflexible, typically cover only a handful of species, and are limited by data scarcity. In this work, we propose BioLingual, a new model for bioacoustics based on contrastive language-audio pretraining. We first aggregate bioacoustic archives into a language-audio dataset, called AnimalSpeak, with over a million audio-caption pairs holding information on species, vocalization context, and animal behavior. After training on this dataset to connect language and audio representations, our model can identify over a thousand species' calls across taxa, complete bioacoustic tasks zero-shot, and retrieve animal vocalization recordings from natural text queries. When fine-tuned, BioLingual sets a new state-of-the-art on nine tasks in the Benchmark of Animal Sounds. Given its broad taxa coverage and ability to be flexibly queried in human language, we believe this model opens new paradigms in ecological monitoring and research, including free-text search on the world's acoustic monitoring archives. We open-source our models, dataset, and code.

Original URL: https://arxiv.org/abs/2308.04978

🙏 Govind Ramnarayan

👀 Anthony Miyaguchi

Paul Bodesheim (paul.bodesheim@uni-jena.de)

2025-08-08 04:22:33

📣 Job alert! The newly founded Senckenberg Institute for Plant Form and Function Jena, Germany (SIP) is looking for a Junior Group Leader in Digital Collectomics - Developing modern technologies for collection-based research. CV/ML/AI skills required for knowledge extraction from historical data, which can be transferred to support present and future monitoring of nature. Please spread the word and forward the attached job announcement to interested people. Many thanks.

Chris Smith (chriscs@colorado.edu)

2025-08-08 10:13:27

Hello everyone! I’m hiring a postdoc to do machine learning for molecular ecology.
Starting to build up a lab at Indiana University Bloomington where I just started a faculty position.
Apply with the below link: https://indiana.peopleadmin.com/postings/30325

Rupa Kurinchi-Vendhan (rupak272@mit.edu)

2025-08-08 16:26:20

🌎 Submit to the CV4E Non-Archival Track! 🌎

We had an amazing diversity of submissions for our proceedings track of the Computer Vision for Ecology Workshop at ICCV 2025 this year--thank you to everyone who sent something in!

Do you have an awesome relevant paper published in the last 6 months? Wanna share in-progress work and get feedback from the community?

You’re in luck! Our non-archival submission deadline is just 1 WEEK AWAY: August 15, 2025. Accepted work will be featured as oral or poster presentations. Submit here: https://openreview.net/group?id=thecvf.com/ICCV/2025/Workshop/CV4E_Non-Archival#tab-recent-activity

🙌 Laura Chrobak, Shir Bar, Justin Kay, Julia Chae, Malte Pedersen, Andy Huynh, Navodita Mathur

Rupa Kurinchi-Vendhan (rupak272@mit.edu)

2025-08-12 18:32:56

Bumping this! T-minus 3 days for the Computer Vision for Ecology (CV4E) Workshop deadline for non-archival submissions! Send in your papers here: https://openreview.net/group?id=thecvf.com/ICCV/2025/Workshop/CV4E_Non-Archival. We're excited to see many of you in Hawaii this October!

🙌 Justin Kay, Jason Holmberg (Wild Me), Oisin Mac Aodha, Malte Pedersen, Xiangteng He, Irina Tolkova, Julia Chae, Timm Haucke, Sara Beery

Antonio Ferraz (antonio.a.ferraz@jpl.nasa.gov)

2025-08-14 00:02:35

Hi Rupa, is the workshop on the 19th or the 20th? Can everyone registered in the ICCV attend the workshop ?

Navodita Mathur (navoditamathur1998@gmail.com)

2025-08-14 02:31:34

Hi Rupa, The deadline is showing as August 22nd. Is it 15th or 22nd ?

Rupa Kurinchi-Vendhan (rupak272@mit.edu)

2025-08-14 09:14:24

Hi @Antonio Ferraz, the workshop is on the 20th! And yes, I believe anyone registered for the ICCV with the workshop pass can attend CV4E: https://iccv.thecvf.com/Conferences/2025/Pricing2.

✅ Antonio Ferraz

Rupa Kurinchi-Vendhan (rupak272@mit.edu)

2025-08-14 09:14:52

Hi @Navodita Mathur! We just extended the deadline...making the official announcement now, but good catch 🙂

Campbell Watson (cwatson@us.ibm.com)

2025-08-12 13:16:34

Hi Everyone – Hope you’re all doing ok! Any suggestions on a ground truth LULC dataset for peatlands? I’m particularly interested in southern Chile. Thanks in advance!

Thijs van der Plas (vdplasthijs@gmail.com)

2025-08-12 14:42:18

Hi! We annotated a fair amount at 12.5 cm resolution, incl a few peatland(-related) classes. In the UK though.. See fig 6/8 for some examples, EO + LC data is linked in the paper (I recently heard someone had an issue downloading the data from the repo, if that's the case let me know): https://doi.org/10.3390/rs15225277

For just a geospatial layer (again just UK though), UK gov recently updated their national-scale 'peaty soil' vector map: https://naturalengland.blog.gov.uk/2025/05/12/a-new-peat-map-for-england/ and https://naturalengland-defra.opendata.arcgis.com/datasets/Defra::peaty-soils-location-england/about

MDPI

Multi-Stage Semantic Segmentation Quantifies Fragmentation of Small Habitats at a Landscape Scale

Land cover (LC) maps are used extensively for nature conservation and landscape planning, but low spatial resolution and coarse LC schemas typically limit their applicability to large, broadly defined habitats. In order to target smaller and more-specific habitats, LC maps must be developed at high resolution and fine class detail using automated methods that can efficiently scale to large areas of interest. In this work, we present a Machine Learning approach that addresses this challenge. First, we developed a multi-stage semantic segmentation approach that uses Convolutional Neural Networks (CNNs) to classify LC across the Peak District National Park (PDNP, 1439 km2) in the UK using a detailed, hierarchical LC schema. High-level classes were predicted with 95% accuracy and were subsequently used as masks to predict low-level classes with 72% to 92% accuracy. Next, we used these predictions to analyse the degree and distribution of fragmentation of one specific habitat—wet grassland and rush pasture—at the landscape scale in the PDNP. We found that fragmentation varied across areas designated as primary habitat, highlighting the importance of high-resolution LC maps provided by CNN-powered analysis for nature conservation.

Original URL: https://doi.org/10.3390/rs15225277

naturalengland.blog.gov.uk

A new peat map for England

By Dr Ruth Waters, Director for Evidence, Natural England, and David Jones, Senior Responsible Officer for the Natural Capital and Ecosystem Assessment, Defra England’s peat deposits are one of our most valuable national assets. They’re our largest carbon store, they …

Original URL: https://naturalengland.blog.gov.uk/2025/05/12/a-new-peat-map-for-england/

naturalengland-defra.opendata.arcgis.com

Peaty Soils Location (England)

Peaty Soils Location (England) - The Peat Layer was produced by Natural England (ARM team) during June - October 2008, with the aim of identifying the extent of three classes of peaty soils for the purposes of the Partnership Project to Protect and Enhance Peat Soils.

Original URL: https://naturalengland-defra.opendata.arcgis.com/datasets/Defra::peaty-soils-location-england/about

🙏 Campbell Watson, Dan Morris

Daniel Olhagaray (dolhagaray@tryolabs.com)

2025-08-12 13:56:19

Hi everyone! I'm Daniel, CSM at Tryolabs, an AI & Machine Learning consulting firm from Uruguay.
We are excited to be working with The Nature Conservancy (TNC) on projects for electronic monitoring of fisheries at sea. Tryolabs is leveraging AI-powered video analysis to create a system that processes raw video directly on fishing vessels. This system helps to detect species, track catch and discards, and generate daily reports in near real-time.

The main goal for TNC is to drive sustainable fisheries management and conserve species and biodiversity. This technology helps to:

Detect unsustainable practices early to prevent harm to ecosystems.

Enforce conservation regulations to protect vulnerable species and critical habitats.

Promote transparency in seafood supply chains.

Enable science-based management decisions by providing reliable, timely data.

You can learn more about this collaboration by checking out our case study: AI on board: near real-time insights for sustainable fishing .
Feel free to reach out to me with any questions, needs, or opportunities where I can collaborate from my end, as I'm enthusiastic about getting more involved in AI for conservation projects!

https://www.youtube.com/watch?v=XOmIKsFlCz4

YouTube

Tryolabs (https://www.youtube.com/@tryolabsteam)

AI insights for fishing transparency

Original URL: https://www.youtube.com/watch?v=XOmIKsFlCz4

🙌 Soledad Rivas, Ndapewa Onyothi

Eugene Galaxy (eugene@animaldetect.com)

2025-08-14 06:09:53

Hey people of the world. 10 days ago I posted here about our new GUI tool where one could drag-n-drop an image and get animals detected on it with bounding boxes.
We have just released the same tool as API endpoint for more tech-oriented members of the community.
Give it a spin and let me know about any problems/feedback/criticism and I will do my best!

Quick start: https://www.animaldetect.com/api

Some technical info on the endpoint:
- Average response time is 1.2-1.5 seconds.
- Image size liimt is 5MB
- Pricing: API is included in the same Data Allowance on your Animal Detect account. (all accounts get 1GB of free data which can be 1000 to 25000 images depending on their size)
- Running MegaDetector+SpeciesNet with possibility to specify detection threshold (default is 0.2)

In the near future we are going to expand the list of API endpoints for different use cases. Some of the are:
1. "Animal detected? Yes/No" - For Real-time monitoring
2. Specific animal detector - Looking for an "elephant" and nothing else? This will work.
3. OCR reading of an image - Your images contain a lot of information both in their metadata, and on the image itself. Timestamps, temperature, humidity, gps tags, etc. We can read all of that and return to you.

Have a nice day everyone!

🎉 Jon Van Oast, Jason Holmberg (Wild Me), Antón Álvarez, Alexander Merdian-Tarko, Alan Stenhouse

🔥 Jason Holmberg (Wild Me), Douglas Mbura

Rupa Kurinchi-Vendhan (rupak272@mit.edu)

2025-08-14 09:22:46

📣 CV4E @ ICCV 2025 – Non-Archival Deadline Extended! 📣

Good news! We've extended the non-archival submission deadline to August 22nd, 2025 (EOD).

We wanna see that awesome paper that got published recently, or any in-progress work that you'd wanna start a conversation around. We can't wait to see your submissions! Send them in here: https://openreview.net/group?id=thecvf.com/ICCV/2025/Workshop/CV4E_Non-Archival

🌍 Oisin Mac Aodha, Timm Haucke, Julia Chae, Jonathan Sauder, Elena Sierra, Yuyan Chen, Luciano Dourado

Rupa Kurinchi-Vendhan (rupak272@mit.edu)

2025-08-21 16:10:49

Reminder that the deadline for non-archival submissions for the CV4E Workshop at ICCV 2025 is TOMORROW!

Pablo Medrano (pabmevi@gmail.com)

2025-08-15 12:42:12

Hello, my name is Pablo Medrano Vizcaíno, a Conservation Biologist and Postdoctoral Fellow at the Oklahoma State University, working at the intersection of Conservation Biology and AI. Together with my supervisor @Atriya Sen, we are applying ILP (Inductive Logic Programming) systems to predict extinction risk based on wildlife traits. As a pilot project we are working on inferring wildlife roadkill risk, also applying ILP, and based on wildlife traits.

I would be glad to connect and exchange ideas with experts in the field, and I look forward to exploring potential collaborations.

😎 Sara Beery, Jason Holmberg (Wild Me), Shir Bar, Atriya Sen, Yu Wang, Luciano Dourado

👋 Viktor Domazetoski, Sam Lapp, Dan Morris, Wenxin Yang

Beckett Sterner (bsterne1@asu.edu)

2025-08-15 17:51:58

Hi @Pablo Medrano @Atriya Sen! 👋

Atriya Sen (atriya@atriyasen.com)

2025-08-18 18:27:31

Hi @Beckett Sterner!! 😃 We're building up some steam (finally!), will be in touch!

Filippo Varini (fppvrn@gmail.com)

2025-08-19 11:59:03

Hello everyone,
We’re excited to announce the release of the Community Fish Detection Dataset 🐠🤖!!
This is a large dataset for marine AI with:

1.9M+ images/frames

935K+ bounding boxes

A single, consistent label: “fish”

Over the past six months, members of this community have collaborated to review more than 30 fish datasets spanning freshwater, marine, and lab-based environments. From these, we selected 17 datasets, cleaned them, and converted everything into the standard COCO format.

Thanks to @Dan Morris, the dataset is now hosted on LILA on https://lila.science/datasets/community-fish-detection-dataset/

We’re already using it to train a universal fish detector (think MegaDetector, but for fish), which we will soon share publicly. While similar models exist, our goal is that this larger and more diverse dataset will produce a detector that is more accurate and generalizable.

We’d love for you to explore the dataset, run your own experiments, and share your results with the community.

A huge thank you to all the contributors: @Dan Morris, @Oc, @Sonny Burniston, @Kevin Barnard, @Laura Chrobak, @Alexander Merdian-Tarko, @Devi Ayyagari, Mona Dhiflaoui, Joshua Chen, @Levi Cai, Aditya Shirvalkar, Chris Jackett and others who collaborated!

And an even bigger thank you to all the individual projects whose commitment to open data made this compilation possible!

❤️ Sonny Burniston, Timm Haucke, Justin Kay, Levi Cai, Alexander Merdian-Tarko, Youssef Bayouli, Jose Ruiz-Munoz, Carly Batist, Mark Goldwater, Rupa Kurinchi-Vendhan, Shir Bar, Malte Pedersen, Andy Huynh, Don Cosseboom, Jason Holmberg (Wild Me), Hitoshi Habe, Ștefan Istrate, Nicole Lopez, Adrian Ibanez, Gaspard Dussert, Hugo, Jaycee Choi, Maria Napolitani, Oc, Jack O'Hara, Devi Ayyagari, Sergei Nozdrenkov, Alan Stenhouse

🐟 Sonny Burniston, Timm Haucke, Dan Morris, Justin Kay, Levi Cai, Alexander Merdian-Tarko, Kevin Barnard, Carly Batist, Shir Bar, Suzanne Stathatos, Malte Pedersen, Andy Huynh, Don Cosseboom, Yseult Hb, Ethan White, Jason Holmberg (Wild Me), Hitoshi Habe, Cameron Trotter, Gaspard Dussert, Kate Rakelly, Jaycee Choi, Jonathan Sauder, Elizabeth Campolongo

🎉 Dan Morris, Justin Kay, Timm Haucke, Xiangteng He, Levi Cai, Alexander Merdian-Tarko, Malte Pedersen, Andy Huynh, Don Cosseboom, Yseult Hb, Leonie Baier, Gaspard Dussert

🪸 Laura Chrobak, Andy Huynh, Don Cosseboom

🙌 Carly Batist, Andy Huynh

Laura Chrobak (lchrobak@mbari.org)

2025-08-19 12:39:41

Such exciting progress @Filippo Varini!! Can't wait to test out the MegaDetector for fishfishcowboy

🙌 Filippo Varini

:fishcowboy: Jaycee Choi

🙌:skin_tone_4: Oc

Malte Pedersen (mape@create.aau.dk)

2025-08-19 14:18:24

Really really cool!! Thank you for the huge effort. Looking forward to test this in Scandinavian waters 🙏

🙌 Filippo Varini

Hugo (hma@reallyarobot.com)

2025-08-27 06:40:11

What about other underwater creatures, such as crabs, sea stars and alike? Are they included or excluded?

Filippo Varini (fppvrn@gmail.com)

2025-08-27 11:18:46

Hi @Hugo, no they are not.

👍 Hugo

Laura Chrobak (lchrobak@mbari.org)

2025-08-27 13:18:54

@Hugo do you need a model with other underwater classes?

Hugo (hma@reallyarobot.com)

2025-08-28 03:25:04

@Laura Chrobak Well not now, but I am always trying to keep myself updated with the "detector" models for trail cameras / underwater cameras and insects, as we may further expand models on Animal Detect. I tried the "MegaFishDetector" before which frankly didn’t work that good, so looking forwards to see how well the model trained on "Community Fish Detection Dataset" will do!

One clear lack when I tested the MegaFishDetector was clearly the ability to spot "non-fish" underwater animals, so just wanted to clarify if this was something which was accounted for or not in this model.

👍 Laura Chrobak

Youssef Bayouli (youssef.projetnemo@gmail.com)

2025-08-20 04:54:36

Hi everyone,
We’re excited to share that WILDLABS has launched a new Edge Computing group, which I’ll be co-leading with @Jenna Kline as part of the 2025–2026 Group Leadership Programme.
This group brings together people working at the intersection of edge AI and conservation, with a focus on real-time, on-device data processing for environmental monitoring. Think tracking bat activity with bioacoustics, identifying species on camera traps in real time, analyzing drone imagery directly in the field, or running resilient sensor networks to monitor ecosystems in remote areas.
It’s a space to share tools, models, and strategies and to tackle common challenges like low connectivity, off-grid power, and building systems that can withstand tough environments.
If this sounds up your alley, we’d love for you to join us 🤗
👉 Full announcement: https://wildlabs.net/article/new-group-announcement-edge-computing
👉 Join the group: https://wildlabs.net/groups/edge-computing

🎉 Rohan Sa, Burooj Ghani, Viktor Domazetoski, Alex Rood, Jenna Kline, Don Cosseboom, Jon Van Oast, Dan Morris, Riley Knoedler, Levi Cai, Luke Meyers, Talia Speaker, Abigail Oppong, Alan Stenhouse

💚 Vanesa Reyes, Aswin, Alex Rood, Jenna Kline, Don Cosseboom, Daniel Olhagaray, Suman Ghosh, Levi Cai, Elizabeth Campolongo

Puneet Kollipara (kollipython@gmail.com)

2025-08-21 15:14:05

Early-bird registration is underway for this year's FOSS4G North America, the premier North American conference on open geospatial technologies! Taking place in Reston, VA (near Washington, DC) on November 3–5, FOSS4G NA 2025 will explore geospatial open-source software, open data, open science, open collaboration, Earth observation / remote sensing (EO/RS), GeoAI, mapping, and more. Check out the full schedule of informative talks and hands-on workshops (with plenty of content for environmental, climate, nature and AI practitioners), and our assortment of activities and socials.

HIGHLIGHTS
Keynote lineup: We're thrilled that Katie Picchione of NASA's Disasters Response Coordination System will deliver one of three top-tier keynote talks. Katie uses her cross-cutting background to get EO/RS data and info into the right hands during disasters. Also keynoting FOSS4G NA 2025: Bruce Momjian (EnterpriseDB and PostgreSQL Global Development Group) and Paul Ramsey (Crunchy Data / Snowflake), two leading figures in the geospatial OS software and data-infrastructure landscape.

For job seekers and hirers/trainers: The first-ever Geo-Career Expo & Ice-Cream Social! We invite geo-job seekers and hiring/training org members to forge connections over an ice-cream treat the first day. It's FREE and open to ALL. No conference badge is needed, but separate registration is required for both job seekers and hirers/trainers.

Get involved and spread the word! As a conference dedicated to open-source communities, we invite everyone to play a role in FOSS4G NA 2025. Here's how:

Present a poster: You have until August 26 to submit your poster ideas!

Become a sponsor: Expand your organization’s reach by supporting open geospatial ecosystems. Read our sponsorship prospectus for details.

Volunteer: Enjoy FREE general admission in exchange for volunteering your time to help FOSS4G NA run smoothly.

Above all, please spread the word among your colleagues and networks!

Act by September 17 to lock in early-bird savings. Feel free to DM me with questions. See you in November!

Puneet Kollipara (kollipython@gmail.com)

2025-10-14 10:43:30

🚨 Tonight (10/14) is the deadline for securing the hotel group rate for FOSS4G North America 2025, which will bring together the Earth/geospatial open-source software, open data, open science and open mapping communities on Nov. 3–5 at the Hyatt Regency Hotel in Reston, VA.

Don't miss out on savings and the convenience of staying in the conference block. This is the absolute final call. Book your group-rate room now!

Still need conference tickets? Get yours here!

Puneet Kollipara (https://aiforconservation.slack.com/team/U06U6QSKXAN)

<a href="https://web.cvent.com/event/5a9e1809-2ffb-484c-8998-4289fcb4199e/register?RefId=FOSS4GNA2025Registration">Early-bird registration is underway</a> for this year's <a href="https://foss4gna.org">FOSS4G North America</a>, the premier North American conference on open geospatial technologies! Taking place in Reston, VA (near Washington, DC) on November 3–5, FOSS4G NA 2025 will explore geospatial open-source software, open data, open science, open collaboration, Earth observation / remote sensing (EO/RS), GeoAI, mapping, and more. Check out the <a href="https://www.foss4gna.org/schedule">full schedule</a> of informative talks and hands-on workshops (with plenty of content for environmental, climate, nature and AI practitioners), and our assortment of <a href="https://www.foss4gna.org/social">activities and socials</a>. HIGHLIGHTS Keynote lineup: We're thrilled that <a href="https://www.foss4gna.org/speakers/katie-picchione">Katie Picchione</a> of NASA's Disasters Response Coordination System will deliver one of three top-tier keynote talks. Katie uses her cross-cutting background to get EO/RS data and info into the right hands during disasters. Also keynoting FOSS4G NA 2025: <a href="https://www.foss4gna.org/speakers/bruce-momjian">Bruce Momjian</a> (EnterpriseDB and PostgreSQL Global Development Group) and <a href="https://www.foss4gna.org/speakers/paul-ramsey">Paul Ramsey</a> (Crunchy Data / Snowflake), two leading figures in the geospatial OS software and data-infrastructure landscape. For job seekers and hirers/trainers: The first-ever <a href="https://www.foss4gna.org/social#career">Geo-Career Expo & Ice-Cream Social</a>! We invite geo-job seekers and hiring/training org members to forge connections over an ice-cream treat the first day. It's FREE and open to ALL. No conference badge is needed, but <a href="https://forms.gle/mtMXWHwuJbjf6aRm8">separate registration</a> is required for both job seekers and hirers/trainers. Get involved and spread the word! As a conference dedicated to open-source communities, we invite everyone to play a role in FOSS4G NA 2025. Here's how: • Present a poster: You have until August 26 to <a href="https://talks.osgeo.org/foss4g-na-2025/cfp">submit your poster ideas</a>! • Become a sponsor: Expand your organization’s reach by supporting open geospatial ecosystems. Read our <https://static1.squarespace.com/static/6660a51220c4d544be347123/t/681a9e723bd3be22bb7ccbf5/1746574962919/FOSS+2025+Exhibitor+%26+Sponsor+Brochure+FINAL.pdf|sponsorship prospectus> for details. • Volunteer: Enjoy FREE general admission in exchange for <a href="http://foss4gna.org/volunteer">volunteering your time</a> to help FOSS4G NA run smoothly. • Above all, please spread the word among your colleagues and networks! Act by September 17 to lock in early-bird savings. Feel free to DM me with questions. See you in November!

Original URL: https://aiforconservation.slack.com/archives/CLWGQ4BJ6/p1755803645858949

🙌 David Green

Hugo (hma@reallyarobot.com)

2025-08-22 08:16:00

Hey all,

I don’t think I made a formal introduction of myself here in the group before. I am the co-founder, together with @Eugene Galaxy, of Animal Detect – www.animaldetect.com

From working with robots and AI nearly a decade, I made the jump into the beautiful field of animal conservation and AI. My most recent update is that in parallel of developing and improving Animal Detect, I’ve also started a PhD in ecology. More specifically an AI initiative by Aalborg University (Denmark) called AI:X EcoNet, where EcoNet is one of the 5 “AI labs” starting up now. Each lab has 4 PhDs, two younger (associate) professors and 1-2 “mentors” which are full professors.

EcoNet is mainly focused on how AI can be used and implemented to look at animal interactions, such as how species interact with each other, and for example insects interacts with flowers, monkeys with fruits, and what would happen if one of these “interaction links” are removed, possibly due to climate change.

My specific PhD will mainly focus on vision transformer models (ViTs) to enable quicker labeling and improved classification of animals (insects, fishes, mammals) and moving into labeling and training a vision language model (VLM) based on annotations of interaction, to see if I can fine-tune a model to detect these interactions.

With all this said, I am currently in search for relevant PhD/doctorate courses in AI and Ecology:

Some keywords, preferably in the contexts of animals (insects, mammals, fishes):

Computer vision
Visual Language Models (VLMs)
Visual Transformers (ViTs)
Convolutional neural networks (CNN)
(other?)

I am more than happy to attend a relevant PhD course anywhere in Europe, and possibly also outside of Europe. Let me know if there is something (Mainly looking at for 2026-2027)

🙌 Nuno Sa, Roshni Biswas, Josh Seltzer, Chris Lange, Jason Holmberg (Wild Me), Ross Gardiner, Antón Álvarez, Eugene Galaxy, Navodita Mathur, Almir Neto, Chijioke Ugwuanyi

😎 Jason Holmberg (Wild Me), Alexander Merdian-Tarko, Almir Neto

🆒 Yunxuan Chai (wsyxbcl), Almir Neto

🙌:skin_tone_3: Alan Stenhouse

Paul Bodesheim (paul.bodesheim@uni-jena.de)

2025-08-26 06:38:59

📣 CamTrapAI Workshop 2025 in two weeks (Sep 9+10)!
The program is available now: https://camtrapai.github.io
Join online or on-site!

✨ Highlights:

5 Keynote Presentations of:
- Ulrik Schultz Lundquist (University of Southern Denmark)
- Ted Schmitt (Ai2)
- Tanya Berger-Wolf (Ohio State University)
- Iain Couzin (Max Planck Institute of Animal Behavior)
- Dan Morris (Google AI for Nature)

4 Paper Sessions on:
- Fine-Grained Species Classification
- Open-source Software for Animal Monitoring
- Wildlife Tracking and Interaction
- Animal Detection and Localization

Looking forward to interesting insights and discussions!

camtrapai.github.io

Camera Traps, AI, and Ecology - International Workshop Archive

Camera Traps, AI, and Ecology - International Workshop Series

Original URL: https://camtrapai.github.io

🔥 Oisin Mac Aodha, Timm Haucke, Andy Huynh, Hugo, Otto Brookes, Elizabeth Campolongo

🙌 Malte Pedersen, Timm Haucke, Josh Seltzer, Dan Morris, Andy Huynh, Viktor Domazetoski, S M Rayeed, Dasha Simons, Alexander Merdian-Tarko, Yunxuan Chai (wsyxbcl)

🙌:skin_tone_2: Cara Appel

🙌:skin_tone_4: Oc

Dan Morris (agentmorris@gmail.com)

2025-08-26 17:12:08

tl;dr: if anyone has camera trap images that are your super-amazing-favorites (in terms of aesthetics), and you don't mind sharing, send them my way (by DM or email to agentmorris@gmail.com)?

Long version... as part of my ongoing effort to Do More Stuff With AI (in the LLM sense), I'm thinking of training a model I've been putting off for a few years: a classifier for "hero images" in camera trap datasets. I.e., given a pile of 2M unlabeled camera trap images, find a handful that make an interesting gallery. IMO the solution is not to send every giant pile of camera trap images to a frontier model, but this is a non-expert task (assessing image aesthetics) that frontier models are pretty good at, so I want to use a frontier model to do most of the labeling during training, and I want to vibe-code a human-in-the-loop workflow that helps me refine that training data. The model I would actually use at scale would be a typical (local) image classifier that has basically distilled my concept and the LLM's concept of what "nice" camera trap images are. So I don't need a volume of examples that adds up to "training data volume", but I'd like to assemble a diverse set of examples that adds up to "few-shot volume" to seed the training data pile. I have a bunch that I've picked e.g. for LILA thumbnails, but I'd like to get up to maybe a couple hundred. Fortunately it's really easy to find boring camera trap images, including boring images that have lots of animals in them, just posed in boring ways. I have an infinite supply of those examples.

Extended/philosophical version... the model is not the point here, it's neat but ultimately not very important. The point is that we're likely to increasingly rely on frontier models for labeling whenever niche expertise is not required, and I want to practice that whole workflow by making One Real Thing that works that way. LLMs can't serve as labelers for every task yet, but if we don't practice doing this, we won't know when that transition happens, and we'll still be spending time and money on labeling when we don't have to. So, others who are interested, for what tasks that you work on do you think LLMs could already serve as (non-expert) labelers? Discuss. And also send me your favorite camera trap images.

🎉 Jon Van Oast, Angus Galloway, Mitchell Rogers, Dasha Simons

❤️ Eugene Galaxy, Hugo, Andrea Zampetti, tom denton, Alan Stenhouse

Ben Weinstein (benweinstein2010@gmail.com)

2025-08-26 17:43:18

Feels like this is related to thumbnail or keynote detection, or summary detection. A good hero image might have all of these qualities. Cool!

Angus Galloway (agalloway@septicsitter.com)

2025-08-26 18:59:30

Re: for what tasks that you work on do you think LLMs could already serve as (non-expert) labelers?

We're characterizing lakebed habitats with underwater drones, and wanted to develop a new segmentation model for empty shells called "shell hash", an important substrate in the North American Great Lakes. We wanted to do an initial screening for "interesting" images that would be worth sending for dense labeling, i.e., non-empty, non-trivial, say between 5-95% coverage. I was impressed by Gemini's "Thoughts" on these images, which went well beyond my shell query. They could possibly be used as a proxy for image difficulty and help us screen for other considerations like submerged aquatic vegetation (SAV). SAM often performs poorly for these images, I'm guessing they're just too OOD.

Dan Morris (agentmorris@gmail.com)

2025-08-26 19:33:55

Difficulty assessment is a good idea here.. in my experience frontier models still struggle with nuanced segmentation tasks, but descriptions of even very complex images are pretty amazing. Even aside from difficulty assessment, diversity in LLM descriptions might even be a useful proxy for diversity in image content, to help with assembling a diverse set for manual labeling.

👍 Angus Galloway

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2025-08-27 09:58:10

Hi @Dan Morris This would be valuable for Wildbook as we often have many photos per individual and predicting and showing a charismatic "hero" image would be helpful.

PG (premsgill@gmail.com)

2025-08-27 09:40:20

Hi all. I'm interested in the ever increasing access to HPCs globally and its growing usage with users outside of computer science, in particular within wildlife conservation. I wonder if anyone has any interesting papers or projects that explores the growing use of HPCs beyond "traditional" users that they could point me to?

Riley Knoedler (mknoedler@west-inc.com)

2025-08-27 14:30:20

I know of CV4Animals at CVPR and CV4Ecology at ICCV. Is there any presence from this community at WACV https://wacv.thecvf.com/?

Malte Pedersen (mape@create.aau.dk)

2025-08-28 04:35:07

And the Joint Workshop on Marine Vision at ICCV! (Not WACV, though).

🙌 Sara Beery

Anthony Hoogs (anthony.hoogs@kitware.com)

2025-10-13 23:36:48

Sometimes there are environmentally-oriented workshops at WACV, but recently there has not been a consistent workshop that recurs year over year. I organized a series of environmental monitoring workshops at WACV 2015-17.

Brittany Solano (brittany@earthspecies.org)

2025-08-27 15:15:56

Hello, all! We are seeking papers and proposals for the inaugural AI for Animal Communication workshop at NeurIPS 2025.

🚨Call for submissions: September 5, 2025 AOE
🌐 More details: https://bit.ly/aiforanimals

aiforanimalcomms.org

AI for Non-Human Animal Communication - NeurIPS 2025 Workshop

This workshop will explore the intersection of artificial intelligence (AI) and bioacoustics, aiming to address challenges in processing complex bioacoustic data and interpreting animal signals in order to advance our understanding of non-human animal communication. Join us for a poster session, keynote talks and a panel discussion as we explore key opportunities to use AI to decipher animal languages and thus deepen our understanding of the natural world.

Original URL: http://aiforanimalcomms.org

aiforanimalcomms.org

AI for Non-Human Animal Communication - NeurIPS 2025 Workshop

Original URL: https://bit.ly/aiforanimals

🙌 Felipe Parodi, Owen Martin, Andy Viet Huynh, Hugo, Sara Beery

Filippo Varini (fppvrn@gmail.com)

2025-08-28 04:44:28

Hello everyone, I wanted to ask a collective advice: what conferences on AI (and AI4C) are you planning to attend or would recommend attending in end of 2025 and 2026?

Filippo Varini (fppvrn@gmail.com)

2025-08-28 04:45:42

Speaking with some friends here are some recommendations I received

Filippo Varini (fppvrn@gmail.com)

2025-08-28 04:46:16

The Joint Workshop on Marine Vision, happening October 19th, 2025 as part of ICCV

❤️ Hugo, Malte Pedersen

Filippo Varini (fppvrn@gmail.com)

2025-08-28 05:07:09

Also NeurIPS 2025 since there’s the CCAI workshop and AI4A

Filippo Varini (fppvrn@gmail.com)

2025-08-28 05:13:55

Also CVPR, since I heard they host interesting workshop such as Earthvision (remote sensing), FGVC, AI4Animals etc

❤️ Mitchell Rogers

Eugene Galaxy (eugene@animaldetect.com)

2025-08-28 06:36:46

Hey hey. I am attending NetworkNature in Brussels, on September 16.

https://networknature.eu/networknature-annual-event-2025-choose-nature

NetworkNature

NetworkNature Annual Event 2025 | Choose Nature.

Choose nature. Unlocking pathways to a resilient economy for people and the planet.

Original URL: https://networknature.eu/networknature-annual-event-2025-choose-nature

Eugene Galaxy (eugene@animaldetect.com)

2025-08-28 06:38:20

There are also these around Europe:

https://www.ibac25.com/
https://ai-environment-summit.com/
https://rewilding.academy/rewilding-our-world-conference-2025/

(2026)
https://conferences.au.dk/tibs-aarhus-2026
https://www.nordicsocietyoikos.org/news/nordic-oikos-2026

Filippo Varini (fppvrn@gmail.com)

2025-08-28 10:01:54

Look super interesting!! Thank you!!

Carly Batist (carly.h.batist@gmail.com)

2025-08-28 11:22:00

Not specifically AI, but lots of AI session! Living Data
Oct. 21-24 in Bogotá, Colombia
I'll be there!

livingdata2025.com

Living Data 2025 - Information that together, transforms

Join us in Bogotá for Living Data 2025, exploring biodiversity data standards, integration, and community engagement.

Original URL: https://www.livingdata2025.com/

➕ Laura Chrobak, Sara Beery

✔️ Jon Van Oast

Carly Batist (carly.h.batist@gmail.com)

2025-08-28 11:28:10

Nature Tech Week will be happening again next year in spring 2026.
This is the 2025 website - https://www.naturetechweek.com/
Follow Nature Tech Collective (NTC) for announcements on that

🙌 Filippo Varini

Laura Chrobak (lchrobak@mbari.org)

2025-08-28 11:45:20

CVPR 26 (FGVC and CV4Animals workshops🎉) and OSM26 (ME001 - Accelerating Discovery in Marine Imagery with Artificial Intelligence)

🙌 Filippo Varini

Sara Beery (sara.beery@gmail.com)

2025-08-28 11:49:57

I'll be at ICCV, NeurIPS, and CVPR probably, there are good workshops at each!

❤️ Mitchell Rogers

🐦 Mitchell Rogers

🙌 Filippo Varini

Riley Knoedler (mknoedler@west-inc.com)

2025-08-28 13:34:09

Also want to shout out that more and more talks and sessions on AI and Conservation are happening at ecology-focused conferences like The Wildlife Society https://twsconference.org/ and Ecological Society of America https://esa.org/events/meetings/

👍 Sara Beery

🙌 Filippo Varini

Malte Pedersen (mape@create.aau.dk)

2025-08-28 14:04:56

ECCV in September 2026 in Sweden is likely also hosting a few relevant workshops. (David Nakath and I are hoping to continue the Joint Workshop on Marine Vision there).

🙌 Filippo Varini, Kalindi Fonda

Devis Tuia (devis.tuia@epfl.ch)

2025-08-29 04:24:12

Some events in Europe, in case you are on this side of the pond:

We just submitted some papers to TIBS in Aarhus. https://conferences.au.dk/tibs-aarhus-2026

Otherwise the World Biodiversity Forum is about to open for submissions: https://worldbiodiversityforum.org

In the messages above, the usual suspects (CVPR, ECCV, etc) are already mentioned 😉

conferences.au.dk

The International Biogeography Society Conference Jan 7-9, 2026 Aarhus

The International Biogeography Society and Aarhus University are excited to welcome you to the 12th Biannual Conference in Aarhus, the city of Smiles.

Original URL: https://conferences.au.dk/tibs-aarhus-2026

worldbiodiversityforum.org

Scientist Feedback Signup

Original URL: https://worldbiodiversityforum.org

❤️ Sara Beery

Filippo Varini (fppvrn@gmail.com)

2025-08-29 08:25:39

Hello! For anyone who is interested in AI for Ocean Conservation, at OceanX we are hosting an long-form hackathon to address challenges of conservation in the deep sea with AI 🙂

Find more about it and apply here!

👀 Hugo, Ghjulia Sialelli, Carly Batist, Rupa Kurinchi-Vendhan, Shir Bar, Andy Viet Huynh, Dasha Simons, Anthony Miyaguchi

😎 Jon Van Oast, Andy Viet Huynh, Elizabeth Campolongo

🌊 Malte Pedersen, Andy Viet Huynh, Alexander Merdian-Tarko, murilo, Elizabeth Campolongo

❤️ Owen Martin, Sonny Burniston, Andy Viet Huynh, Maria Napolitani, Wilhelm

🙌 Marco Ghilardi

Hugo (hma@reallyarobot.com)

2025-08-29 08:28:13

I have a few people I will forward this to, is it physical attendance, and where?

Filippo Varini (fppvrn@gmail.com)

2025-08-29 08:31:54

Amazing @Hugo!
It’s remote-first, but it includes one in-person event in New York in November and one in the beginning of 2026 if the team is shortlisted to the finals.

❤️ Hugo, gyeongsu, James Liu

Marco Ghilardi (mghilardi@quasiscience.com)

2025-09-07 18:21:48

Hi @Filippo Varini,
Thanks for sharing looks like a fun opportunity. Would you be able to share a link to the application please? The one embedded in the PDF does not work for me.
Many thanks,
Marco

Filippo Varini (fppvrn@gmail.com)

2025-09-08 04:25:57

Sure! Here is the link to the challenge and application.

oceanx.org

OceanX

Join us as we build a global community deeply engaged with understanding, enjoying, and protecting our oceans.

Original URL: https://oceanx.org/science-impact-challenge/

Hanane Moussa (moussa.45@buckeyemail.osu.edu)

2025-09-03 14:45:33

Paid Opportunity to Participate in User Study

Hello 👋!

We are a group of researchers from The Ohio State University and Allen Institute for AI, and we are looking for participants to give feedback on our multi-domain AI-based research proposal evaluator.

We are particularly looking for participants with a background in Chemistry, Neuroscience, and Ecology, who have published at least 1 paper in their field. Participants will be compensated with up to $100 (as Amazon gift cards).

As a participant, your task will consist of creating a research proposal (this can be an unpublished idea or from your already published work), generating an evaluation of the idea using our system, and filling out a rubric scoring the output. Your unpublished ideas will not be shared publicly without permission.

If you are interested please sign-up through this link 👉 https://go.osu.edu/scholar-eval-signup

Looking forward to receiving your responses!

N.B: Participants should be available for the user study at some point in the next 2 weeks. The task itself should take no longer than 2 hours!

Google Docs

We are a group of researchers from The Ohio State University and Allen Institute for Artificial Intelligence seeking feedback on our multi-domain, literature-based research proposal evaluator. Your Task Your task is to create your own research proposal, generate an evaluation from our system (blind and randomized with another baseline system), and fill out a basic rubric scoring the output. Your proposal can be from an ongoing project, prior published/unpublished work, or something you have yet to start. Your ideas will not be shared publicly without permission. Time & Compensation The expected time for the user to create a research proposal, generate the evaluation, and score the rubric is 1 hour. Each user may submit up to 4 different research proposals for evaluation, paid $25 each for a total of $100 (Amazon gift cards). Eligibility We are primarily targeting those with 1+ paper published in any field. Questions Please use this form or contact <a href="mailto:dasilva.30@buckeyemail.osu.edu">dasilva.30@buckeyemail.osu.edu</a> for any questions

Original URL: https://go.osu.edu/scholar-eval-signup

🙌 Mamoon, Carly Batist, Anna Vallery

🙌:skin_tone_3: Alan Stenhouse

Irina Tolkova (itolkova@g.harvard.edu)

2025-09-10 19:02:38

Hi Hanane, sounds like an interesting study. Could I ask, what will happen to the submitted proposals? Will they be stored for further use, or published anywhere? Will they be used to develop/train models for commercial purposes?

Viktor Domazetoski (viktor.domazetoski@hotmail.com)

2025-09-08 14:36:37

Hi everyone!

We are excited to share that @Anna Vallery (conservation biologist and coastal avian ecologist), @Josh Seltzer (independent researcher, and founder of AI company) and I (PhD student researching responsible AI for Conservation) will be co-leading the AI for Conservation group at WILDLABS (https://wildlabs.net/groups/ai-conservation) over the next 12 months! If you’re not already a member of WILDLABS, we encourage you to join! It’s a vibrant community bringing together conservation AI and tech from many angles, so you’re sure to find resources, ideas, and people relevant to your work.

As one of the largest communities at the intersection of artificial intelligence and conservation, our goal is to foster collaboration, bridge the disciplines and spark new ideas and research directions. Most importantly, we want to shape this space and our work around your needs and interests! To help us do that, we’ve created a poll that aims to act as a community census, so we can better understand where you’re coming from, what excites you and what kind of activities would be most useful.

https://forms.gle/cxXPoF9YhUVyiqYa8 (deadline 5th October)

We’d love to hear your thoughts! Please fill out the poll at your convenience, join the thread on WILDLABS, or reach out to any of us. We’re looking forward to working together to not only grow this community, but more importantly, make it more connected and valuable for everyone.

Google Docs

WILDLABS AI for Conservation - Group Interests

The WILDLABS AI For Conservation group is intended to unite and inspire all WILDLABS community members—whether already involved in AI for conservation, or not—to understand how to use and/or directly contribute to open-source research and development efforts. As we plan for the year ahead, we're eager to hear from our members (and future members) about what you'd like to see, what events you're interested in attending, and how the group leaders can best serve this community! *All questions below are optional

Original URL: https://forms.gle/cxXPoF9YhUVyiqYa8

🔥 Sara Beery, Carly Batist, Jason Holmberg (Wild Me), Oisin Mac Aodha, Josh Seltzer, Yseult Hb, Ross Gardiner, Prabath Gunawardane, Carlos Rodríguez - Pardo

💚 Nicolas Arrieta Larraza, Vanesa Reyes, Elizabeth Campolongo, Sara Beery, Alexander Merdian-Tarko

🎉 Anna Vallery

🙊 Josh Seltzer

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2025-09-08 15:16:37

Thank you!

❤️ Viktor Domazetoski, Josh Seltzer

Anthony Miyaguchi (acmiyaguchi@gmail.com)

2025-09-09 07:48:22

Hey y’all, in case anyone is interested, the LifeCLEF 2025 workshop is starting in 30 minutes at 14:15CEST, covering species distribution, animal reidentification, multilabel plant detection, fungi detection, and bird all classification. A teams link can be found at the homepage https://www.imageclef.org/LifeCLEF2025

🙌 Oisin Mac Aodha, Lukas Picek, Juan Sebastián Cañas, Adrian Cheung, Xiangteng He, murilo

Tanya Berger-Wolf (berger-wolf.1@osu.edu)

2025-09-09 22:12:42

From GBIF:
📣 Calling all biologists, botanists, ecologists, students and researchers!

If you’ve ever said “I’ll publish that dataset later...” ChatIPT is the “later.” 😎

Developed by Rukaya Johaadien of #GBIFNorway, this clever chatbot allowes non-technical users to easily share their data to GBIF.

A new version of the tool has just been release, and Rukaya is looking for feedback! The newest version integrates ORCID login, dataset management, post-publication editing and improved functionality.🤖

This easy-to-use platform handles the technical components of sharing data to GBIF and aims to unlock data hidden in spreadsheets!

Interested? Give it a go today 👉 https://lnkd.in/dmvnrdps
Learn more: https://lnkd.in/dX9PWAjD

chatipt.svc.gbif.no

ChatIPT

Publish your spreadsheets to GBIF through a chat interface

Original URL: https://lnkd.in/dmvnrdps

lnkd.in

This link will take you to a page that’s not on LinkedIn

Original URL: https://lnkd.in/dX9PWAjD

🔥 Josh Seltzer, Ankita Shukla, Shir Bar, Viktor Domazetoski, Elizabeth Campolongo, Sara Beery, Anna Vallery, Alan Stenhouse

🤩 Malte Pedersen, Maria Napolitani

Josh Seltzer (jyseltz@gmail.com)

2025-09-09 22:21:27

Very cool, thank you for sharing! (@Anna Vallery maybe this could be useful / good inspiration for what you have in mind?)

🎉 Anna Vallery

Otto Brookes (otto.brookes@bristol.ac.uk)

2025-09-10 12:12:28

The second day of the 2025 workshop is starting now! Feel free to join online on short notice via GoogleMeet: https://camtrapai.github.io
Iain Couzin beginning his talk in 5 minutes...

camtrapai.github.io

Camera Traps, AI, and Ecology - International Workshop Series

Original URL: https://camtrapai.github.io

🔥 Oisin Mac Aodha, Dante Wasmuht, Timm Haucke, Andy Viet Huynh, Andy Huynh, Christoph Praschl

🙌 Matthias Zuerl, Timm Haucke, Josh Seltzer, Andy Viet Huynh, Mitchell Rogers, Andy Huynh

🙌:skin_tone_2: Cara Appel

Sara Beery (sara.beery@gmail.com)

2025-09-10 19:35:57

If you're located in the Northeast, come join us at our first local meetup!!!

❤️ Irina Tolkova, Rupa Kurinchi-Vendhan, Justin Kay, Aaron Serianni, Levi Cai, Andy Huynh, Elena Sierra, Christoph Praschl, Shir Bar, Timm Haucke, Sara Si-Moussi, Shayan Azhar, Oisin Mac Aodha, Konstantin Klemmer, Risa Shinoda, Ed Miller, Daniel Yang, Aleksis Pirinen, Nayan Mehta

😎 Jason Holmberg (Wild Me), Jon Van Oast, Levi Cai, Andy Huynh, Elena Sierra, Timm Haucke

:female_fairy: Burak Ekim

Sara Beery (sara.beery@gmail.com)

2025-09-10 19:49:09

Help spread the word!

https://x.com/sarameghanbeery/status/1965924809235378359
https://bsky.app/profile/sarameghanbeery.bsky.social/post/3lyjfv3sx222c

X (formerly Twitter)

University of Oxford (@UniofOxford) on X

Are you interested in #AIforConservation #AIforBiodiversity #AIforWildlife or #AIforNature? Are you located in the Boston Area? If so, come join us!! The AI for Conservation Slack community is doing our first local-area Boston meetup, partnering with iNaturalist and TEDx Boston

Original URL: https://x.com/sarameghanbeery/status/1965924809235378359

bluesky

Sara Beery (https://bsky.app/profile/sarameghanbeery.bsky.social)

Are you interested in #AIforConservation #AIforBiodiversity #AIforWildlife or #AIforNature?? Are you located in the Boston Area? If so, come join us!! The AI for Conservation Slack community is doing our first local-area Boston meetup, partnering with iNaturalist and TEDx Boston!

Original URL: https://bsky.app/profile/sarameghanbeery.bsky.social/post/3lyjfv3sx222c

Sara Beery (sara.beery@gmail.com)

2025-09-23 18:27:01

You can still register and join us in Boston this weekend!! First northeast meetup!!!

📣 Elena Sierra

Dan Morris (agentmorris@gmail.com)

2025-09-11 17:25:42

This is slightly old news now, but since I never formally announced it... I released new MegaDetector models a little over a month ago. I wanted to wait to announce this until a few people other than me had kicked the tires, and they have, so here we are:

https://github.com/agentmorris/MegaDetector/blob/main/docs/release-notes/mdv1000-release.md

I will keep this post short; the "release notes" turned out to be closer to a "release novel" that is more about other random MegaDetector stuff than it is about this new release, but if you want 100x your recommended daily dose of random information about MegaDetector, the release notes have you covered.

The most important thing to mention here is that no one using MD should rush to "upgrade"; I believe the largest of the new models is more accurate than MDv5 more often than it's not, but it's never as simple as "this is just obviously better and everyone should stop using previous MD models".

As always, feedback is welcome; I especially love hearing about MegaDetector failures. Give me your blurry, your partially-submerged, your inexplicably-missed wildlife yearning to be detected.

🙌 Carly Batist, Timm Haucke, Andy Huynh, Shir Bar, Paul Bodesheim, Roberta Hunt, Eugene Galaxy, Alexander Merdian-Tarko, Elizabeth Campolongo, Sam Lapp, Peter Bull, Adam Elzinga, Viktor Domazetoski, Hugo, Nayan Mehta

😎 Jon Van Oast

🙌:skin_tone_2: Cara Appel

🙌:skin_tone_3: Alan Stenhouse

Dan Morris (agentmorris@gmail.com)

2025-09-11 17:39:01

It was great to see so many people here in the Pacific Northwest for the CamtrapAI workshop this week! Bonus points to the one person who took me up on my offer to join me and my dog for a 5:30am nature walk; in addition to the expected beavers and osprey and muskrats, we got to see a Northern saw-whet owl at ~five feet away. I confess I had never heard of this owl, but it turns out they are adorable.

[pausing while everyone goes and looks at pictures of northern saw-whet owls]

But I'm not here posting just to talk about owls. During my talk at the workshop, I promised that if I didn't get to the section of the talk where we collectively discuss what LLMs mean for our field, I would make it my mission to organize a follow-up discussion. As it happened, I got nowhere near that part of the talk, so, I'm making good on my promise to organize a follow-up session. I'll talk through a couple of the slides I would have shown, but mostly I am aiming for informal discussion, basically "virtual coffee and chat about how LLMs will change camera trap workflows".

If you are interested, email me at agentmorris@gmail.com and tell me which of the following times you are available (choosing early PST times to attempt Europe-friendliness):

Thursday 9/18, 7am PT or 8am PT

Friday 9/19, 7am PT or 8am PT

Wednesday 9/24, 7am PT

Thursday 9/25, 7am PT or 8am PT

Friday 9/26, 7am PT or 8am PT

If >=1 person replies, I will pick a time, and post a meeting link back here. Everyone is welcome, but FWIW in the interest of making good on my promise, I am aiming to have a discussion specifically about the "AI for camera traps" field, though most things that I think will come up will apply to other data processing domains as well. But if I steer the conversation toward camera traps, don't say you weren't warned.

🙌 Carly Batist, Ted Schmitt, Josh Seltzer, Andy Huynh, Elena Sierra, Elizabeth Campolongo, Carly Batist, Angela Zhu

👍 Jon Van Oast, Josh Seltzer, Paul Bodesheim, Ed Miller

🦉 Cara Appel, Elizabeth Campolongo, Jennifer, Ed Miller, Nayan Mehta

🙌:skin_tone_3: Alan Stenhouse

Dan Morris (agentmorris@gmail.com)

2025-09-15 22:13:59

Plenty of replies (lots more than 1), and the time that maximized availability was 7am PT on Thursday 9/25. On second thought I'm not going to post a link, but if anyone else wants to join, just DM or email me.

If folks want some optional homework in advance... the slides I presented (a little bit) at the CamtrapAI workshop are here. The (short) LLM-stuff section starts here. Most slides have open questions to which I don't claim to have answers. So, pontificate away, and bring your pontification next week.

In particular, if someone wants to spend time with their favorite agentic CLI between now and next week, I'm particularly interested in one possible homework item... I keep a list of publicly-available AI models for camera traps here. As fodder for the discussion of how LLMs will change workflows (as opposed to using LLMs as vision models), I'm interested in a prompt like "hey, AI, Dan keeps a list of camera trap models online, here's a folder of camera trap images, run all the models on those images and compare them" (with no other guidance). Or, alternatively, "here's a link to a dataset on LILA, download it and run all the models on those images". I think AI would almost be able to do that now, and I expect that AI will be able to do that with minimal supervision by the end of December. But I'm not sure. It will still require a human there to tell it where to put stuff and where to run stuff, but I can't point to any specific step in that process - finding model documentation, figuring out the image data, running azcopy or gsutil, validating the output, fixing random TF/PT errors, thinking at a high level about how to compare models with different taxonomic outputs - etc. that AI definitely can't do right now.

This kind of thing likely doesn't make sense in a web-based interface, it likely has to be something with local access like Gemini CLI, Claude Code, or Claude Desktop. But maybe I'm not imaginative enough when it comes to browser-based LLM interfaces.

The more random stuff people have had successes and failures with, the more we'll have to talk about, so, try random stuff in advance!

🙌:skin_tone_3: Alan Stenhouse

Howard L Frederick (simbamangu@gmail.com)

2025-09-12 02:11:28

Interesting and somewhat controversial AI and satellite paper in PNAS a few days ago, attempting a total enumeration of the Serengeti wildebeest migration using satellite imagery and a couple of models.
https://academic.oup.com/pnasnexus/article/4/9/pgaf264/8249339

Owen Martin (owma6084@colorado.edu)

2025-09-12 10:29:53

Super interesting! Could you speak to the controversy a bit?

Howard L Frederick (simbamangu@gmail.com)

2025-09-12 11:45:00

High-resolution aerial photographic sample counts (used for past decades) show a population of around 1,300,000 (in 2023).

The satellite count claims to be far more accurate, and shows:

330,000 in 2022

530,000 in 2023

❤️ Owen Martin

Ted Schmitt (teds@allenai.org)

2025-09-12 17:15:44

And that’s controversial because of the implications for management and conservation based on which number you believe, correct? For instance, have authorities in Tanzania reacted to this discrepancy?

Owen Martin (owma6084@colorado.edu)

2025-09-12 17:16:33

Wow huge discrepancy

Ted Schmitt (teds@allenai.org)

2025-09-12 17:26:23

Yes and surprising because AI models have generally shown that traditional methods undercount rather than over count. I have big questions.

Mitchell Rogers (mitchell.rogers@auckland.ac.nz)

2025-09-12 20:05:40

For anyone looking for a short-form summary of the paper: https://x.com/UniofOxford/status/1965759773518749765
It is interesting to point out that one of the authors (Tony Sinclair: https://www.theserengetirules.com/tony-sinclair) has been heavily involved in the previous aerial counts, so this is a continuation of previous surveying works rather than a new and independent survey.

X (formerly Twitter)

NEW: A pioneering study has applied AI for the first time to count the Great Wildebeest Migration from satellite images. Unexpectedly, the results showed fewer than 600,000 individual wildebeest – less than half the previous estimate of 1.3 million animals. #OxfordClimate

Original URL: https://x.com/UniofOxford/status/1965759773518749765

The Serengeti Rules

Tony Sinclair — The Serengeti Rules

Tony Sinclair has conducted ecological research on the role of biodiversity in the functioning of ecosystems around the world including Africa, Australia, New Zealand and Canada.

Original URL: https://www.theserengetirules.com/tony-sinclair

❤️ Owen Martin

Howard L Frederick (simbamangu@gmail.com)

2025-09-13 03:27:05

@Ted Schmitt there’s a press release on the way from the Tanzania Wildlife Research Institute; there could be implications for management and funding down the road too.

Howard L Frederick (simbamangu@gmail.com)

2025-09-13 03:28:59

One of the implications from the paper is that sample counts from aircraft are somehow overestimating badly, and have always done so - but the aerial methods were validated against other data over the decades, and our experience with using AI to count the imagery suggests (as you say) that the humans undercount!

Keiller Nogueira (keillernogueira@gmail.com)

2025-09-12 11:08:40

3rd Data-Centric Land Cover Classification Challenge

We're thrilled to announce the 3rd Data-Centric Land Cover Classification Challenge as part of the Workshop on Machine Vision for Earth Observation and Environment Monitoring (MVEO) @BMVC 2025 in Sheffield, England, UK!

Building on the success of last years' challenges, this year participants will have to develop AI-based models that achieve the highest possible performance on a small, highly imbalanced, multi-modal semantic segmentation dataset.

Key Details:
Challenge Details: https://mveo.github.io/challenge.html
CodaBench Competition Page: https://www.codabench.org/competitions/10453/

Challenge Deadline: 11 November 2025
Workshop Date: 27 November 2025
Location: Sheffield, UK

The top performers will present their methods at the Workshop and be invited to co-author a journal paper summarizing the challenge outcomes.

mveo.github.io

Workshop on Machine Vision for Earth Observation and Environment Monitoring

Original URL: https://mveo.github.io/challenge.html

Dan Morris (agentmorris@gmail.com)

2025-09-12 12:08:50

Folks in the DC area (which does not include me, but I would come if I were there), consider joining our friends at Development Seed for an in-person (!!!) conservation-themed vibe coding party:

https://luma.com/gukb6kc4

😮 Jon Van Oast, Sara Beery, Alex Logan, Catherine

🙌 Carly Batist

Catherine (catherine.m.breen@gmail.com)

2025-09-12 13:10:30

Love this! Thanks for sharing !

Jon Van Oast (jon@wildme.org)

2025-09-12 13:45:11

shared with our dc office. thanks.

Josh Seltzer (jyseltz@gmail.com)

2025-09-12 14:33:58

And for peeps anywhere in the world - you're invited to join me for a virtual Conservation Tech Vibe Coding coding session next Wednesday @ 8am PT / 11am ET / 5pm CET.

The format is pretty simple - I'll spend ~30 minutes live coding a simple app (which we'll release open-source), and then will stick around for another ~30 minutes to discuss, help with questions, or just hang out. I'll also be demo'ing KIRO which looks pretty cool.

You can register on the link or DM me if you would like to attend on another date.

Should be a fun time yay-frog

luma.com

Conservation Vibe Coding · Luma

Vibe Coding party to explore agentic tools, with a focus on building solutions for conservation tech.

Original URL: https://luma.com/adcnqdq5

Kiro

Kiro is an agentic IDE that helps you go from prototype to production with spec-driven development.

Original URL: https://kiro.dev/

🔥 Sara Beery, Dan Morris, Sam Lapp, Ștefan Istrate, Viktor Domazetoski

🙌 Carly Batist

Prabath Gunawardane (prabathg@gmail.com)

2025-09-14 18:31:11

I'd like to make it but unfortunately I'll be out of town for a couple of weeks starting from next Wednesday. Would love to join the next one 🙂 !

🙌 Josh Seltzer

Ryan Anderson (rkanderson15@gmail.com)

2025-09-17 11:05:31

Hi Josh,
I registered for the event, but am having trouble joining. The calendar event doesn’t seem to have an associated google meet room, and the Luma link just says “google meet” — without providing any link

👀 Josh Seltzer

Yihang She (yihshe@student.ethz.ch)

2025-09-16 06:10:19

We have just released a synthetic data pipeline and dataset to help build robust forest vision systems. Please check it out if you are interested!

linkedin.com

Cambridge Arboreal Modelling Panoptic 3D (CAMP3D) Dataset | Yihang She

🌲🚀 We’ve just released 𝗖𝗔𝗠𝗣𝟯𝗗, a synthetic data pipeline and dataset to help scale up 𝗳𝗼𝗿𝗲𝘀𝘁 𝘃𝗶𝘀𝗶𝗼𝗻 𝘀𝘆𝘀𝘁𝗲𝗺𝘀. Do you work with 𝗟𝗶𝗗𝗔𝗥 𝗳𝗼𝗿 𝗳𝗼𝗿𝗲𝘀𝘁 𝗶𝗻𝘃𝗲𝗻𝘁𝗼𝗿𝘆 and find it costly and time-consuming to label tree point clouds for training forest vision systems? Do you ever play 𝘃𝗶𝗱𝗲𝗼 𝗴𝗮𝗺𝗲𝘀 𝘄𝗶𝘁𝗵 𝗳𝗼𝗿𝗲𝘀𝘁 𝘀𝗰𝗲𝗻𝗲𝘀 𝗶𝗻 𝗨𝗻𝗿𝗲𝗮𝗹 𝗘𝗻𝗴𝗶𝗻𝗲 and know that procedural foliage generation shares roots with forest growth theory — and can be repurposed for ecological research? Do you train 𝗮𝘂𝘁𝗼𝗻𝗼𝗺𝗼𝘂𝘀-𝗱𝗿𝗶𝘃𝗶𝗻𝗴 𝗼𝗿 𝗿𝗼𝗯𝗼𝘁𝗶𝗰𝘀 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 𝘄𝗶𝘁𝗵 𝘀𝘆𝗻𝘁𝗵𝗲𝘁𝗶𝗰 𝗱𝗮𝘁𝗮, and know that a similar approach can scale up forest vision systems — an essential infrastructure for understanding global carbon cycling and forest dynamics? We introduce 𝗖𝗮𝗺𝗯𝗿𝗶𝗱𝗴𝗲 𝗔𝗿𝗯𝗼𝗿𝗲𝗮𝗹 𝗠𝗼𝗱𝗲𝗹𝗹𝗶𝗻𝗴 𝗣𝗮𝗻𝗼𝗽𝘁𝗶𝗰 𝟯𝗗 (𝗖𝗔𝗠𝗣𝟯𝗗) — a new synthetic data generation pipeline and dataset. 📄 Preprint: <https://lnkd.in/gNGrYY72> 💻 Code & pipeline: <https://lnkd.in/gmJRd6Jr> 📦 Dataset: <https://lnkd.in/gNAwG5g2> What's new? • A 𝗽𝗶𝗽-𝗶𝗻𝘀𝘁𝗮𝗹𝗹𝗮𝗯𝗹𝗲 𝘀𝘆𝗻𝘁𝗵𝗲𝘁𝗶𝗰 𝗱𝗮𝘁𝗮 𝗽𝗶𝗽𝗲𝗹𝗶𝗻𝗲 for forest vision tasks that combines advances in game engines (Unreal Engine) and physics-based LiDAR simulation (HELIOS++). • A 𝗰𝗼𝗺𝗽𝗿𝗲𝗵𝗲𝗻𝘀𝗶𝘃𝗲, 𝗱𝗶𝘃𝗲𝗿𝘀𝗲, 𝘀𝘆𝗻𝘁𝗵𝗲𝘁𝗶𝗰 𝗟𝗶𝗗𝗔𝗥 𝗱𝗮𝘁𝗮𝘀𝗲𝘁 covering 12 forest scenes (coniferous + deciduous), totalling 75 hectares, with semantic (leaf/wood/ground) and instance labels for tree segmentation. • Experiments demonstrating the effectiveness of the dataset, and showing that 𝗽𝗵𝘆𝘀𝗶𝗰𝘀-𝗯𝗮𝘀𝗲𝗱 𝘀𝗶𝗺𝘂𝗹𝗮𝘁𝗶𝗼𝗻, 𝘀𝗰𝗲𝗻𝗲 𝗱𝗶𝘃𝗲𝗿𝘀𝗶𝘁𝘆, 𝗮𝗻𝗱 𝗱𝗮𝘁𝗮𝘀𝗲𝘁 𝘀𝗰𝗮𝗹𝗲 are all crucial for synthetic data to be effective in training. 🌍 We believe that synthetic data can scale up the learning of robust forest vision systems to help address the climate challenges facing our planet. We hope our pipeline and dataset will support the community in moving towards this goal. #SyntheticData #ComputerVision #ForestEcology #LiDAR #RemoteSensing #MachineLearning #Forestry #UnrealEngine #AI4Science #ClimateTech

Original URL: https://www.linkedin.com/posts/yihang-she-9b5240161_cambridge-arboreal-modelling-panoptic-3d-activity-7373654997395279872-D7rc?utm_source=share&utm_medium=member_desktop&rcm=ACoAACan8ZkBY8Mnmp2vaClJhUdbCW_RAZNYaNI

🙌 David Russell, Benjamin Hoffman, Nate Harada, Don Cosseboom, Josafat-Mattias Burmeister

🌲 Dan Morris, Nate Harada, Alexander Merdian-Tarko, Don Cosseboom, Elizabeth Campolongo

❤️ Josh Seltzer, Nate Harada, Don Cosseboom

😎 Jon Van Oast, Don Cosseboom, Sara Beery

Johannes Pichler (johannes.pichler07@gmail.com)

2025-09-17 12:14:28

🐀 Innovative funding opportunity - Humane rodent control technology by Nitin Sekar 🐁

TL;DR: Conservation X Labs (CXL) partnership developing fertility control alternatives to rodenticides for island conservation. Need $242k to unlock major foundation/corporate funding.

The Problem

Island eradications use rodenticides that harm non-target native species

Growing public backlash + impending EU bans creating market pressure

Conservation groups actively seeking better solutions

The Solution

CXL partnership launching innovation prize for fertility control technology

Target: Field-applicable solutions for island conservation

CXL track record: $12M prizes → $570M follow-on funding (45x ROI)

The Ask $242k bridge funding to secure full ~$1.4M program from biodiversity foundations and corporate partners. Without it, partnership dissolves.

🚨 Interested? Contact nitin@conservationxlabs.org

👍 Dante Wasmuht, Peter Bull

Anna Vallery (annacvallery@gmail.com)

2025-09-17 12:30:32

Hi All,

We're still hoping for feedback on what folks would like to see out of the WILDLABS AI for Conservation Group over the next year! Whether it's certain discussion topics, more webinars, networking time, or support for your ongoing projects, we're eager to hear what it is you need!

If you have 5 minutes, help guide the group by filling out this form. The deadline is October 5th!

If you have other thoughts or questions, feel free to message myself @Josh Seltzer or @Viktor Domazetoski here.

Thanks so much!

https://forms.gle/cxXPoF9YhUVyiqYa8

Google Docs

WILDLABS AI for Conservation - Group Interests

Original URL: https://forms.gle/cxXPoF9YhUVyiqYa8

👀 Viktor Domazetoski, Pen-Yuan Hsing, Sara Beery, Alexander Merdian-Tarko, Ross Gardiner

👍:skin_tone_3: Pen-Yuan Hsing

Robert Long (Robert.Long@zoo.org)

2025-09-17 16:59:51

Hi folks,

Just joining the channel and wanted to say hello. I’m a carnivore biologist with Woodland Park Zoo in Seattle, so more on the “use” side of AI tools than the “creation” side. Background with noninvasive wildlife survey methods (especially camera trapping), habitat connectivity analysis, wildlife survey design, road ecology, urban ecology, and conservation technology. I know some of you, and look forward to learning lots from all of you!

😎 Jason Holmberg (Wild Me), Gracie Ermi, Sara Beery, Ben Seleb

👋 Matt Weldy, Jon Van Oast, Andy Huynh, Viktor Domazetoski, Siddharth Mayya, Dan Morris, Boe Zhou, Anna Vallery, Klára Janoušková, Alexander Merdian-Tarko, Shir Bar, Elizabeth Campolongo, Angela Zhu

👋:skin_tone_2: Cara Appel, Patrik Reiske

🦁 Irina Tolkova

Aleksis Pirinen (aleksis.pirinen@ri.se)

2025-09-19 06:45:46

🌍✨ Climate AI Nordics is thrilled to announce “Workshop on AI for Climate and Conservation (AICC)”, which has been accepted at EurIPS 2025 in Copenhagen! 🎉

📢 Call for Participation is now open!
Details, confirmed speakers, and program updates: https://sites.google.com/g.harvard.edu/aicceurips

📅 Key dates:

Submission deadline: October 10

Notifications: October 31

Camera-ready poster: November 28

Workshop: December 6 or 7 (TBD) at the Bella Center, Copenhagen

💡 What’s AICC about?
The workshop will highlight open problems at the intersection of climate, conservation, and AI — while also celebrating success stories where AI has already made a positive impact. Our goal: to connect domain experts and ML researchers, spark collaboration, and inspire contributions to some of the most urgent challenges of our time.

👉 Confirmed speakers: Philippe Ciais (Laboratoire des Sciences du Climat et de l’Environnement), Sagar Vaze (Mistral AI), Verena Griess (ETH Zurich), Natalie Eva Iwanycki Ahlstrand (University of Copenhagen), Drew Purves (Google DeepMind)

:femalescientist::malescientist: Organizers: Joakim B. Haurum, Nico Lang, Oisin Mac Aodha, Lynn H. Kaack, Ankit Kariryaa, Isabelle Tingzon, Lucia Gordon, Aleksis Pirinen

📝 Program committee members: Christian Lange, Stefano Puliti, Olof Mogren, Sigrid P. Hellan, Malte Pedersen, Olaya Álvarez-Tuñón

🤝 Sponsors: Pioneer Center for AI + more TBA

We’re also looking for volunteer reviewers to make AICC a great success: https://forms.gle/ZXiwP2hQ8BwiosFP6

We look forward to seeing your contributions and discussions in Copenhagen!

#EurIPS2025 #ClimateAI #AIforConservation #Sustainability #ClimateAction #ArtificialIntelligence #ConservationTech #ClimateChange #AI4Good

--------------------------------------------

A LinkedIn post is here: https://lnkd.in/p/dWSeXeCC

A BlueSky post is here: https://bsky.app/profile/climateainordics.com/post/3lz6oal3pls2r

Please share far and wide 🙂

👍 Joakim Bruslund Haurum, Ghjulia Sialelli, Viktor Domazetoski, Arron Watson, Robin Zbinden, Jan Dirk Wegner, Hugo

🎉 Eugene Galaxy, Oisin Mac Aodha, Nico Lang, Alexander Merdian-Tarko, Sara Beery, Malte Pedersen, Robin Zbinden, Carl Boettiger, Thijs van der Plas, Gerard, Alan Stenhouse

Sara Beery (sara.beery@gmail.com)

2025-09-23 18:26:04

There's still time to apply to be an instructor for the CV4Ecology Program 2026! We’ll keep the application open, but we're planning to start reviewing the applications on Monday!

CV4Ecology (https://cv4ecology.caltech.edu/) is an interdisciplinary AI education program focused on building AI expertise in the ecological community. Each CV4Ecology student applies to the program with their own interdisciplinary AI+Ecology project, and brings their own data. The instructors work hands-on mentoring small groups of students on their projects, helping them frame their ecological challenges as tractable AI projects, choose methodology, work with open source code and models, and define evaluation criteria.

This is an amazing and fulfilling opportunity to bridge the gap between research and practice and make a lasting impact, helping to train the next generation of ecologists to build their own AI tools. Come join our team!!

For more information on the logistics, instructor responsibilities, and how to apply, see here: https://drive.google.com/file/d/1i3E5H0sopjCsL3aE-jC45FYKub_EuJQO/view?usp=sharing

📣 Oisin Mac Aodha, Elena Sierra, Sam Lapp, Jose Ruiz-Munoz

James Brown (jamesbrown@lincoln.ac.uk)

2025-09-24 06:01:17

Hi there! Myself and colleagues at the University of Lincoln (UK) are organising a stakeholder workshop for AI-Hab – a project to develop automated habitat type classification methods in collaboration with UK Centre for Ecology & Hydrology (UKCEH). This project is funded by the UK's Natural Environment Research Council.

This workshop will give you the opportunity to:
🧪 Test and provide feedback on the AI-Hab model prototype
💡 Share ideas for future development and applications
🤝 Connect with others working across ecology, conservation, and technology

Your insights will directly shape the design, usability, and future scope of AI-Hab. We’re also keen to grow our network and welcome new voices into the conversation.

🗓️ Save the date: 30th October at 10 AM GMT
👉 Sign up here: https://forms.office.com/e/0Hrzz9Ekjp

We look forward to collaborating with you and building the next generation of tools for biodiversity monitoring together. Please share with your networks!

📣 Oisin Mac Aodha, Aleksis Pirinen

Adrien Pajot (adrien.pajot@wildlabs.net)

2025-09-24 13:57:44

The first International Conservation Technology Conference!

Lima, Peru - 18-20 February 2026

Hey there!

Along with a coalition of global and local leading organisation, WILDLABS is pleased to share this Save The Date with you!

The call:
https://www.linkedin.com/posts/wildlabs-community_technology-biodiversity-conservation-activity-7376652050388115456-v_sQ?utm_source=share&utm_medium=member_ios&rcm=ACoAACVxhRsBnmIWJI0ZSB43TuYjJiOEABCr3Vc

Share your interest with this form: https://docs.google.com/forms/d/e/1FAIpQLSfWjrH3Z35iznWo2lNP60EEFBOxoXCYTwpgtzIP5WY4kyMzGQ/viewform

Learn more here: https://wildlabs.net/article/save-date-first-international-conservation-technology-conference

Looking forward to seeing this community there!

🎉 Justin Kay, Irina Tolkova, Elena Sierra, Andy Viet Huynh, Sara Beery, Andy Huynh, Shir Bar, Jon Van Oast, Josh Seltzer, Kalindi Fonda, Eugene Galaxy, Cameron Trotter, Viktor Domazetoski, Carly Batist, Dan Morris, Matt Hron, Jose Ruiz-Munoz, Alexander Merdian-Tarko, Carly Batist, Ed Miller, Antón Álvarez, Aleksis Pirinen

Ben Williams (ben.williams.20@ucl.ac.uk)

2025-09-25 05:07:20

Next Thursday a few us will be speaking on a webinar hosted by the International Coral Reef Initiative: Harnessing Artificial Intelligence to Monitor Coral Reefs 🪸 Topics will include, vision models for benthic classification, 3D modelling, acoustic monitoring and more! 8am EDT, 1pm BST 3pm AST, 10pm AEST.

https://icriforum.org/events/forcoral-webinar-ai-monitoring/

💙 Gerard, Juan Sebastián Cañas, Sonny Burniston, murilo, Rupa Kurinchi-Vendhan, Antón Álvarez, Nathaniel Rindlaub, Risa Shinoda

🪸 Shir Bar, Juan Sebastián Cañas, Viktor Domazetoski, Patrik Reiske, takatomi-k, Yuyan Chen, Levi Cai, Rupa Kurinchi-Vendhan, Alexander Merdian-Tarko, Risa Shinoda, Jonathan Sauder

Sara Beery (sara.beery@gmail.com)

2025-09-27 16:21:41

Anyone going to ICCV want to help organize the BIRD WALK?????

Sara Beery (sara.beery@gmail.com)

2025-09-27 16:22:52

Hawaii bird walk will go hard! We need to pick a morning to do it, pick a cool bird location near the conference center, and make a flyer!

Risa Shinoda (shinoda.lisa.47z@gmail.com)

2025-09-28 07:55:28

I think I can help you 🙌

Sara Beery (sara.beery@gmail.com)

2025-09-28 09:31:15

Awesome! I'll DM you

👍 Risa Shinoda

Mohamed Elhoseiny (mohamed.elhoseiny@gmail.com)

2025-10-15 03:05:59

would lov to join gthis time.

Youssef Bayouli (youssef.projetnemo@gmail.com)

2025-09-28 07:14:37

Hey community

does anyone have suggestions or experience with AI-based approaches to count animal populations, ideally with the ability to estimate the sex ratio as well?

I'm exploring methods that can automate or assist in both counting and sex identification, and would love to hear about any tools, models, or datasets you've come across.

Thanks in advance!

🐘 Peter van Lunteren

Peter van Lunteren (peter@addaxdatascience.com)

2025-09-28 07:56:26

Hi @Youssef Bayouli, can you tell us a bit more about the data? Are you referring to drone / overhead / aerial imagery? In that case, have a look at https://github.com/Alexandre-Delplanque/HerdNet

Youssef Bayouli (youssef.projetnemo@gmail.com)

2025-09-28 08:15:33

Yes exactly ! I mean aerial images , mainly from small planes or drones .
Thank you so much for sharing this , I will check it out .

Youssef Bayouli (youssef.projetnemo@gmail.com)

2025-09-28 08:17:48

Also thinking of combining many sources of truth like camera traps aerial images acoustic if needed and historic data to tackle the same challenge

Peter van Lunteren (peter@addaxdatascience.com)

2025-09-28 09:28:29

Population counts are always difficult, herd counts less so. For population counts you basically need some form of individual identification so you can do a mark recapture study to estimate population size

Roberta Hunt (r.hunt@di.ku.dk)

2025-09-29 08:32:19

Maybe this is relevant from Ben Koger a couple years back https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/1365-2656.13904

Youssef Bayouli (youssef.projetnemo@gmail.com)

2025-09-29 08:43:10

@Peter van Lunteren thank you so much for your valuable insights , I will definitely explore the approach you mentioned to estimate the population size . I am facing a problem to run the HerdNet project on my macos , as some packages to be installed should have windows env . Is there anyone who have faced the same issue?

Youssef Bayouli (youssef.projetnemo@gmail.com)

2025-09-29 08:43:40

Hi @Roberta Hunt ! Thank you so much I will check it out :)

Nicholas Osner (nicholasosner@gmail.com)

2025-10-07 10:26:13

Hi @Youssef Bayouli, WildEye Conservation has an AI-powered web application for processing aerial survey data called SurveyScope. At present, we use models to detect wildlife (and classify to a lesser extent). However, we don't have any models for identifying animal sex (and aren't aware of any), so we are having that sort of data (and other demographic data) manually annotated on the surveys we are assisting with where the end goal is to develop a model once the training data exists.

https://wildeyeconservation.org/surveyscope/

Peter van Lunteren (peter@addaxdatascience.com)

2025-10-10 13:04:53

Hi @Nicholas Osner , I usually working with camera traps, but lately getting more and more into the aerial world. I’m keen to learn more and would love to try out SurveyScope. Can I just create an account for some trials? Are WildEye’s model weights also released open-source? Thanks!

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2025-09-29 18:12:27

Tuesday, 14 October 2025
15:00–17:00 CET | Online
This event will bring together policy practitioners, researchers, and advocates to exchange strategies and thinking around pressing issues, including:

Reshaping narratives and building solidarity between climate justice and digital rights at COP30 and beyond;

Strengthening strategies to hold the line on hard-won protections in the era of the EU’s de-regulation agenda.

Engaging in “Digital Sovereignty" debates with a strong vision for equitable futures;

Exploring visions for “Public AI” through the lenses of openness and equity.

As a fellow in Aspiration Tech's recent AI Openness and Equity Leadership Cohort, I can attest to the fact that this should be a very thoughtful/inclusive discussion on so-called "AI" beyond the hype!

🎉 Jon Van Oast

Pen-Yuan Hsing (penyuanhsing@posteo.is)

2025-09-29 18:17:28

Repost from WILDLABS relevant to this Slack group:
The Open Source Solutions group at WILDLABS is looking for feedback on what the community would like to see this year. Open source solutions overlaps with most/all other WILDLABS groups, and we'd love to know how we can help serve the wider conservation tech community outside of WILDLABS as well.
So, please consider completing and sharing/amplifying our OSS Survey in your group!

Sky Sheng (skysheng7@gmail.com)

2025-09-30 18:33:55

Hi friends, we are looking for data science mentors who are interested in cultivating the next generation of young data scientists! The UBC Vancouver Master of Data Science (MDS) program will be starting the sixth year of its mentoring program this November. If you are interested in sharing your experiences and providing invaluable insights/advice to current MDS students, consider becoming a mentor. You will also have the opportunity to network with other mentors and professionals during the events we will organize as part of the program. Watch this short video to hear more about the program: https://youtu.be/y3cOB2TA1lk

We are currently recruiting Data Science professionals who are interested in mentoring current students from the MDS program. Please note you will be paired with one or two mentees and must be able to commit an estimated total of 10-15 hours of time to this program.

If you are interested in being a mentor, please complete the online application (link below) by end of day Friday October 17. Here is the application link: https://ubc.ca1.qualtrics.com/jfe/form/SV_0VQNP56ETHr9yxU

Your required time commitment includes:

If available, attend the November 18 kickoff event and April 2026 program wrap up event.

A total of two meetings with your mentee(s) (or more if you wish; the first meeting ideally being held within a month of the kickoff). You have the option of holding your meetings in-person or online.

Please note there is no guarantee that if you apply that we will be able to match you with a student.

YouTube

UBC Master of Data Science (https://www.youtube.com/@ubcmasterofdatascience)

Introducing the UBC MDS Mentoring Program

Original URL: https://youtu.be/y3cOB2TA1lk

ubc.ca1.qualtrics.com

MDS Mentor Application 2025

The most powerful, simple and trusted way to gather experience data. Start your journey to experience management and try a free account today.

Original URL: https://ubc.ca1.qualtrics.com/jfe/form/SV_0VQNP56ETHr9yxU

👍 Xiangteng He, Don Cosseboom, Alexander Merdian-Tarko

Tanya Berger-Wolf (berger-wolf.1@osu.edu)

2025-10-01 15:56:51

https://www.nytimes.com/2025/10/01/obituaries/jane-goodall-dead.html?unlocked_article_code=1.qE8.ygHx.aNBLyFbdTCMF&smid=nytcore-ios-share&referringSource=articleShare

The New York Times

By Keith Schneider

Jane Goodall, Eminent Primatologist Who Chronicled the Lives of Chimps, Dies at 91

Her discoveries in the 1960s about how chimpanzees behaved in the wild broke new ground and represented what was called “one of the Western world’s great scientific achievements.”

Original URL: https://www.nytimes.com/2025/10/01/obituaries/jane-goodall-dead.html?unlocked_article_code=1.qE8.ygHx.aNBLyFbdTCMF&smid=nytcore-ios-share&referringSource=articleShare

😭 Chris Smith, Sudhanshu Agarwal, Yuyan Chen, Shayan Azhar, Polina Stucke, Mitchell Rogers, Jess Tam, Boe Zhou, Patrik Reiske, gyeongsu, Carly Batist, Moos Middelkoop, Prabath Gunawardane, Brittany Aguilar, Don Cosseboom, Sky Sheng, Lucy Dimitrova, Alan Stenhouse

💗 Irina Tolkova, Eric Cunningham, Yseult Hb, Levi Cai, Lily Xu, Aaron Serianni, Owen Martin, Boe Zhou, Suman Ghosh, Sara Beery, Chris Lange, Devis Tuia, Hayley Smith, Don Cosseboom, Dan Cher

😢 Elena Sierra, Shir Bar, Aaron Serianni, Simon Ilić, Alexander Merdian-Tarko

🥺 Shayan Azhar, Don Cosseboom, Valentin Gabeff

Josh Seltzer (jyseltz@gmail.com)

2025-10-03 10:26:50

Looking to meet up with some fellow AI for conservation peeps from around the world?

Join us next Friday for our (rescheduled) AI For Conservation — Informal Networking event!
(Registration is quick & free.)

It'll be a great opportunity to meet other people, share what you're working on (or would like to be working on), without any pressure or presentations. @Anna Vallery & @Viktor Domazetoski will be joining me to kick off the discussion, and we hope to see you there!

luma.com

AI For Conservation — Informal Networking · Luma

Grab your morning coffee (or evening beer) and join us for a casual hour of chatting about how AI is being used to protect wildlife, track ecosystems, and…

Original URL: https://luma.com/mgjy4rtl

🎉 Anna Vallery, Xiangteng He, Navodita Mathur, Viktor Domazetoski, Prabath Gunawardane

Dan Morris (agentmorris@gmail.com)

2025-10-03 11:54:04

A couple of weeks ago, I asked folks for some of their favorite camera trap images; I was going to use a combination of an LLM and human labeling (i.e., me) to rate a bunch of camera trap images, then train a local model to score new images. Sara said (paraphrasing) "VLMs are pretty good at this, maybe you should try that instead of training a model" and I said (paraphrasing) "I don't know, I'm stuck in 2021 and I want to train a model", and it turns out Sara was a 100% right: VLMs are very good at this, even local VLMs (as long as you use relatively giant local VLMs), and IMO there's no reason to train a model. I didn't really build a full working pipeline, but I did a vibe-coding-party-level prototype that convinced me VLMs are the way to do things like this. Posting back here to earn Karma since people sent me images.

Specifically, I did the following (I wrote zero code manually to do this, vibe coding FTW):

I started with ~3M images from Snapshot Serengeti, on which I already had MD+SpeciesNet results. I gave AI a link to the file format for those results, and a pointer to the results files and the images, and I asked it to apply some heuristics to select images that are likely to represent a range of aesthetic quality. Things like "a large animal near the middle of the image is probably good" and "only very tiny animals in the distance is probably bad". A bunch of heuristics like that, but I was not looking just for images that checked the boxes, rather a mix of images that did and didn't check the boxes, just to get a range of quality (because at the time I still thought I was going to train a model, so I needed both good and bad images for training). I ran that code to sample 5k images. AI was awesome at just digesting file formats from descriptions, I didn't have to explain anything about how MD/SpeciesNet results files work. Bananas.

Using Python code that AI wrote for me, I gave those 5k images to (a) Gemini 2.5 Flash, via the Gemini Batch API, and (b) a whole bunch of local VLMs, via vLLM and Ollama, and a prompt like "you are an expert wildlife photography curator evaluating camera trap images for aesthetic appeal...". I asked for scores from 0 to 10 for each image, and I asked for explanations for each score. I did zero prompt engineering. AI was amazing at figuring out how to use the Gemini Batch API, and at getting me up and running with vLLM and Ollama for local VLMs. Also vLLM and Ollama are both bananas-good pieces of software.

I asked AI to write another script to make nice Web pages showing the results in a format that made it reasonable to compare them.

You can see the results for the six models I tinkered with here:

https://lila.science/public/misc/hero-image-samples/snapshot-serengeti-by_score/

In that link, each model's page is sorted by score, so you'll see its "favorites" at the top. In this link, they're sorted by filename, which makes it easier to compare models if you open them in tabs:

https://lila.science/public/misc/hero-image-samples/snapshot-serengeti-by_filename/

For example, here's the output from Gemini 2.5 Pro, Gemini 2.5 Flash, and Quen2.5VL-72B, respectively, sorted by score:

https://lila.science/public/misc/hero-image-samples/snapshot-serengeti-by_score/gemini_sync_labels_gemini-2.5-pro_20251008_093911.html
https://lila.science/public/misc/hero-image-samples/snapshot-serengeti-by_score/gemini_batch_labels_20250925_210142.html
https://lila.science/public/misc/hero-image-samples/snapshot-serengeti-by_score/ollama_local_labels_20250928_190018.html

I find the explanations to be super interesting, and overall I think all three do an awesome job picking good images. Scroll to the bottom if you need to be reminded how aesthetically awful random camera trap images are. Subjectively, I found these models to be a cut above all the smaller models I tried.

In some cases, it's like Gemini is just showing off when it casually identifies a hartebeest, e.g.:

"This image boasts exceptional composition, with the prominent hartebeest perfectly centered and making direct eye contact, creating a powerful and engaging focal point. The low-angle shot against the bright, clear sky emphasizes the animals' stature and provides excellent lighting, making the image sharp and visually impactful. The presence of other hartebeest around the main subject adds depth and context without detracting from its hero status."

Overall, this was a really interesting vibe coding exercise, and a fun way to learn about running VLMs. Also this is the first time I felt like the 2x4090 setup - which seemed huge when I built it - was limiting: the 72B parameter Qwen2.5-VL model is the largest I could fit on my GPUs, but it's not even nearly the largest open-weights VLM available (e.g. Qwen3-VL came out last week at 235B (!) parameters).

Code is here, in case anyone wants to pick this up and tinker with prompts more:

https://github.com/agentmorris/hero-images

🙌 Boe Zhou, Bowen Zhang, Sara Beery, Elena Sierra, Justin Chung Clark, Malte Pedersen, Burak Ekim, Owen Martin, Xiangteng He, Jason Holmberg (Wild Me), Mitchell Rogers, Eugene Galaxy, Justin Kay, Rupa Kurinchi-Vendhan, Hayley Smith, Angus Galloway, Elizabeth Campolongo, Hugo Magaldi

🤯 Shir Bar, Sonny Burniston, Ben Seleb, Jon Van Oast, Malte Pedersen, Sam Lapp, Jason Holmberg (Wild Me), Justin Kay

👏 Bowen Zhang, Sonny Burniston, Jason Holmberg (Wild Me), Yseult Hb, Justin Kay, Juan Sebastián Cañas

💯 Bowen Zhang, Owen Martin, Justin Kay, Sky Sheng

😍 Jon Van Oast, Mitchell Rogers, Justin Kay, Sky Sheng

🙌:skin_tone_5: Prabath Gunawardane

🙌:skin_tone_3: Alan Stenhouse

🔍 Kalindi Fonda

Ben Weinstein (benweinstein2010@gmail.com)

2025-10-03 14:10:29

The killer one for me is this one, the sense that there is a 'main' wildebeast, and that its executing a behavior associated with that label

💯 Sara Beery, Dan Morris

Angus Galloway (agalloway@septicsitter.com)

2025-10-07 14:22:44

Really enjoyed your write up!

Sky Sheng (skysheng7@gmail.com)

2025-10-04 14:06:41

📊 Got conservation data that needs analysis? UBC Master of Data Science students can help!

Do you have datasets collecting dust, or a data-driven conservation idea that needs extra hands? UBC’s Master of Data Science program is looking for Capstone partners for May–June 2025. Please note that we welcome remote partners too!

What we’re offering: Our MDS students spend the final 2 months of their program on real-world Capstone projects. Past partners have gotten reproducible pipelines, custom visualizations, interactive dashboards, and analysis tools—all well-documented for full reproducibility.
Check out past projects: https://ubc-mds.github.io/capstone/past_projects/

Why partner with us:

Fresh perspectives on your conservation data challenges

Rigorous analysis backed by faculty mentors

Deliverables you can actually use (pipelines, reports, dashboards, scripts)

Free data science support for research and projects that matters

Timeline:
Deadline for early feedback on proposals: October 15, 2025
Final deadline on proposals without feedback: November 30, 2025

Interested?

Submit a proposal: https://ubc.ca1.qualtrics.com/jfe/form/SV_6rqTW70L2PfkAt0
Partner info: https://ubc-mds.github.io/capstone/partner_info/
Full timeline: https://ubc-mds.github.io/capstone/timeline/
Questions? DM me or reach out to Maymay Milad at maymay@science.ubc.ca

(This could be a great fit for camera trap data, habitat analysis, climate datasets, or any conservation challenge that needs data science expertise!)

ubc-mds.github.io

Past Projects

For the MDS Capstone course

Original URL: https://ubc-mds.github.io/capstone/past_projects/

ubc.ca1.qualtrics.com

Online Survey Software | Qualtrics Survey Solutions

Qualtrics sophisticated online survey software solutions make creating online surveys easy. Learn more about Research Suite and get a free account today.

Original URL: https://ubc.ca1.qualtrics.com/jfe/form/SV_6rqTW70L2PfkAt0

ubc-mds.github.io

Partner Info

For MDS Capstone Projects

Original URL: https://ubc-mds.github.io/capstone/partner_info/

ubc-mds.github.io

Timeline

For MDS Capstone Projects

Original URL: https://ubc-mds.github.io/capstone/timeline/

❤️ Nicole Lopez, Sara Beery, Xiangteng He, Elizabeth Campolongo

Ghjulia Sialelli (ghjulia.sialelli@inf.ethz.ch)

2025-10-08 08:17:13

🎥 The recordings from the AI+Environment Summit in Zurich last week are now available on YouTube! Check it out: https://youtube.com/playlist?list=PLHeEcDvb8uC_5L3Ie4xb7UFMjTAIySInq&si=fb_TrsSU9P49mQY-

YouTube

AI + Environment Summit 2025

The AI + Environment Summit 2025 brought together researchers, innovators, and practitioners at Switzerland Innovation Park Zurich to explore one guiding que...

Original URL: https://youtube.com/playlist?list=PLHeEcDvb8uC_5L3Ie4xb7UFMjTAIySInq&si=fb_TrsSU9P49mQY-

👍 Binbin Xiang, Chijioke Ugwuanyi, Samantha Biegel, Emilia Arens, Valentin Gabeff, Gözde Cilingir, Nina van Tiel, Oisin Mac Aodha, Gerard, Sonny Burniston, Viktor Domazetoski, Lukas Drees, Alexander Merdian-Tarko, Robin Zbinden, Andy Viet Huynh, Jonathan Sauder, Carly Batist, Jan Dirk Wegner, Boe Zhou, Hiroto, Juan Sebastián Cañas

🙌 Shayan Azhar, Robin Zbinden, Andy Viet Huynh, Lily Xu, Aleksis Pirinen

🎉 Jon Van Oast, Andy Viet Huynh, Jan Dirk Wegner, Luhan, Aleksis Pirinen

👍:skin_tone_3: Alan Stenhouse

Finn Manning (hillary.b.manning@gmail.com)

2025-10-17 16:00:04

Hey hey: I was excited to check this out, but wanted to let you know I'm getting the following errors for the playlist and the videos linked on the website.

Ghjulia Sialelli (ghjulia.sialelli@inf.ethz.ch)

2025-10-17 18:12:28

Oh weird, I’ll investigate! Thanks for letting me know

:gratitude_thank_you: Finn Manning

Ghjulia Sialelli (ghjulia.sialelli@inf.ethz.ch)

2025-10-19 10:34:09

The link is working again! YouTube had blocked the channel (we still don't know exactly what happened), a decision we appealed, and the channel has been back online since yesterday afternoon. @Finn Manning

Finn Manning (hillary.b.manning@gmail.com)

2025-10-19 17:24:28

Yay, thank you @Ghjulia Sialelli!

Ben Weinstein (benweinstein2010@gmail.com)

2025-10-08 10:11:55

I am looking for informal test datasets to validate our workflow for reducing double counting in airborne object detection among overlapping images. Imagine you have a collection of photos of a herd of bison, flocks of birds, or colony of animals (seals, penguins, etc). These photos do not have precise enough geographic data to overlay them into an orthomosaic. Instead we have to use traditional structure-through-motion and feature matching to find similar angles and align detections to get a unique count within the set. I originally implemented this in https://github.com/Weecology/DoubleCounting and a user has picked it up and wants to bring into DeepForest toolset and take it further. I would be interested in knowing failure modes, how far we can push changes in angles and perspectives, maybe underwater (coral reefs?), and anything else that might make it break. Maybe @Fadel Seydou @Casey Youngflesh @Howard L Frederick @Anna Vallery @Maddie Hayes or others have just a few images for us to play with, ideas for improvements and interest. Also has some agriculture uses (https://www.mdpi.com/2073-4395/12/2/440).

MDPI

Culling Double Counting in Sequence Images for Fruit Yield Estimation

Exact yield estimation of fruits on plants guaranteed fine and timely decisions on harvesting and marketing practices. Automatic yield estimation based on unmanned agriculture offers a viable solution for large orchards. Recent years have witnessed notable progress in computer vision with deep learning for yield estimation. Yet, the current practice of vision-based yield estimation with successive frames may engender fairly great error because of the double counting of repeat fruits in different images. The goal of this study is to provide a wise framework for fruit yield estimation in sequence images. Specifically, the anchor-free detection architecture (CenterNet) is utilized to detect fruits in sequence images from videos collected in the apple orchard and orange orchard. In order to avoid double counts of a single fruit between different images in an image sequence, the patch matching model is designed with the Kuhn–Munkres algorithm to optimize the paring process of repeat fruits in a one-to-one assignment manner for the sound performance of fruit yield estimation. Experimental results show that the CenterNet model can successfully detect fruits, including apples and oranges, in sequence images and achieved a mean Average Precision (mAP) of 0.939 under an IoU of 0.5. The designed patch matching model obtained an F1-Score of 0.816 and 0.864 for both apples and oranges with good accuracy, precision, and recall, which outperforms the performance of the reference method. The proposed pipeline for the fruit yield estimation in the test image sequences agreed well with the ground truth, resulting in a squared correlation coefficient of R2apple = 0.9737 and R2orange = 0.9562, with a low Root Mean Square Error (RMSE) for these two varieties of fruit.

Original URL: https://www.mdpi.com/2073-4395/12/2/440

👍 Casey Youngflesh, Aleksis Pirinen

Fadel Seydou (fadel.seydou@gmail.com)

2025-10-08 10:20:38

Hi @Ben Weinstein
We are working on releasing some data pretty soon hopefully. In the meantime, you can have a look at savmap dataset if you haven't done so aleady. I have attempted to reduce double counting using GPS data when available. I'd love to know if that could be useful to you as well

👍 Ben Weinstein

Howard L Frederick (simbamangu@gmail.com)

2025-10-08 12:25:38

@Ben Weinstein super worthwhile as overlap removal is a big part of the workflow; I’ll send you some.

Anna Vallery (annacvallery@gmail.com)

2025-10-09 11:04:07

Hi Ben, this is absolutely a critical part of the workflow and I'd be happy to send some images over! Are there specific guidelines for the types of image sets that would be helpful for testing it out?

Ben Weinstein (benweinstein2010@gmail.com)

2025-10-09 12:44:19

just a small handful of images taken of one central source, like a bird colony. We don't need any ground truth or counts yet, just informal, does this work and should we pursue it further.

👍 Anna Vallery

Dan Morris (agentmorris@gmail.com)

2025-10-08 13:50:15

New dataset on LILA, thanks to California Fish & Wildlife:

https://lila.science/datasets/california-small-animals/

Downward-facing small animal cameras continue to be a significant gap in the space of available camera trap data and models, and this dataset (~2.3M images from downward-facing cameras) goes a long way toward closing that gap, especially when combined with the Ohio Small Animals dataset that was released earlier this year.

As always, camera trap datasets on LILA should only sort of be viewed as independent entities; this has also been mapped into the shared taxonomy to make it easier to train models across datasets, in this case primarily across the Ohio Small Animals and California Small Animals datasets.

🎉 Jon Van Oast, Cara Appel, Shir Bar, Timm Haucke, Alessandra Vidal Meza, Peter Bull, Fagner Cunha, Justin Kay, Alan Stenhouse, Jason Holmberg (Wild Me), Carly Batist, Prabath Gunawardane, Brooke Goodman, Aakash Gupta, Dasha Simons, Valentin Gabeff, Elizabeth Campolongo, Aleksis Pirinen

😮 Jon Van Oast, Timm Haucke, Peter Bull, Jason Holmberg (Wild Me)

🐝 Michael Bunsen, Jason Holmberg (Wild Me)

Jon Van Oast (jon@wildme.org)

2025-10-08 13:52:06

way to sell this dataset! 🥹

❤️ Cara Appel, Shir Bar, Timm Haucke, Peter Bull, Kalindi Fonda, Mitchell Rogers, Alan Stenhouse, Jason Holmberg (Wild Me), Brittany Aguilar, Luhan, Elizabeth Campolongo, Angela Zhu

Dan Morris (agentmorris@gmail.com)

2025-10-08 13:53:34

Credit to Lindsey from CDFW, who not only led all the hard work that allowed this dataset to happen, but - equally importantly - picked that thumbnail.

🎉 Jon Van Oast, Alan Stenhouse, Jason Holmberg (Wild Me), Elizabeth Campolongo

Michael Bunsen (notbot@gmail.com)

2025-10-08 16:37:26

Is anyone planning to attend the Entomological Society conference next month in Portland, Oregon? I will be there talking about data management and ML for automated insect monitoring!

https://entomology2025.eventscribe.net/index.asp?presTarget=3138022

I'd love to help organize an unofficial gathering related to AI for Conservation, if there interest.

Sara Beery (sara.beery@gmail.com)

2025-10-13 13:01:17

@Ben Weinstein is in Portland!

🙌 Michael Bunsen

👍 Michael Bunsen

Ben Weinstein (benweinstein2010@gmail.com)

2025-10-13 13:01:45

sure am

👍 Michael Bunsen

Angela Zhu (azhu@umass.edu)

2025-10-13 16:25:31

I'm in Corvallis, OR! I don't know anything about insects, but could make the drive up if y'all end up doing something

Michael Bunsen (notbot@gmail.com)

2025-10-14 20:47:45

Okay awesome!

I'll be in Corvallis next weekend for the PNW Lepidopterists Workshop on the OSU campus. We could do a pre-meet if you have time on Sunday afternoon. Or come Saturday 3-5pm if you want to see the insect collection!
https://osac.oregonstate.edu/events/46th-annual-pacific-northwest-lepidopterists-workshop

osac.oregonstate.edu

46th-annual-pacific-northwest-lepidopterists-workshop | Oregon State Arthropod Collection

Mark your calendar for the 45th Pacific Northwest Lepidopterists WorkshopThe official program for this year's event has yet to be created, if you are interested in presenting, please reach out to Paul Hammond (<a href="mailto:copablepharon@gmail.com">copablepharon@gmail.com</a>). The workshop is in person, but portions will be watchable online via Zoom. To attend 'virtually' you will need to pre-register (see below for preregistration). Highlighted events at this year's meeting:

Original URL: https://osac.oregonstate.edu/events/46th-annual-pacific-northwest-lepidopterists-workshop

Hiroto (hirotogoto522@gmail.com)

2025-10-13 11:49:52

Hi all,

My name is Hiroto and I recently joined the channel. I'm a graduate student at the University of British Columbia in the Integrated Remote Sensing Studio (IRSS), where I work with computer vision models to detect animals from drone footage. I'm thrilled to join this channel and learn from all of you, and would love to hear from others who are also working with drone data!

🙌 Josh Seltzer, Xiangteng He, Tim Zhou, Viktor Domazetoski, Shayan Azhar, Sara Beery, Sonny Burniston, Alexander Merdian-Tarko, Shir Bar, Don Cosseboom, Ethan White, Elena Sierra, Aleksis Pirinen

🙌:skin_tone_2: Cara Appel

Joe Nash (joe@jna.sh)

2025-10-13 16:25:51

Hey folks 👋

Just joined today (thanks Sara!). I’ve been working in the technology sector for the past 10 years, but have spent the last couple trying to apply myself to biodiversity conservation and related areas. I briefly got to help out WILDLABS in 2019 as project manager on the TECH HUB (3 month accelerator), and have just finished a MSc in biodiversity, wildlife, and ecosystem health from University of Edinburgh.

Given my software background, I found myself falling into applications of AI to conservation, mostly currently in the audio realm, where I’ve been doing some bioacoustics and species classification projects. Excited to be here and get to see all the incredible work everyone is doing!

👋 Sara Beery, Shayan Azhar, Viktor Domazetoski, Justin Kay, Josh Seltzer, Shir Bar, Andy Huynh, Jon Van Oast, Julia Chae, Elena Sierra, Alexander Merdian-Tarko, Paridhi Jain, Ethan White

Lily Xu (lily.x@columbia.edu)

2025-10-13 17:36:02

Columbia University is launching a new Master of Science in Biodiversity Data Analytics to start in fall 2026! This 1-year program will focus on applying analytics and data science to nature especially for business, nature financing, and policy.

There will be a launch event next month with a panel on Why Biodiversity Analytics Matters Now, both in-person in NYC or remote.

Wednesday, November 5
6–8pm
Columbia University in NYC, or remote

😮 Angela Zhu, Robin Zbinden, Hemal Naik, Elena Sierra

🔥 Sara Beery, Justin Kay, Connor Jordan, Julia Chae, Jana Huisman, Carly Batist, Elena Sierra, Millie Chapman, Jenya Bogacheva, Elizabeth Campolongo

👍 takatomi-k, Judith Dekkers, Aleksis Pirinen

❤️ Owen, Jennifer, Cameron Trotter, Alan Stenhouse

Lily Xu (lily.x@columbia.edu)

2025-10-13 20:55:41

Info / registration: https://sps.columbia.edu/events/nature-numbers-why-biodiversity-analytics-matters-now

sps.columbia.edu

Nature in Numbers: Why Biodiversity Analytics Matters Now

Introducing Columbia University’s Master of Science in Biodiversity Data AnalyticsHow do we measure biodiversity? How do we bring nature data…

Original URL: https://sps.columbia.edu/events/nature-numbers-why-biodiversity-analytics-matters-now

Lorenzo Girotti (lollogiro2@gmail.com)

2025-10-14 10:13:54

Hi everyone!

I'm Lorenzo, a Master’s student in Artificial Intelligence (with a Bachelor’s in Computer Science) at the University of Bologna. I’ve recently been working on projects related to mammals and bioacoustics classification, and I’m now looking for an internship in ecological computer vision, which will form the basis of my master’s thesis.

I’m open to opportunities both in research labs or industry, and I’m particularly interested in anything related to wildlife monitoring, including insects, drones, or multimodal approaches. Looking forward to connecting and exploring potential collaborations!

👋 Sara Beery, Joe Nash, Jon Van Oast, Alexander Merdian-Tarko, Aleksis Pirinen

Karina Torres (karina.torres@siu.edu)

2025-10-14 16:30:21

Hi all! I am a first-year MS student in Forestry at Southern Illinois University, and I am passionate about integrating technology and conservation. My undergrad research focused on wildfire remote-sensing and moth species identification with machine and deep learning. For my MS thesis, I am particularly interested (at least for one chapter) in improving the extraction of secondary data from images, such as from iNaturalist, for ecological studies. I am still developing ideas for my thesis, so I am excited to be here and continue learning about different AI/tech applications in conservation! I am always happy to chat if you have similar interests.

👋 Andy Huynh, Jon Van Oast, Nate Harada, Joe Nash, Josh Seltzer, Dan Morris, Lorenzo Girotti, Jason Holmberg (Wild Me), Viktor Domazetoski, Jenya Bogacheva, Elizabeth Campolongo, Alexander Merdian-Tarko, Aleksis Pirinen, Hugo, Angela Zhu

Nathan Jacobs (jacobsn@wustl.edu)

2025-10-15 10:31:06

Welcome to the Slack. If you are interested, I can connect you with one of my PhD students at WashU who has been doing this kind of work.

Karina Torres (karina.torres@siu.edu)

2025-10-15 15:57:47

Thanks @Nathan Jacobs! I actually have some of your work on my to-read list, since I have been starting to explore some ideas with multi-modal SDMs and spatial observation biases. I would love to connect with one of your students!

Yip Fai Tse (tseyipfai@gmail.com)

2025-10-15 11:27:37

Hi everyone! I am a PhD student on bioethics and AI ethics in the National University of Singapore. I would like to share with you a piece of my publication that got online yesterday. Hopefully it's relevant and interesting to some of you! Thank you!

https://link.springer.com/article/10.1007/s13347-025-00979-1

SpringerLink

AI Alignment: The Case for Including Animals

Philosophy & Technology - AI alignment efforts and proposals try to make AI systems ethical, safe and beneficial for humans by making them follow human intentions, preferences or values....

Original URL: https://link.springer.com/article/10.1007/s13347-025-00979-1

👏 Viktor Domazetoski, Felipe Parodi, Jose Ruiz-Munoz, Vincent Lostanlen, Karina Torres, Lorenzo Girotti, Ragib Nihal, Cameron Trotter, Alexander Merdian-Tarko, Aleksis Pirinen

Carly Batist (carly.h.batist@gmail.com)

2025-10-15 11:56:01

Anyone else also attending Living Data Oct. 21-24 in Bogotá? :flag_co:
Would be great to meet up with AI folks during it 🙂

livingdata2025.com

Living Data 2025 - Information that together, transforms

Join us in Bogotá for Living Data 2025, exploring biodiversity data standards, integration, and community engagement.

Original URL: https://www.livingdata2025.com/

🔥 Sara Beery

Elly Knight (ecknight@ualberta.ca)

2025-10-15 12:01:29

I’ll be there Carly, would love to meet up!

❤️ Carly Batist

Matt Hron (matt.hron@wildlifeprotectionsolutions.org)

2025-10-15 12:32:41

I'll be there as well and would be happy to meet

❤️ Carly Batist

Wenxin Yang (wenxinyang@ucsb.edu)

2025-10-15 13:04:54

I will be there too and would be happy to meet! : )

❤️ Carly Batist

Kalindi Fonda (kalindi.fonda@gmail.com)

2025-10-15 13:36:05

Me too! 🥳

❤️ Carly Batist

Carly Batist (carly.h.batist@gmail.com)

2025-10-15 15:37:05

FYI - https://wildlabs.net/event/join-us-wildlabs-meetup-bogota-veni-al-encuentro-de-wildlabs-en-bogota

🌱 Kalindi Fonda, Wenxin Yang

😃 Elly Knight, Sara Beery

Owen Martin (owma6084@colorado.edu)

2025-10-15 20:14:40

I unfortunately cannot attend but it is such a cool looking event!

Molly Blank (mblank@naturalstate.org)

2025-10-22 13:44:54

I have acute FOMO and would love to know if there was anyone working on/presenting about metadata standards for observation-level/in situ data (e.g. images, audio files). I am interested in the work by B-cubed but they seem to be focused on occurrence time series (very awesome, not what I'm after).

Raj (therajsmile@gmail.com)

2025-10-16 10:32:21

Hello, just wondering about openly accessible fish datasets which can be used? I meant, fish videos from previous studies, or any other open source fish database which can be used to train for tracking? Thanks in advance. Specially moving fish, preferably from laboratory data.

Dylan Van Bramer (she/her) (dcv26@cornell.edu)

2025-10-16 12:12:18

Just sent you a DM!

Karina Torres (karina.torres@siu.edu)

2025-10-16 16:50:29

@Raj Have you looked at datasets on LILA?

LILA BC

Accelerating Scientific Research with Vision AI · Zoom · Luma

LILA BC - List of other conservation-related data sets

Written by

lilawp

Est. reading time

45 minutes

Original URL: https://lila.science/otherdatasets#images-marine

Dan Morris (agentmorris@gmail.com)

2025-10-16 17:19:59

A group of us (led by @Filippo Varini) have been working for a while to collect fish-related datasets (some of which include tracking information) here:

https://github.com/filippovarini/fish-datasets

That is a pretty structured repository that tries to collate all of those datasets into a common format; if you just want a list, it's largely derived from a more matter-of-fact list that I maintain here:

https://lila.science/otherdatasets#images-marine-fish

Raj (therajsmile@gmail.com)

2025-10-16 21:43:03

Thank you for help Dylan @Dylan Van Bramer (she/her).

Raj (therajsmile@gmail.com)

2025-10-16 21:43:45

Thanks for sharing the link Karina @Karina Torres.

Raj (therajsmile@gmail.com)

2025-10-16 21:44:43

Dan, thanks a lot for this information @Dan Morris. It is a nice information.

Raj (therajsmile@gmail.com)

2025-10-16 22:01:27

Thank you all for sharing information on fish datasets, greatly appreciated. Thanks!

Jana Huisman (janasannehuisman@gmail.com)

2025-10-16 10:39:20

Hi all,

Is there anyone here who has worked on using remote sensing or animal tracking data for microbiology or virology purposes? Specifically, I would be interested in using estimates of the density of animal hosts and/or their interaction networks (which could be the output of an AI model) to model the spread of infectious diseases or the risk of zoonotic transmission to humans.

My background is in microbial ecology and modeling infectious disease dynamics, but I am less familiar with these sources of data, so I’d be happy for pointers to any relevant work! Also very happy to exchange with anyone here who would be interested in these topics.

Aakash Gupta (aakash@thinkevolveconsulting.com)

2025-10-16 12:57:08

Accelerating Scientific Research with Vision AI

If you have millions of insect specimens, how do you digitize them and transform the imagery into valuable research data?
Join us for a live interview with Elizabeth Postema, Postdoctoral Researcher at the Field Museum of Natural History. We'll explore how the museum, along with researchers from several institutions, are using computer vision to digitize over 8 million insect specimens, transforming them into valuable data for large-scale scientific research.
Elizabeth will share insights into the development of DrawerDissect, a custom software solution that automates the transcription of specimen data and enables rapid analysis of insect features like color and size. She'll explain how they used Roboflow to build accurate vision models, accelerating a process that would have been impossible with traditional methods.

https://luma.com/zjm41bcl

luma.com

If you have millions of insect specimens, how do you digitize them and transform the imagery into valuable research data? Join us for a live interview with…

Original URL: https://luma.com/zjm41bcl

🐝 Michael Bunsen, Alan Stenhouse

👋 Michael Bunsen

👍 Michael Bunsen

Alan Stenhouse (alan.stenhouse@csiro.au)

2025-10-31 00:59:23

Hi @Aakash Gupta, was this recorded and made available online somewhere? Would be great to see it. Thanks!

Aakash Gupta (aakash@thinkevolveconsulting.com)

2025-10-31 07:10:59

@Alan Stenhouse
https://roboflow.com/case-studies/field-museum

roboflow.com

Case Study: Field Museum Accelerates Scientific Research with Vision AI

The Field Museum used vision AI to digitize their massive insect collection, accelerating scientific research, and opening new possibilities for scientists and curators.

Original URL: https://roboflow.com/case-studies/field-museum

🙌:skin_tone_3: Alan Stenhouse

Justin Kay (justinkay92@gmail.com)

2025-10-16 18:51:53

Join us for an AI for Conservation Happy Hour at ICCV! All are welcome. We will arrive at 6 and try to get some tables together.

🙌 Xiangteng He, Mitchell Rogers, Navodita Mathur, Cameron Trotter, Sara Beery, Subhransu Maji

🙌:skin_tone_3: Jess Tam

🐢 Rupa Kurinchi-Vendhan, Subhransu Maji, Anthony Hoogs, Malte Pedersen, Lukas Picek, Jonathan Sauder, Sara Beery

🔥 Sara Beery

Sara Beery (sara.beery@gmail.com)

2025-10-17 18:56:12

See you there!!

Johannes Pichler (johannes.pichler07@gmail.com)

2025-10-17 06:41:57

Postdoctoral Funding Opportunity: Rodent Fertility Control 🎓
Hi everyone!
Mal Graham (Strategy Director at Wild Animal Initiative) has a line on potential postdoctoral funding for someone interested in working on rodent fertility control research.

Requirements:

PhD in a relevant field (ecology, animal welfare, conservation, etc.)

What to do: If you're interested or know someone who might be, please send me a DM and I can share more details with you and connect you with Mal.

Please note: This is support to help with your application—funding is not guaranteed.

Sara Beery (sara.beery@gmail.com)

2025-10-17 19:24:10

BIRD WALK BIRD WALK BIRD WALK (IN HAWAII!!)

😮 Jon Van Oast

🐦 Dan Morris, Xiangteng He, Risa Shinoda, Rupa Kurinchi-Vendhan, Mitchell Rogers, Cheng Zhang, Shir Bar, Viktor Domazetoski, Cameron Trotter, Jess Tam, Jonathan Roberts, Anna Vallery, Hugo, Dasha Simons, Jenya Bogacheva

🚶 Mitchell Rogers

Sara Beery (sara.beery@gmail.com)

2025-10-17 19:28:35

I went back and forth a bunch on location, but since local transport in Waikiki can be a bit tricky decided to stay within walking distance of the conference center. We'll be taking an early-morning walk through Ala Moana Park, and will make it back before the first oral session!

Of course, there are lots of AMAZING places to see birds on the island, so consider this an opportunity to find bird buddies to coordinate bigger adventures with later in the week :)

Sara Beery (sara.beery@gmail.com)

2025-10-17 19:38:43

Help spread the word!

https://x.com/sarameghanbeery/status/1979330310811848805

https://bsky.app/profile/sarameghanbeery.bsky.social/post/3m3ggtvhfqk25

https://www.linkedin.com/posts/sara-beery-1ba89166_iccv2025-cvbirdwalk-activity-73850[…]mber_desktop_web&rcm=ACoAAA4DIN4BIW_3ubfcfe3GqvBT-NI--ZBxMzQ

X (formerly Twitter)

#iccv2025 #cvbirdwalk | Sara Beery

Are you coming to @ICCVConference?? Do you like BIRDS?!?! The #CVBirdWalk is happening!!! Oct. 21, 7am, Ala Moana Park. Kick off your conference right!! #ICCV2025 #AIforConservation

Original URL: https://x.com/sarameghanbeery/status/1979330310811848805

bluesky

Sara Beery (https://bsky.app/profile/sarameghanbeery.bsky.social)

Are you coming to <a href="https://bsky.app/profile/iccv.bsky.social">@iccv.bsky.social</a>?? Do you like BIRDS?!?! The #CVBirdWalk is happening!!! Oct. 21, 7am, Ala Moana Park. Kick off your conference right!! #ICCV2025 #AIforConservation

Original URL: https://bsky.app/profile/sarameghanbeery.bsky.social/post/3m3ggtvhfqk25

linkedin.com

Are you coming to #ICCV2025??? Do you like BIRDS?!?!? #CVBirdWalk is happening!!! Oct. 21, 7am, Ala Moana Park. Kick off your conference right!!

Original URL: https://www.linkedin.com/posts/sara-beery-1ba89166_iccv2025-cvbirdwalk-activity-7385096883725320192-YRnI?utm_source=social_share_send&utm_medium=member_desktop_web&rcm=ACoAAA4DIN4BIW_3ubfcfe3GqvBT-NI--ZBxMzQ

Sara Beery (sara.beery@gmail.com)

2025-10-17 19:48:58

And thanks a billion to @Risa Shinoda who helped me scout out what locations within walking distance to the conference center might have cool birds!!!!

🙌 Risa Shinoda, Justin Kay

❤️ Risa Shinoda

Xiangteng He (xiangteng.he@ubc.ca)

2025-10-19 13:51:16

I saw many birds on the way to the Convention Center.🐦

😎 Sara Beery

Malte Pedersen (mape@create.aau.dk)

2025-10-18 02:50:57

Hello to all the people attending ICCV. Why only go to a single happy hour, when you can get two two days in a row?! Join us for a no-host happy-hour/dinner at the Maui Brewing Company, Waikiki on Sunday 19th at 6.30pm!

❤️ Viktor Domazetoski, Cameron Trotter, Jess Tam, Rupa Kurinchi-Vendhan, Robin Zbinden, Hugo

Tanya Berger-Wolf (berger-wolf.1@osu.edu)

2025-10-19 10:24:22

https://www.linkedin.com/posts/connected-conservation-foundation_satellites-for-biodiversity-award-round-activity-7378386840975732736-zAYq?utm_medium=ios_app&rcm=ACoAAAC6PyAB_nuBo3L661sZgV2tdBwOiQJa_Co&utm_source=social_share_send&utm_campaign=copy_link

linkedin.com

Satellites for Biodiversity Award – Round 4 is open! :satellite: We’re calling on conservationists worldwide to harness the power of high-resolution satellite imagery, artificial intelligence, and local… | Connected Conservation Foundation

Satellites for Biodiversity Award – Round 4 is open! 📡 We’re calling on conservationists worldwide to harness the power of high-resolution satellite imagery, artificial intelligence, and local community knowledge—to protect our planet’s most vulnerable species and ecosystems. What you can win: ✅ Access to Airbus’s Pléiades & Pléiades Neo (INC. HD15) satellite data ✅ $6,000 ✅ ArcGIS Pro software from Esri ✅Mentorship and technical support from CCF and Airbus geospatial experts We’re looking for bold projects tackling wildlife conservation, invasive species, ecosystem management and restoration. Together, we can turn complex data into actionable strategies that protect biodiversity and enable community-led impact. 📅 Applications close 19 Dec 2025 👉 Apply now: <https://lnkd.in/eiRcJzf9> Airbus Defence and Space Airbus Geospatial and Secure Connectivity Solutions

Original URL: https://www.linkedin.com/posts/connected-conservation-foundation_satellites-for-biodiversity-award-round-activity-7378386840975732736-zAYq?utm_medium=ios_app&rcm=ACoAAAC6PyAB_nuBo3L661sZgV2tdBwOiQJa_Co&utm_source=social_share_send&utm_campaign=copy_link

🙌 Mitchell Rogers, Dasha Simons, Alexander Merdian-Tarko

❤️ Mitchell Rogers, Alan Stenhouse

Sara Beery (sara.beery@gmail.com)

2025-10-21 12:49:14

For the folks heading to the bird walk, rain is clearing! We will wait a few minutes at our starting point for any stragglers who were avoiding the downpour earlier

Sara Beery (sara.beery@gmail.com)

2025-10-21 13:09:46

Through a series of unfortunate events (downpour plus Áqqtwo Uber drivers cancelling on me) I am running quite late but will be there soon

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2025-10-21 14:20:42

I am bird walking with you all in spirit.

https://media3.giphy.com/media/v1.Y2lkPTc5MGI3NjExc2ViZnl0cnBubGRuMWRnN2ljbmhudWNodWh[…]12MV9pbnRlcm5hbF9naWZfYnlfaWQmY3Q9Zw/RzKHvdYC3uds4/giphy.gif

Original URL: https://media3.giphy.com/media/v1.Y2lkPTc5MGI3NjExc2ViZnl0cnBubGRuMWRnN2ljbmhudWNodWhtbmI5OGJzNHZoYmluaiZlcD12MV9pbnRlcm5hbF9naWZfYnlfaWQmY3Q9Zw/RzKHvdYC3uds4/giphy.gif

🦆 Malte Pedersen, Navodita Mathur, Sara Beery, Alan Stenhouse

Dan Morris (agentmorris@gmail.com)

2025-10-21 22:12:46

What was the most exciting sighting of the bird walk?

Sara Beery (sara.beery@gmail.com)

2025-10-21 23:07:00

For me probably kolea and manu-o-ku :)

😎 Jason Holmberg (Wild Me)

🐦 Dan Morris

Dan Morris (agentmorris@gmail.com)

2025-10-22 17:59:12

I confess I had to look both of those up.

Sara Beery (sara.beery@gmail.com)

2025-10-24 15:56:59

I learned them when I saw them! I never said I was a GOOD birder 🤣

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2025-10-24 16:36:23

EVIL birder?

Simone Franceschini (smn.franceschini@gmail.com)

2025-10-21 19:10:07

Hey everyone! I’m thrilled to share that I’ll be co-hosting (together with a group of fantastic scientists) an AI session at ICRS 2026 in Auckland (New Zealand):

#123 | The human–AI nexus: Scaling trustworthy coral reef monitoring and management

Artificial intelligence is transforming how we monitor and manage coral reefs, unlocking new capacity to process imagery, acoustics, and sensor data at scale. Yet ensuring these tools remain ecologically valid, equitable, and trustworthy is crucial for real-world impact.
We’re inviting contributions that span the full AI lifecycle, from ecological validation and community-science integration to Indigenous co-design, monitoring platforms, and policy dashboards.

Would love to see some of you there. It’s going to be an amazing space for conversations on how to make AI truly work for reefs and communities.

📢 Submit your abstract and join us at ICRS 2026 in Auckland!

https://www.icrs2026.nz/sessions-and-themes

ICRS 2026

Sessions and themes

Original URL: https://www.icrs2026.nz/sessions-and-themes

🪸 Malte Pedersen, Dan Morris, Sara Beery, Rupa Kurinchi-Vendhan, Shir Bar, Devis Tuia, Alexander Merdian-Tarko, Kalindi Fonda, Jonathan Sauder, Alan Stenhouse, Aleksis Pirinen

:flag_nz: Mitchell Rogers, Jess Tam

Alex Logan (alex@cecil.earth)

2025-10-22 12:41:34

Towards mutualism: reconciling differences between field and EO data by Chris, Lead Scientist of the Forest Ecosystems team at Planet
Super excited to have this newsletter out in the wild - it's the first time we've had an external team write the nature data newsletter. It's a great sign post for the partnership we have with Planet that dates back to the Salo Science days. So grateful for Dave and Chris' friendship and collaboration.

I thought this topic might be interesting to some folks in this Slack. So please read the newsletter and dive into the 15,000 words, 29 figures and 61 total pages of the manuscript 📡

🙌 Molly Blank, Anthony Miyaguchi, Nuno Sa, Aleksis Pirinen

Anthony Miyaguchi (acmiyaguchi@gmail.com)

2025-10-23 15:49:19

Cool to see Planet contributing to the community, I work on infra there so it’s nice to see the outward impact that folks are making

🙌 Alex Logan

Catherine Breen (cbreen@uw.edu)

2025-10-24 13:07:45

For anyone in the DC area, WILDLABS Geospatial will have a meetup on Wednesday Nov 5 at Franklin Beer Hall!! Feel free to spread the word! Register here so we can get a headcount!

❤️ Carly Batist, Jason Holmberg (Wild Me), Shir Bar, Yseult Hb, Sara Beery, Aleksis Pirinen

Youssef Bayouli (youssef.projetnemo@gmail.com)

2025-10-25 06:46:14

WILDLABS Edge Computing Group: Kickoff Meetup : D

Hey everyone! We’re starting something new , a space for anyone working with edge tech in conservation. If you’re deploying AI models in the field, setting up sensor networks, or experimenting with on-device intelligence, this is for you.
Our first meetup is a chance to connect, swap stories, and share ideas (and challenges!) with others pushing the boundaries of what’s possible at the edge. Let’s explore how real-time, resource-efficient, and autonomous tech can make conservation smarter and more effective.

📅 When: 4 November 2025
🕚 Time: 11:00 AM – 12:00 PM EST (UTC-5)
🔗 Register here: https://wwfus.zoom.us/meeting/register/NVIZALAPRqeMACFesSxwSQ#/registration
Link to the event page : https://wildlabs.net/event/wildlabs-edge-computing-group-meetup

Come share your projects, meet others in this space, and help shape the start of the WILDLABS Edge Computing community.

🎉 Ed Miller, Jason Holmberg (Wild Me), Vanesa Reyes, Anna Vallery, Prabath Gunawardane, Sara Beery

Ed Miller (ed@hypraptive.com)

2025-10-25 17:24:01

Be careful on the time! Daylight savings in the US ends on November 2, so this sessions should be in EST time as noted in the image, not EDT as noted in the text!

🙌 Youssef Bayouli

Jenna Kline (jennamkline@gmail.com)

2025-10-27 11:49:47

Thanks for the catch Ed! The event is 11 AM EST (UTC-5)

👍 Ed Miller

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-10-27 09:02:16

Hi Sara, just saw the presentation of Ciira in the AI4Good - cool stuff! thanks for setting it up!

Puneet Kollipara (kollipython@gmail.com)

2025-10-27 17:44:28

🚨 Last call for advance registration for FOSS4G North America 2025, taking place Nov. 3–5 just outside Washington, DC near Washington-Dulles Int'l. Airport. Save $100 on general admission to this gathering of the open-source geospatial/GIS tech community by registering in advance instead of on-site. Day passes are also available, and students save 50% on general admission! REGISTER HERE.

Why should the AI for Conservation community attend FOSS4G NA? Here's just a few of the many reasons:

Learn from leading figures in GeoAI and remote sensing for planetary health: Hundreds of geospatial/GIS developers, ML/AI engineers, executives, researchers and other practitioners across North America will attend FOSS4G NA, where you'll have no shortage of chances to meet with and learn from them through talks, workshops and community-building socials.

Numerous sessions on 11/4–11/5 tackling geospatial data and GeoAI for conservation:

Open-Access High-Resolution Data for a Livable Planet

Atlas: Postgres Meets GeoAI at Scale

Microsoft’s Planetary Computer: Building a Planetary-Scale Data Platform

Remote Sensing for Plants That Are Hard to See

Open-Access High-Resolution Data for a Livable Planet

And more! Check out the full schedule.

Gain new skills, career insights and business connections: Whether you hope to hire talented geospatial data/AI professionals, find research collaborators, sharpen your skills or explore career opportunities, check out these activities on Monday, 11/3:

Technical workshops, all day: $100 each, tickets limited. Optional add-on to general admission.

Geo-Career Expo & Ice-Cream Mixer, 1:00–3:00 PM: FREE, and general admission not required, separate RSVP.

B2B session, 4:30–6:00 PM: $80, tickets limited, and general admission not required.

See many of you next week near DC for FOSS4G NA 2025!

Filippo Varini (fppvrn@gmail.com)

2025-10-28 12:43:30

🐟 New release: Community Fish Detector (CFD)

Hey everyone, we’ve just released the Community Fish Detector (CFD), a YOLOv12x model trained with the goal of detecting any fish, anywhere, under a single class: “fish.”

The CFD model was trained on the Community Fish Detection Dataset, built thanks to the incredible collective effort of folks from this community, ultimately combining >1.9M images and >935k fish bounding boxes from 17 open datasets spanning marine, freshwater, and lab environments.

You can find the model and instructions on how to run it here: https://github.com/WildHackers/community-fish-detector

This is the first version. It performs reasonably well across many environments, but there’s still lots of room to improve.

If you’d like to contribute or build on it, we’d love your help!

A huge thank you to all the contributors: @Dan Morris, @Oc, @Sonny Burniston, @Kevin Barnard, @Laura Chrobak, @Alexander Merdian-Tarko, @Devi Ayyagari, Mona Dhiflaoui, Joshua Chen, @Levi Cai, @Adrian Ibanez, Aditya Shirvalkar, Chris Jackett and others who collaborated!

🐟 Levi Cai, Malte Pedersen, Timm Haucke, Shir Bar, Sonny Burniston, Atul Ingle, Rupa Kurinchi-Vendhan, Justin Kay, Dan Morris, John Walsh, Brian Geuther, Gaspard Dussert, Viktor Domazetoski, Nate Harada, Maria Napolitani, Hiroto, Sara Beery, Prabath Gunawardane, Alan Stenhouse, Jenya Bogacheva, Carly Batist, Elizabeth Campolongo, Aditya Shirvalkar, Sofía Miñano

🐠 Levi Cai, Timm Haucke, Shir Bar, Sonny Burniston, Atul Ingle, Rupa Kurinchi-Vendhan, Justin Kay, Alexander Merdian-Tarko, Nate Harada, Sara Beery, Jenya Bogacheva, Elizabeth Campolongo

🎉 Levi Cai, Malte Pedersen, Timm Haucke, Shir Bar, Sonny Burniston, Rupa Kurinchi-Vendhan, Justin Kay, Cameron Trotter, Kevin Barnard, Ross Gardiner, Sal Keith, Elizabeth Campolongo

🎣 Atul Ingle

:fishcowboy: Laura Chrobak, Joost Daniels

Levi Cai (levi@levicai.com)

2025-10-28 12:47:26

Sweet!!! The github link seems to be dead for me though?

Filippo Varini (fppvrn@gmail.com)

2025-10-28 12:50:52

Can you try now @Levi Cai?

Jon Van Oast (jon@wildme.org)

2025-10-28 12:52:30

links seems good now (to me). this is great - thanks!

🙌 Filippo Varini

Gerard (gerardcalvo26@gmail.com)

2025-10-28 12:52:46

It works now, yes!

Levi Cai (levi@levicai.com)

2025-10-28 12:52:49

Yup, works now, thanks!

Maria Napolitani (napolitanima@gmail.com)

2025-10-28 21:37:14

@Luke Meyers

🙌 Laura Chrobak

Kevin Barnard (kbarnard@mbari.org)

2025-10-29 19:41:11

@Filippo Varini, great work! I spun up a demo interface for the model for folks who want to try it out: https://huggingface.co/spaces/FathomNet/community-fish-detector

huggingface.co

Community Fish Detector - a Hugging Face Space by FathomNet

WildHackers Community Fish Detector (CFD)

Original URL: https://huggingface.co/spaces/FathomNet/community-fish-detector

🙌 Shir Bar

Filippo Varini (fppvrn@gmail.com)

2025-10-30 08:30:10

That’s amazing @Kevin Barnard !! Thank you for that

Laura Chrobak (lchrobak@mbari.org)

2025-10-30 13:37:55

This is one of my FathomNet favorites! Localization looking sharp!

Filippo Varini (fppvrn@gmail.com)

2025-10-30 13:39:52

That’s amazing !!! 😁
Hope it helps

Shir Bar (shirbar@tauex.tau.ac.il)

2025-10-31 16:15:47

Pretty impressive on Red Sea reef fish as well! Cheers :)

👀 Oc

Filippo Varini (fppvrn@gmail.com)

2025-10-31 16:22:59

Amazing!!!

❤️ Shir Bar

Simone Franceschini (smn.franceschini@gmail.com)

2025-10-28 16:10:30

Hey everyone!

Just sharing FYI, Conservation Mind is a new NGO founded by a few friends from different backgrounds, ranging from WWF conservation programs to AI software engineering. The goal is to democratize access to environmental technologies like AI, camera traps, drones, and software for conservation practitioners, NGOs, and local/indigenous communities.

They’re working on projects from camera trap individual species recognition (e.g., jaguars and elephants) to satellite-based deforestation and illegal fisheries monitoring, aiming to make these AI pipelines and tools freely available and easy to use.

Thought some of you here might find it interesting! 🐯🤖

https://conservationmind.org/

Conservation Mind

Conservation Mind | Connecting Conservation and Technology

Explore how Conservation Mind uses AI and remote sensing to accelerate ecosystem protection and community engagement today.

Original URL: https://conservationmind.org/

❤️ Jon Van Oast, Sara Si-Moussi, Sonny Burniston, Cameron Trotter, DEV MEHROTRA, Peggy Bevan, James Liu, Sara Beery, Alan Stenhouse

👀 Jon Van Oast, Navodita Mathur, Alexander Merdian-Tarko

🎉 Filippo Varini

Tom August (tomaug@ceh.ac.uk)

2025-10-29 06:56:23

If you live or work in a lower- or middle- income countries we would be really grateful if you could fill in this questionnaire about the use of biodiversity monitoring technology. This will be used to help inform future funding from the UK government to overseas projects

🔥 Sara Beery

Danilo Ortelli (danilo.ortelli92@gmail.com)

2025-10-29 11:10:50

Hi everyone! 👋

I’ve been quietly part of this group for a little while now, and it has been a great inspiration: your enthusiasm and the interesting projects and events you shared here actually encouraged me to start an MSc in Artificial Intelligence after completing my engineering degree, with the goal of shifting my career toward AI for conservation!

I am writing because from fall/mid-2026 I’ll begin working on my Master’s thesis and I’d want to develop a project related to AI for conservation, if possible combining my thesis work with a remote internship.

If anyone knows of organizations, labs, companies that may be interested in hosting MSc students for a thesis and/or internship I’d be very grateful for any advice!

Some areas I’m particularly interested in exploring include wildlife monitoring using computer vision, bioacoustic modeling, ecosystem change detection using remote sensing, but I'm open to anything in the conservation field 😊

Happy to connect and discuss possible directions or just exchange ideas!

I share my LinkedIn profile if someone wants to connect: www.linkedin.com/in/danilo-ortelli/
Thanks so much for any help 🙏

🔥 Sara Beery, Nicole Lopez, Joe Nash, Alexander Merdian-Tarko, Laura Madrid

🤞 Youssef Bayouli

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-10-30 05:15:06

we regularly host MSc students at WWF - DE and given we have offices almost everywhere I can share your cv around. Send it to me and some details (like start, end dates, topics of interest) and I'll share it. nuno.cesar.sa@wwf.de

❤️ Danilo Ortelli

Danilo Ortelli (danilo.ortelli92@gmail.com)

2025-10-30 15:09:03

Thank you so much, Nuno! 🙏
That’s amazing to hear, I really appreciate your offer and I’ll send you my CV along with a few details about my timeline and topics of interest shortly!

Peter van Lunteren (peter@addaxdatascience.com)

2025-10-30 06:19:07

The last few days I've been breaking my head on how to finetune the always-crop PyTorch SpeciesNet on a custom dataset. Together with my two pals Claude and Codex we managed to figure something out. It hasn't been thoroughly tested, but it seems to work much better than just training a pretrained efficientnetv2m. For everybody interested, you can find my main findings and a simple script to try it out here, so that you don't have to redo the same steps (which will save you some headaches!).

https://github.com/PetervanLunteren/AddaxAI/blob/main/markdown/finetune-speciesnet-simple.md

If you're on the same route and find anything, let me know! @Dan Morris @Peter Bull

🙌 Matthias Zuerl, Shir Bar, Don Cosseboom, Sara Beery, Simone Franceschini, Alexander Merdian-Tarko, Youssef Bayouli, Hugo

🤯 Paul

🎉 Jon Van Oast, Dan Morris, Sara Beery, Sam Lapp

🙌:skin_tone_2: Cara Appel

🙌:skin_tone_5: Prabath Gunawardane

🙌:skin_tone_3: Alan Stenhouse

Joe Nash (joe@jna.sh)

2025-10-30 06:21:11

you can find my main findings and a simple script to try it out here

looks like the link may have dropped out of the message @Peter van Lunteren!

Peter van Lunteren (peter@addaxdatascience.com)

2025-10-30 06:21:41

Yes, sorry! I pressed enter to quickly 😉

👍 Joe Nash

Peter Bull (peter@drivendata.org)

2025-10-30 13:37:10

Cool! I was working on a timm-compatible translation of the keras weights, which should make the finetuning easier. Dan and I have some time to chat about that tomorrow. We'll report back!

Sara Beery (sara.beery@gmail.com)

2025-10-31 09:08:24

Hey everyone, it's last minute but I just saw this opportunity: https://www.un.org/independent-international-scientific-panel-ai/en/open-call

"As requested by the General Assembly, the United Nations Secretary-General has launched an open call for candidates to serve on the Independent International Scientific Panel on Artificial Intelligence"

There is a limit of 2 panelists per country, and it would be AWESOME to see some representation in the climate and biodiversity space.

👍 Oisin Mac Aodha, Andy Huynh, Dasha Simons, Jose Ruiz-Munoz, Hayley Smith, Malte Pedersen, Steffen Knoblauch, Hugo, Yu Wang

😍 Jon Van Oast, Aleena Munshi

Alex Rood (alex.rood@wildlabs.net)

2025-11-03 13:48:37

You're invited to participate in the first International Conservation Technology Conference (ICTC), hosted by WILDLABS and a coalition of leading partner organisations this February in Lima, Peru! Over two days of pre-conference workshops and three days of the core event, we’ll bring together conservation technology adopters, innovators, scalers, and enablers from across various sectors. By connecting the full spectrum of those shaping conservation technology, ICTC will spark collaboration and unlock new pathways to scale innovation and access to conservation tech!

We are currently accepting applications for conference sessions, pre-conference workshops, and exhibitor booths. Registration to attend will open soon.

Conference Sessions : Do you have an idea, insight, or innovation to share? This is your chance to showcase a tool transforming your work, sharing lessons and questions from the field, or sparking dialogue on the challenges and opportunities ahead. Submit your session proposals by November 15.

Pre-Conference Workshops : Two days of pre-conference workshops for ICTC participants on February 16 and 17 provide an opportunity to run hands-on sessions, technical exchanges, and more intimate community engagements in a collaborative, creative setting ahead of the main event. Submit a workshop by November 15!

Call for Exhibitors : We are now accepting applications until December 1 for exhibitor booths in the ICTC Discovery Zone. Share your work and spark new collaborations with hundreds of innovators, practitioners, and decision-makers. Exhibitions will take place during the core conference days, February 18-20, 2026. Booth spaces are limited and will be accepted on a rolling basis. Apply by December 1 to secure yours!

You can find more information about the conference at https://ictc2026.com.

Additionally, if you are interested in collaborating with us, whether through sponsorship, partnership, or joint initiatives that help make ICTC more inclusive and impactful, please don’t hesitate to contact us at conference@wildlabs.net.

-The WILDLABS team

PS: If you want to stay tuned about future information regarding the conference, please sign up for updates through this form.

🎉 Jon Van Oast, Prabath Gunawardane, Dan Morris, Viktor Domazetoski, Jenna Kline, Andy Huynh, Talia Speaker, Cameron Trotter

💚 Jenna Kline, Andy Huynh, Filippo Varini, Daniel Olhagaray

👀 Gerard, Andy Huynh

Sara Beery (sara.beery@gmail.com)

2025-11-03 13:57:56

COME WORK WITH ME!!!

Very exciting open faculty position at MIT, with ecosystem and biodiversity modeling as a focus!

https://cee.mit.edu/people/shared-faculty-search/

The Department of Civil and Environmental Engineering (CEE), together with the Schwarzman College of Computing (SCC) at the Massachusetts Institute of Technology (MIT) in Cambridge MA, seeks candidates at the level of tenure-track Assistant Professor beginning July 1, 2026 or on a mutually agreed date thereafter.

We seek candidates who can make foundational contributions at the intersection of computing and one of two domains, either (i) ecosystem and biodiversity modeling, such as computational tools for conservation planning, species and ecosystem monitoring, and sustainable agriculture, or (ii) sustainable infrastructure and urban systems, such as AI-augmented planning for electrification, renewable integration, and climate-resilient energy and water systems. In both cases we are interested in candidates who are advancing the domain through the use and development of modern AI and machine learning techniques.

The successful candidate will have a shared appointment in both the Department of Civil and Environmental Engineering (CEE) and the Schwarzman College of Computing (SCC) in either the Department of Electrical Engineering and Computer Science (EECS) or the Institute for Data, Systems, and Society (IDSS), depending on best fit.

cee.mit.edu

Faculty Position in the Department of Civil and Environmental Engineering and the Schwarzman College of Computing - cee.mit.edu

The Department of Civil and Environmental Engineering (CEE), together with the Schwarzman College of Computing (SCC) at the Massachusetts Institute of Technology (MIT) in Cambridge MA, seeks candidates at the level of tenure-track Assistant Professor beginning July 1, 2026 or on a mutually agreed date thereafter. [...]Read More...

Est. reading time

3 minutes

Original URL: https://cee.mit.edu/people/shared-faculty-search/

‼️ Justin Kay, Suzanne Stathatos, Tim Zhou, Jason Holmberg (Wild Me), murilo, Jana Huisman, Angela Zhu, Andy Huynh, Mitchell Rogers, Juan Pierella Karlusich, Vincent Lostanlen, Kalindi Fonda, Paul, Irina Tolkova, gyeongsu, Matthias Zuerl, Nicole Lopez, Ted Schmitt, Hayley Smith, Hiroto, Rupa Kurinchi-Vendhan, Burak Ekim, Don Cosseboom

🤩 Jon Van Oast, Jason Holmberg (Wild Me), Viktor Domazetoski, Shir Bar, Andy Huynh, Mitchell Rogers, Vincent Lostanlen, Xiangteng He, Jess Tam, Yseult Hb, Dasha Simons, Cameron Trotter, Hiroto, Burak Ekim, Riccardo de Lutio, Don Cosseboom, Alan Stenhouse

👍 Holger Klinck, Andy Huynh, Vincent Lostanlen, Paul, Jan Dirk Wegner

Ted Schmitt (teds@allenai.org)

2025-11-04 14:30:52

Hi all, I want to make you aware of the launch of our latest program at Ai2 - OlmoEarth. As with all of our offerings, we make it freely available to mission-aligned partners. Our overarching goal is to make foundational AI infrastructure that is accessible (open) and affordable (free) available to this community in a way doesn't just provide the latest models, but also the tooling and architecture to fit into your workflows. Leveraging AI has been too big a lift for organizations in our space. We hope OlmoEarth lowers that lift significantly. Take a look here.

allenai.org

OlmoEarth | Ai2

Ai2, a non-profit research institute founded by Paul Allen, is committed to breakthrough AI to solve the world’s biggest problems.

Original URL: https://allenai.org/olmoearth

linkedin.com

This is my fourth major program working for the Allen organizations, and I couldn&#39;t be more excited about the potential OlmoEarth has to enhance humans&#39; ability to understand and manage the Planet… | Ted Schmitt

This is my fourth major program working for the Allen organizations, and I couldn't be more excited about the potential OlmoEarth has to enhance humans' ability to understand and manage the Planet for the benefit of both people and nature. This is a game-changer. And we can only do it with deep collaboration and partnerships. We aim to contribute to providing the foundational AI infrastructure needed. Please let us know how we can help your organization achieve its mission to create a better planet.

Original URL: https://www.linkedin.com/posts/tedschmitt_introducing-the-olmoearth-platform-state-of-the-art-activity-7391545724762701824-v6wr

😎 Sara Beery, Jason Holmberg (Wild Me), Andy Huynh, Sara Si-Moussi, Peter Bull, Yi Jin Toh, Nuno Sa, Elizabeth Campolongo, Daniel Rice, Ankit Kumar Upadhyay, Don Cosseboom

🎉 Jon Van Oast, Jason Holmberg (Wild Me), Andy Huynh, Sara Si-Moussi, Peter Bull, Yi Jin Toh, Carly Batist, Molly Blank, Matt Hron, Boe Zhou, Subhransu Maji, Alan Stenhouse

🙌 Laura Chrobak, Dante Wasmuht, Andy Huynh, Sara Si-Moussi, Gerard, Peter Bull, Yi Jin Toh, Connor Jordan

💙 Patrick Beukema

👀 Matt Ziegler

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-11-05 06:13:53

Phenomenal, great! thanks for sharing

Jack O'Hara (jack@centaurlabs.com)

2025-11-07 15:08:49

Happy Friday everyone! I've been a passive member for the last few months... putting it out there, would love to connect with anyone who might be working on model validation, data curation, or any active annotation projects. I'm looking to learn more about what kind of specialized datasets that research groups, CV models, or other platforms may use + how they are fine tuned...

Also curious to learn more about Field Research and how experts are involved in data collection -> preparation for models - specifically for conservation. I'm with a Series B Startup called Centaur.ai - spun out from MIT, YComb - we've created a quality management system for annotating/validating - most of our experience is in Healthcare, and I'm curious about overlap within Conservation!

👍 Ankita Shukla, Shir Bar

Josh Seltzer (jyseltz@gmail.com)

2025-11-09 16:54:35

Hi everyone, as part of our efforts to facilitate more networking opportunities amongst our community, the AI For Conservation WILDLABS group will be hosting an AI for Conservation - Speed Networking Event on November 19th and we'd love to see you there 🎉 !

Everyone is invited and you can sign up easily @ https://luma.com/g59crnky

Feel free to reach out if you have any questions / would like to attend but aren't available at this time. We'd love to hear from you!

luma.com

AI For Conservation — Speed Networking Event · Zoom · Luma

Looking to meet people and find new connections among the amazing WILDLABS / AI For Conservation community? Look no further! Following a few minutes to welcome…

Original URL: https://luma.com/g59crnky

🙌 Connor Jordan, Viktor Domazetoski, Catherine, Tiffany Vlaar, Elizabeth Campolongo, Don Cosseboom, Simone Franceschini, Shravan Ambudkar, Mark Goldwater, Alexander Merdian-Tarko, Jenya Bogacheva, Talia Speaker

🙌:skin_tone_5: Prabath Gunawardane

🌱 Kalindi Fonda, Aleksis Pirinen, Mark Goldwater

Sal Keith (sally.a.keith@gmail.com)

2025-11-11 05:05:42

Hi All. Check out the current PhD projects advertised under the NERC-funded ExaGEO: https://www.exageo.org/phd-student-projects/ Deadline for applications is 9th Jan.

👀 Sara Beery

❤️ Tiffany Vlaar

Sal Keith (sally.a.keith@gmail.com)

2025-11-11 05:07:22

Projects include: Changing Ecological Role of Coral Reef Marine Protected Areas; Scalable Deep Learning for Biodiversity Monitoring under Real-World Constraints; Modelling threatened biodiversity at national, continental and planetary scales; Forests in the Exascale Era: High-resolution Modelling of Global Biomass Drivers, Loss and Recovery

Isaac Corley (isaac.a.corley@gmail.com)

2025-11-11 14:19:47

Hi all, Isaac from TorchGeo & Wherobots here -- can't believe I wasn't aware of this awesome slack workspace! Excited to lurk and catch up on AI 4 Conservation news and papers!

👋 Alex Logan, Timm Haucke, Peter Bull, Shir Bar, Burak Ekim, Dan Morris, Simone Franceschini, Iván Higuera-Mendieta

Alex Logan (alex@cecil.earth)

2025-11-11 14:21:13

we have at least 4 of the same slack workspaces in common now Isaac 😆

💯 Isaac Corley

Peter Bull (peter@drivendata.org)

2025-11-11 14:22:28

I've only got two! What am I missing??

Isaac Corley (isaac.a.corley@gmail.com)

2025-11-11 14:26:26

@Alex Logan Why stop at 4? Let's make it 10!

😆 Alex Logan

Isaac Corley (isaac.a.corley@gmail.com)

2025-11-11 14:27:07

@Peter Bull What's up Peter! Good to see you here. I recently joined the Cecil slack as well to expand my lurking

👋 Peter Bull

Alex Logan (alex@cecil.earth)

2025-11-11 14:34:49

I am in CNG, Source Coop, and I run the Cecil Community Slack too with about 450 nature data folks

Peter Bull (peter@drivendata.org)

2025-11-11 14:42:41

Nice! I am in torchgeo, CNG, and this one. Sounds like there are a few other good ones to lurk in.

🙌 Alex Logan, Isaac Corley

Depanshu Sani (depanshus@iiitd.ac.in)

2025-11-12 00:55:20

Hi everyone!
Since animal Re-ID is highly relevant to this community, I'm sharing a short summary of our findings on existing foundation, unsupervised (USL) and active learning (AL) based Re-ID techniques.
This work, "Active Learning for Animal Re-Identification with Ambiguity-Aware Sampling," just got accepted at the AAAI Conference on Artificial Intelligence (AI for Social Impact track).

💡 Key Findings & Our Solution

Limited Gains: We found that existing foundational models, USL and AL methods offer surprisingly limited gains over a simple pre-trained ResNet-50 baseline across 13 animal Re-ID benchmarks.

Ambiguity-Aware Sampling (AAS): Our solution is a novel active learning strategy that leverages the disagreements between two complementary clustering algorithms—DBSCAN (density-based) and FINCH (nearest-neighbor)—to identify the most informative, uncertain and diverse pairs of images for annotation.

📈 Conservation Impact & Results

Superior Performance: We outperform existing foundational, USL and AL techniques on a wide range of wildlife and person Re-ID benchmarks.

Early Stopping: AAS can assess if further annotations will help, actively reducing wasted annotation effort.

Massive Annotation Reduction: We achieved state-of-the-art using only 0.033% of total annotations, compared to the 0.1% used by other AL baselines. When the total available budget is increased from 0.1% to 10%, we achieve near-skyline performance while still only utilizing 0.037% of the annotations.

Metric Improvements (Closed World Setting): We report an average improvement of 10.49% (over foundational), 11.19% (over USL) and 3.99% (over AL) in mAP across 13 wildlife datasets.

Open-World Readiness: We also show significant improvements on AUC ROC (11.09%, 8.2% and 2.06% for foundation, USL and AL, resp.) for identifying unknown individuals in open-world settings.

Feel free to catch up and ask any questions or suggestions if you want.

🙌 Rohan Sa, Viktor Domazetoski, Gerard, Antón Álvarez, Navodita Mathur, Sara Beery, Peter Bull, Raphael, Elizabeth Campolongo, Aakash Gupta, Aleksis Pirinen, Fadel Seydou, Yu Wang, Yuyan Chen

👍 Lukas Picek, Jack Etheredge, Brian Geuther, Karl Ahrendsen, Yu Wang

Sara Beery (sara.beery@gmail.com)

2025-11-12 18:14:26

Allen Family Philanthropies, formerly known as the Paul G. Allen Family Foundation, has issued a Request for Proposals (RFP) for up to $10 million in funding to accelerate Natural Climate Solutions (NCS). These are nature-based actions that protect, restore, and improve management of forests, wetlands, grasslands, oceans, and agricultural lands, mitigating climate impacts by working with natural ecosystems. This new RFP is seeking to support efforts to develop science and technology solutions to accelerate NCS implementation in the United States.

We would be grateful if you shared this link with non-profit organizations advancing NCS science and practice, which describes eligibility and the process for submitting a Letter of Interest, which is due January 15, 2026.

We will host an informational webinar on December 9, 2025 at 9am PST with a live Q&A about this opportunity. You can register to attend the webinar here. In the meantime, we are also seeking to understand the various barriers to NCS implementation and possible solutions for overcoming them and invite you and partners to fill out this 10-minute survey.

allenphilanthropies.org

Accelerating NCS RFP | Allen Family Philanthropies

Accelerating NCS in the United States

Original URL: https://www.allenphilanthropies.org/accelerating-ncs-rfp

sites.google.com

Individual NCS Survey

Original URL: https://sites.google.com/view/individual-ncs-survey/home

👍 Aleksis Pirinen

👀 Noah Wilson-Rich, PhD

🔥 Alan Stenhouse

Carly Batist (carly.h.batist@gmail.com)

2025-11-17 12:47:11

Session & conference proposals due today for the International Conservation Technology Conference!

Carly Batist (https://aiforconservation.slack.com/team/U08JJMAJL0K)

WILDLABS & partners are hosting the 1st International Conservation Technology Conference! • Location: Lima, Peru • Dates: February 16-20, 2026 • Program: Three core days of knowledge exchange, collaboration, and action (Feb 18-20), with workshop days Feb. 16-17 • Submit sessions <a href="https://whova.com/web/9Lxr4HlVRezcmc-5vjQTb9JxCZlni4Bia%40LDwkfAab4%3D/Submit%20a%20Session/|here">https://whova.com/web/9Lxr4HlVRezcmc-5vjQTb9JxCZlni4Bia%40LDwkfAab4%3D/Submit%20a%20Session/|here</a> (due today!) and apply to be an exhibitor <a href="https://whova.com/web/vKXxRYCYAKvvk25vdKeKfIrNzFk2kDi2XzbJbKyZxks%3D/Call%20for%20Exhibitors/|here">https://whova.com/web/vKXxRYCYAKvvk25vdKeKfIrNzFk2kDi2XzbJbKyZxks%3D/Call%20for%20Exhibitors/|here</a> (due Dec. 1) <a href="https://whova.com/web/vKXxRYCYAKvvk25vdKeKfIrNzFk2kDi2XzbJbKyZxks%3D/">https://whova.com/web/vKXxRYCYAKvvk25vdKeKfIrNzFk2kDi2XzbJbKyZxks%3D/</a>

🎉 Jon Van Oast, Timm Haucke, Joe Nash, Andy Viet Huynh, Boe Zhou, Prabath Gunawardane

⏳ Jon Van Oast, Andy Viet Huynh

❤️ Daniel Olhagaray

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2025-11-17 13:01:39

The web site says they are due tomorrow the 18th.

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2025-11-17 13:02:08

trying to avoid a personal, panic rush to submit right now :)

➕ Timm Haucke

😂 Carly Batist

Carly Batist (carly.h.batist@gmail.com)

2025-11-17 13:03:05

whoops, sorry for the panic! I've updated the post 🙂

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2025-11-17 13:03:34

Whew! Thank you!

Dan Morris (agentmorris@gmail.com)

2025-11-18 10:19:45

New dataset on LILA, thanks to @Erin Connolly and colleagues:

https://lila.science/datasets/biome-health-project-maasai-mara

~37k images from Kenya, with ~100 labeled categories (mostly species-level).

Come for the training data, stay for the zebra foal.

🎉 Peter van Lunteren, Justin Kay, András Zábó, Timm Haucke, Shir Bar, Dante Wasmuht, Jason Holmberg (Wild Me), Yi Jin Toh, Jon Van Oast, Andy Huynh, Carly Batist, Matt Hron, Ankit Kumar Upadhyay, Elizabeth Campolongo, Simone Franceschini, Andy Viet Huynh, Peter Bull, Juan Sebastián Cañas

:zebra_face: Erin Connolly, Oisin Mac Aodha, Shir Bar, Viktor Domazetoski, Jason Holmberg (Wild Me), Kalindi Fonda, Yi Jin Toh, Jon Van Oast, Timm Haucke, Andy Huynh, Carly Batist, Ankit Kumar Upadhyay, Meredith Palmer, Alexander Merdian-Tarko, Andy Viet Huynh, Hugo, Malte Pedersen, Peter Bull, Christopher Crawford, Juan Sebastián Cañas, Eric Cunningham, Alan Stenhouse

❤️ Josh Seltzer, Abigail O'Neill, Viktor Domazetoski, Jason Holmberg (Wild Me), Yi Jin Toh, Timm Haucke, Ankit Kumar Upadhyay, Saket Anand, Hayley Smith, Sara Beery, Andy Viet Huynh, Malte Pedersen, Robert Pless, Peter Bull, Dasha Simons, Juan Sebastián Cañas, Peggy Bevan, Jennifer

stefano puliti (stefano.puliti@nibio.no)

2025-11-19 05:46:05

Hi everyone,

I am looking for the best available model for plant species identification from images and while many apps exist for that I am struggling to find a model that I can run locally. I was a bit surprised to see that the model/code underlying iNaturalist is actually not open, any idea why?

Any suggestions on best models for my task?

🙏

🎉 Jon Van Oast

Lukas Picek (lukaspicek@gmail.com)

2025-11-19 06:16:02

Hi Stefano, the best option would be to use the PlantNet API.
Or, this model could work for you as well --> https://zenodo.org/records/10848263

stefano puliti (stefano.puliti@nibio.no)

2025-11-19 07:10:30

Thanks @Lukas Picek ! I have been trying the pl@ntNet API but I need something I can run locally and with a bit more control of the model. The Zenodo example would be good but for Northern Europe :) anything else out there?

Lukas Picek (lukaspicek@gmail.com)

2025-11-19 07:45:27

I'll check with some colleagues, and come back to you if I got something better suited.
Anyway, I bet that the majority of the flora does not differ that much and majority of species you might need is already in that PlantCLEF checkpoint as it includes ~8k species.

The "fallback" option could be to get data about your "flora" from GBIF and train your own model.

Elizabeth Campolongo (e.campolongo479@gmail.com)

2025-11-20 13:36:58

You could use BioCLIP 2 and provide it a list of the taxa for your specific region to narrow the options. We have a section in the pybioclip docs providing guidance on such an implementation. We used the GBIF May 2024 snapshot (+ EOL, and non-plant sources) for training. It's all open-source.

Youssef Bayouli (youssef.projetnemo@gmail.com)

2025-11-20 05:01:11

Hello Community 🤗

We’re excited to share that our Edge Computing Group will be hosting a session on Project SPARROW, a real-world Edge AI initiative bringing artificial intelligence directly into the wild. SPARROW helps conservation teams make sense of camera-trap photos and acoustic recordings right where the data is collected , even in some of the most remote places.

Dr. Zhongqi Miao from Microsoft’s AI for Good Lab will join us to talk about how this project is supporting wildlife monitoring and the people doing the work on the ground.
Join us: https://wildlabs.net/event/ai-conservation-edge-project-sparrow

👍 Dante Wasmuht, Jason Holmberg (Wild Me), Adam Elzinga, Daniel Olhagaray, Elizabeth Campolongo, Yi Jin Toh, Simone Franceschini

🙌 Dante Wasmuht, Jason Holmberg (Wild Me), Yi Jin Toh, Alexander Merdian-Tarko

🎉 Jenna Kline, Jason Holmberg (Wild Me), Catherine Breen, Boe Zhou, Sara Beery, Yi Jin Toh

👍:skin_tone_5: Prabath Gunawardane

Youssef Bayouli (youssef.projetnemo@gmail.com)

2025-11-20 05:03:01

[ MESSAGE TEXT EMPTY ]

Dante Wasmuht (dante@conservationxlabs.org)

2025-11-20 05:08:03

Related to the SAM3 release yesterday, we just released a comparably large camera trap video dataset (SA-FARI - yes, the acronym is forced) that was used to train SAM3 plus an accompanying short paper. There are other large video datasets @Otto Brookes @Valentin Gabeff @Jenna Kline @Hemal Naik etc. But this one particularly focuses on extremely meticulous annotations for multi-animal tracking + a wide range of species (and camera traps of course). It is VERY 'jungle leaning' but hopefully we can grow this to other biomes! Special shoutout to @Otto Brookes @Maximilian Schall @Christopher Beirne and many others that (might be) in this channel!

conservationxlabs.com

SA-FARI: Open Video Dataset

Wildlife video dataset with segmentation, bounding boxes, and annotations. Made using Meta's SAM3. 10,000+ meticulously annotated videos. 99 species.

Original URL: https://www.conservationxlabs.com/sa-fari

🙌 Youssef Bayouli, Robin Zbinden, Jenna Kline, Peter Bull, Carly Batist, Elizabeth Campolongo, Navodita Mathur, Alexander Merdian-Tarko

😎 Matthias Zuerl, Robin Zbinden, Peter Bull, Adam Elzinga, Sara Beery

🎉 Jon Van Oast, Aleksis Pirinen

🙌:skin_tone_3: Alan Stenhouse

Jenna Kline (jennamkline@gmail.com)

2025-11-20 11:01:41

Amazing work! Congratulations to the team! I'm super excited to try out SAM3 on my data! 🙌

Ben Weinstein (benweinstein2010@gmail.com)

2025-11-20 13:38:48

improving. The concept mapping is interesting. "Trees" versus "Tree".

Ben Weinstein (benweinstein2010@gmail.com)

2025-11-20 13:40:50

one more, best was, "individual tree canopy"

Dante Wasmuht (dante@conservationxlabs.org)

2025-11-20 14:02:37

Someone go try a camera trap video 😉 I think it was fed some drone video data too so that might work ok too. Plus underwater videos as mentioned by @Levi Cai

Ben Weinstein (benweinstein2010@gmail.com)

2025-11-20 14:05:21

Its hard to know where to set expectations. Drone imagery for small objects is still hard, which is totally reasonable, normal tiling approaches still apply. It can't find the heron in the corner, but if you tile in, it can. We are going to explore putting this into DeepForest for others to play with. The challenge is that you need to request the weights from meta, so each user will have to do it individually.

Ben Weinstein (benweinstein2010@gmail.com)

2025-11-20 14:06:32

sorry I dropped the crop in, here is the original, you can see why it doesn't get it.

Dante Wasmuht (dante@conservationxlabs.org)

2025-11-20 14:11:38

👍 also true for camera trap domain. Small animals (or I should say small looking animals because e.g. far away from camera trap) is what it struggles most with (some very rudimentary benchmarking is in the paper I attached)

Ben Weinstein (benweinstein2010@gmail.com)

2025-11-20 14:17:52

I don't really think this is a problem on its own, for traditional models we were tiling and passing the chip to the model anyway, so this is part of what is expected. Perhaps because the model is large this will be a heavy GPU problem, I'll report back.

👍 Dante Wasmuht

Kevin Barnard (kbarnard@mbari.org)

2025-11-20 17:53:27

Also related to SAM 3: nearly 130k images and 300k annotations of ocean animals from the FathomNet Database were used to train and evaluate SAM 3, lending to better performance when applied to underwater imagery and taxa. You can download the training and test sets here and read more in our blog post! 🐟 🦑 🪸

❤️ Daniel Yang, Lilli, Peter Bull, James Liu, Sara Beery, Aleksis Pirinen, Ghjulia Sialelli, Lukas Picek, Asa DeHaan, Jason Holmberg (Wild Me), Sonny Burniston

🌊 Yuyan Chen, Lilli, Peter Bull, Jason Holmberg (Wild Me), James Liu, Dan Morris, Sara Beery, Cameron Trotter, Sonny Burniston

🐠 Shir Bar, Elizabeth Campolongo, Alexander Merdian-Tarko, Sonny Burniston

🎉 Carly Batist

👀 Ben Williams

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2025-11-21 03:16:20

very cool stuff, i used your playground to test with a satellite video from vivid-i https://earthi.space/vivid-i/ - was a bit wonky but it did the trick. Would be very cool to use satellite video for wildlife tracking

Filippo Varini (fppvrn@gmail.com)

2025-11-25 05:15:31

That is amazing work! Congratulations to the team at MBARI, ConservationXLabs and Osa Conservation 🔥

Catherine Breen (cbreen@uw.edu)

2025-11-20 09:47:38

Mark your calendar for the first-ever WILDLABS AI+Geospatial speaker panel! We have four amazing speakers lined up discussing the intersection of AI + geospatial data, covering topics included (but not limited to!) AI agents for earth observation, deep learning for wildlife conservation, and multi-modal model geospatial applications. Come on Thursday, December 11 at 12PM EST (5 PM GMT) to hear from Dr. Mahya Hashemi (NASA), Dr. Isla Duporge (Princeton), Valérie Zermatten (EPFL), and Dr. Li Mi (ETH)!! 🐘🐻🌲 @Viktor Domazetoski @Anna Vallery @Josh Seltzer

Register now here! https://wildlabs.net/event/wildlabs-aigeospatial-speaker-event

❤️ Viktor Domazetoski, Devis Tuia, Nina van Tiel, Josh Seltzer, Justin Kay, Jack O'Hara, Youssef Bayouli, Tanner Corless Smith, Oisin Mac Aodha, Don Cosseboom, Ghjulia Sialelli, Ellie Bowler, Kristin Davis, Ross Gardiner, Carly Batist, Daniel Olhagaray, Boe Zhou, Elizabeth Campolongo, Andy Huynh, Wenxin Yang, Simone Franceschini, Navodita Mathur, Sara Beery, Aleksis Pirinen, Depanshu Sani, Hayley Smith, Alexander Merdian-Tarko, Alan Stenhouse

Carly Batist (carly.h.batist@gmail.com)

2025-11-21 13:11:18

Is there an event registration page or calendar invite?

Catherine Breen (cbreen@uw.edu)

2025-11-21 14:42:12

Yes ! now added to the post!

❤️ Carly Batist

Atul Ingle (ingle@uwalumni.com)

2025-11-23 16:40:40

Megadetector question... I was processing this image and megadetector didn't find anything in here. On visual inspection (I'm not a field ecology expert), I can't find any animals in this image either. But.... [More in thread!]

Atul Ingle (ingle@uwalumni.com)

2025-11-23 16:41:22

... but when I looked at a burst image sequence, lo and behold, I could see the ground squirrel!

Atul Ingle (ingle@uwalumni.com)

2025-11-23 16:41:41

Which brings me to my megadetector question: Is there a way to exploit temporal sequences like these to improve detections?

Atul Ingle (ingle@uwalumni.com)

2025-11-23 16:44:05

(PS: I'm using MDV5A. I also tried thresholds as low as 0.0005.)

Atul Ingle (ingle@uwalumni.com)

2025-11-23 16:53:34

This is the single frame in that sequence where I didn't see a squirrel initially, neither does megadetector.

Lukas Picek (lukaspicek@gmail.com)

2025-11-24 03:33:19

Hi @Atul Ingle, there are multiple different implementations. some work better then other. If you are using this one, I would suspect you are doing something wrong as the animals are clearly visible.

If you did everything all right, I bet @Dan Morris will look into that. He loves those "non working" cases 🙂

Dan Morris (agentmorris@gmail.com)

2025-11-24 12:20:21

Lukas is correct, I love difficult and/or catastrophic MD cases. I would expect MD to see the squirrel in the frame where it's most obvious, but, stranger things have happened than MD missing that. @Atul Ingle, can you share the original images, either here or by email (agentmorris@gmail.com)?

Atul Ingle (ingle@uwalumni.com)

2025-11-24 13:36:18

Thank you!

Dan Morris (agentmorris@gmail.com)

2025-11-24 14:57:32

First, Atul clarified by email that MDv5a has no trouble with the squirrel when it's standing up and obvious.

Dan Morris (agentmorris@gmail.com)

2025-11-24 14:58:55

For the more difficult images in this sequence, I propose that Atul is actually asking two questions:

Is there a way to use the temporal information in the sequence?

Regardless of the answer to (1), what can we do with difficult images like this?

➕ Atul Ingle

Dan Morris (agentmorris@gmail.com)

2025-11-24 15:05:03

Re: temporal information...

AFAIK this is in the category of "lots of research papers, no production systems that do this".

Here are a few papers that incorporate some version of frame-to-frame temporal information:

Beery S, Wu G, Rathod V, Votel R, Huang J. Context r-cnn: Long term temporal context for per-camera object detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition 2020 (pp. 13075-13085).

Shashidhara BM, Mehta D, Kale Y, Morris D, Hazen M. Sequence Information Channel Concatenation for Improving Camera Trap Image Burst Classification. arXiv preprint arXiv:2005.00116. 2020 Apr 30.

Muhammad S, Xiang W, Mann S, Han K, Nair S. TemporalSwin-FPN Net: A Novel Pipeline for Metadata-Driven Sequence Classification in Camera Trap Imagery. 2024 International Conference on Digital Image Computing: Techniques and Applications (DICTA) 2024 Nov 27 (pp. 616-623). IEEE.

Wang Y, Wang J, Shang Y. New Methods for Animal Detection in Camera Trap Image Sequences. In2024 IEEE Intelligent Mobile Computing (MobileCloud) 2024 Jul 15 (pp. 32-37). IEEE.

Park B, Kim J, Cho S, Kim H. Balancing Domain Experts for Long-Tailed Camera-Trap Recognition. arXiv preprint arXiv:2202.07215. 2022 Feb 15.

Wei W, Luo G, Ran J, Li J. Zilong: A tool to identify empty images in camera-trap data. Ecological Informatics. 2020 Jan 1;55:101021.

👀 Atul Ingle

Dan Morris (agentmorris@gmail.com)

2025-11-24 15:07:16

Re: what can we do with difficult images...

First I'll skip to the punchline and show you a version of the results on these more difficult images with all the bells and whistles turned on, then I'll tell you what the bells and whistles are.

🌟 Atul Ingle

🙌 Lukas Picek

Dan Morris (agentmorris@gmail.com)

2025-11-24 15:15:57

Here are the things I try when I'm presented with difficult cases (not necessarily because of camouflage, rather for any type of difficult image):

Turning on test-time augmentation

Trying different MD models (among MDv5a, MDv5b, and MDv1000-redwood)

Increasing the inference size (the default size, and the training size, is 1280px, but increasing that inference size as high as 1920px can still be beneficial)

Tiling the image into smaller images

Aggressive use of repeat detection elimination, which lets you operate at lower confidence thresholds (though I never go below around 0.01, and the default MD inference script has an inference-time threshold of 0.005).

(1), (2), and (3) are all built in to run_detector_batch (the default script for running MD, and what Atul used in this case).

(4) is not literally built in to rundetectorbatch, but run_tiled_inference has a syntax that is almost identical to rundetectorbatch.

(5) is a standalone process that I do for every MD job, whether it's difficult or not.

In this case, I used (1) (I enabled image augmentation), (3) (I set the inference size to 1600), and (4) (image tiling). I couldn't have done repeat detection elimination even if I wanted to here, since I only have these five images, not the whole camera record, but it wasn't necessary. I didn't try other MD models since MDv5a is working fine here with augmentation and tiling.

Atul Ingle (ingle@uwalumni.com)

2025-11-24 15:41:29

Thanks Dan! I'll try these strategies out on my other images!

👍 Dan Morris

Catherine Breen (cbreen@uw.edu)

2025-11-24 15:54:13

As mentioned last week, we have an AI+Geospatial Speaker Panel coming up on Thursday, December 11 at 12PM EST (5 PM GMT)! The flier and registration link is now live!!

We are so excited to be joined by Dr. Mahya Hashemi (NASA), Dr. Isla Duporge (Princeton), Valérie Zermatten (EPFL), and Dr. Li Mi (ETH)!! 🐘🐻🌲 @Viktor Domazetoski @Anna Vallery @Josh Seltzer

Register now here! https://wildlabs.net/event/wildlabs-aigeospatial-speaker-event

❤️ Viktor Domazetoski, Anna Vallery, Hiroto, Gracie Ermi, Fonya Irvine, Ghjulia Sialelli, Robin Zbinden, Boe Zhou, Prabath Gunawardane

🙌 Josh Seltzer, Holly Houliston, Ghjulia Sialelli, Malte Pedersen, Elizabeth Campolongo, Mitchell Rogers

🎉 Jonathan Sauder, Robin Zbinden, Hayley Smith

Patrick Beukema (patrickb@allenai.org)

2025-11-24 23:40:54

How many folks here are using embeddings from earth-observation foundation models (e.g., AlphaEarth, Tessera, TerraMind, etc.) in research or production?

We’re considering releasing a similar set of embeddings based on our recently released OlmoEarth model and would love to hear from people who have used existing options — what you like, what’s missing, and any pain points.

Feel free to reply here, DM me, or email patrickb@allenai.org 💙🌏💙

🙌 Aakash Gupta, Timm Haucke, Esther Rolf, Hayley Smith, Andy Viet Huynh

Aakash Gupta (aakash@thinkevolveconsulting.com)

2025-11-25 01:08:45

This is super cool. I just logged into the channel to post my query regarding the WildMe embedding function (check the next message on the slack! 🙂 ). Not currently using any of the frontier EO foundation models. But maybe in the future.
Curious as to the license under which these embeddings will be released?

Ghjulia Sialelli (ghjulia.sialelli@inf.ethz.ch)

2025-11-25 04:29:23

In my experience, a key points is how fast it is to download the embeddings at a large scale 🙂

Timm Haucke (timm@haucke.xyz)

2025-11-25 09:36:28

We've been using AlphaEarth, Tessera (via geotessera), and a couple other models for occupancy modeling (preprint). I'd love to see & benchmark some easy to use OlmoEarth embeddings! Obviously AlphaEarth / Google Satellite Embeddings is difficult to beat on ease of access. I also really like geotessera's approach of providing a thin wrapper around downloading pre-computed tiles. My personal dream would be accessing embeddings via STAC, which would make custom clients superfluous and make the embeddings accessible from any language (provided it has a STAC client)

stacspec.org

STAC: SpatioTemporal Asset Catalogs

The STAC specification is a common language to describe geospatial information, so it can more easily be worked with, indexed, and discovered.

Original URL: https://stacspec.org/en/

👍 Wenxin Yang

Patrick Beukema (patrickb@allenai.org)

2025-11-25 12:21:38

Thanks @Ghjulia Sialelli — great to know. If you don’t mind, would say a few hours be acceptable or are you looking for immediate retrieval/limited only by the internet speed? (e.g. for AEF there is a requester pays bucket that enables parallel downloads)

Ghjulia Sialelli (ghjulia.sialelli@inf.ethz.ch)

2025-11-25 12:26:38

i'm working on really large scales so for me the most useful is to be able to directly download .tif files, e.g. like for the ESA worldcover (although their tiling is really coarse). in that case it's limited by the internet speed

Patrick Beukema (patrickb@allenai.org)

2025-11-25 12:29:12

Gotcha makes sense @Ghjulia Sialelli — you might check out these COGs from source-coop which just got published: https://source.coop/tge-labs/aef alongside Taylor Geospatial. These are not yet available as STAC but that might be forthcoming. Either way, Jed would be keen to hear that feedback.

source.coop

Source Cooperative

A data publishing utility for the open web.

Original URL: https://source.coop/tge-labs/aef

👀 Ghjulia Sialelli, Timm Haucke

Patrick Beukema (patrickb@allenai.org)

2025-11-25 12:30:07

@Timm Haucke makes sense thanks! And thanks for sharing your paper! Did you find that annualized embeddings were sufficient for this kind of occupancy modeling — would higher temporal granularity be useful to you or not so much?

Timm Haucke (timm@haucke.xyz)

2025-11-25 13:23:24

@Patrick Beukema we didn't explicitly evaluate annualized vs. finer temporal granularity since we found that the annualized Tessera and AlphaEarth embeddings outperformed finer-grained embeddings (roughly monthly resolution) obtained from older models (though that would likely change with better models). Since we're modeling occupancy as constant over time for a single site, higher temporal resolution would likely not help this specific use-case, but it would be of course be amazing for studying seasonality and change over time

Patrick Beukema (patrickb@allenai.org)

2025-11-25 18:18:26

great to know @Timm Haucke! thank you for the insights. much appreciated,

🙌 Timm Haucke

Thijs van der Plas (vdplasthijs@gmail.com)

2025-11-28 04:05:16

This is great @Patrick Beukema, I've worked with AlphaEarth so far but planning to do model comparisons soon and would be great to include OlmoEarth 🙂 Pre-computed embeddings help a lot imo. My use case is downloading sets of 10^3 to 10^5 image patches, either 128x128 ish or just a few pixels each. So for us, what's really helpful is being able to crop using a bbox online and just download the patches, using eg a python api. Using google earth engine/geemap api, this takes about 1-2 seconds per patch for alphaearth, to give you an idea. Hope that helps and looking forward to this!! Thanks!

🙏 Patrick Beukema

Aakash Gupta (aakash@thinkevolveconsulting.com)

2025-11-25 01:07:36

Does the channel members know what is the license under which WildMeId models have been released. Their git repo and HF page doesnt give the necessary information

Dante Wasmuht (dante@conservationxlabs.org)

2025-11-25 02:05:43

@Jason Holmberg (Wild Me) @Lasha Otarashvili

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2025-11-25 09:45:43

Hi @Aakash Gupta We haven't assigned a license yet.

Nathan Fox (foxnat@umich.edu)

2025-11-25 11:21:49

I’ve been thinking a lot about how we can use AI to help safeguard sensitive biodiversity data, especially when those data come from open, citizen-sourced platforms. I recently wrote up some ideas on this and would love to connect with anyone exploring similar questions or facing related challenges in their work. Happy to chat, compare approaches, or think about collaborative directions.
Paper: https://www.sciencedirect.com/science/article/pii/S1574954125004522

😎 Jason Holmberg (Wild Me)

🎉 Jon Van Oast

Sara Beery (sara.beery@gmail.com)

2025-11-25 11:50:49

How do we do more with biodiversity data we've already collected?

I gave a TED Talk on scientific discovery in ecological databases at a joint TED Countdown and Bezos Earth Fund event for NY Climate Week this year, and it's now live!

Watch the full talk at http://go.ted.com/sarabeery 🌍

If you're interested in learning more about the science behind it, we've just released a preprint describing the methodology and the tool we've built: https://lnkd.in/eWtSpS9F

This work is a collaboration with a large team of diverse ecologists, led by my students Edward Vendrow, @Julia Chae, and @Rupa Kurinchi-Vendhan. Many of the awesome folks here were involved!! @Oisin Mac Aodha @Laura Pollock @Marta Jarzyna @Ruth Oliver @Lauren Shrack Isaac Eckert, Jazlynn Hall, Reymond Miyajama, Scott Yanco, and wouldn't have been possible without the team at iNat (shout out Scott Laurie et al)

It started, as many AI efforts to address new challenges do, with the development of a benchmark: https://inquire-benchmark.github.io/ (NeurIPS 24)

🙌 Justin Kay, Xiangteng He, Madison Van Horn, Rupa Kurinchi-Vendhan, Esther Rolf, Robin Zbinden, Patrick Beukema, Levi Cai, Peter Bull, Timm Haucke, Jenna Kline, Louis Freeland-Haynes, Gracie Ermi, Cheng Zhang, Viktor Domazetoski, Joe Nash, Yseult Hb, Simone Franceschini, Oisin Mac Aodha, Carly Batist, Dante Wasmuht, Owen Martin, Gustavo Perez, Polina Stucke, Nuno Sa, Jonathan Sauder, Aleksis Pirinen, Nico Lang, Matthias Zuerl, Hayley Smith, Malte Pedersen, Takumi Sato, Aaron Serianni, Navodita Mathur, Andy Huynh, Danilo Ortelli, Juan Sebastián Cañas, Ed Miller, Vincent Lostanlen, Spencer Bialek, Ben Williams

🎉 Jon Van Oast, Rupa Kurinchi-Vendhan, Sara Si-Moussi, Cara Appel, Peter Bull, Timm Haucke, Alexander Merdian-Tarko, Owen Martin, Alex Chan, Holly Houliston, Nico Lang, Aaron Serianni, Andy Huynh, sunny Kumar, Ross Gardiner

💙 Patrick Beukema, Tiziana Gelmi Candusso, Owen Martin, Andy Huynh

🌻 Burak Ekim, Andy Huynh

🌟 Laura Chrobak, Andy Huynh, Emily Fenwick

🙌:skin_tone_3: Alan Stenhouse

Sara Beery (sara.beery@gmail.com)

2025-11-25 12:02:35

If you want to help broadcast:

https://x.com/sarameghanbeery/status/1993357371515498755

https://bsky.app/profile/sarameghanbeery.bsky.social/post/3m6hrmhtdvk2y

https://www.linkedin.com/posts/sara-beery-1ba89166_nyclimateweek-activity-73991251328[…]m=member_desktop&rcm=ACoAAA4DIN4BIW_3ubfcfe3GqvBT-NI--ZBxMzQ

X (formerly Twitter)

#nyclimateweek | Sara Beery

How do we do more with biodiversity data we've already collected? I gave a TED Talk on scientific discovery in ecological databases at a @TEDCountdown and @BezosEarthFund event for #NYClimateWeek this year, and it's now live! @TEDTalks @inaturalist #AIforConservation

Original URL: https://x.com/sarameghanbeery/status/1993357371515498755

bluesky

Sara Beery (https://bsky.app/profile/sarameghanbeery.bsky.social)

Original URL: https://bsky.app/profile/sarameghanbeery.bsky.social/post/3m6hrmhtdvk2y

linkedin.com

How do we do more with biodiversity data we've already collected? I gave a TED Talk on scientific discovery in ecological databases at a joint TED Countdown and Bezos Earth Fund event for #NYClimateWeek this year, and it's now live! Watch the full talk at <https://lnkd.in/e6Mm9GTZ> 🌍 If you're interested in learning more about the science behind it, we've just released a preprint describing the methodology and the tool we've built: <https://lnkd.in/eWtSpS9F> This work is a collaboration with a large team of diverse ecologists, led by my students Edward Vendrow, Julia Chae, and Rupa Kurinchi-Vendhan. It started, as many AI efforts to address new challenges do, with the development of a benchmark: <https://lnkd.in/engwkDKs> (NeurIPS 24)

Original URL: https://www.linkedin.com/posts/sara-beery-1ba89166_nyclimateweek-activity-7399125132885037056-MH3f?utm_source=share&utm_medium=member_desktop&rcm=ACoAAA4DIN4BIW_3ubfcfe3GqvBT-NI--ZBxMzQ

Ed Miller (ed@hypraptive.com)

2025-11-29 13:59:06

@Sara Beery - Great TED talk! I will share it widely. I am also interested to see if INQUIRE can be integrated with data management packages like TRAPPER!

❤️ Sara Beery

David Rolnick (dsrolnick@gmail.com)

2025-11-27 18:10:39

Going to NeurIPS San Diego? Escape the conference for a couple hours with a morning birding walk in the trails of Balboa Park. Meet 7 am on Thurs, Dec. 4 at the SW corner of the park: https://maps.app.goo.gl/KRCTQ4ZteqKj3jj66

Bring binoculars if you have them, but not necessary and we'll share.

🦤 Mitchell Rogers, Tim Zhou, Owen Martin, Yseult Hb, Shir Bar, Robin Zbinden, Jason Holmberg (Wild Me), Sarthak Arora, Justin Kay, Timm Haucke, Maria Jose Guerrero, Felipe Parodi, Maximilian Schall, Aleksis Pirinen

🔥 Sara Beery, Tim Zhou, Owen Martin, Oisin Mac Aodha, Jason Holmberg (Wild Me), Catherine Villeneuve, Justin Kay, Timm Haucke

Sara Beery (sara.beery@gmail.com)

2025-11-27 21:38:16

BIRD WALK CONTiNUES!!!!

💚 David Rolnick, Subhransu Maji, Esther Rolf, Navodita Mathur

🎉 Jon Van Oast

David Rolnick (dsrolnick@gmail.com)

2025-11-28 09:15:40

Help spread the word 🙂

https://www.linkedin.com/feed/update/urn:li:activity:7400166180872757248/
https://bsky.app/profile/drolnick.bsky.social/post/3m6p32d7qa22o
https://x.com/david_rolnick/status/1994409364589817998

David Bennett (dsbennett2023@gmail.com)

2025-12-03 06:39:05

Hello everyone!

The World Biodiversity Forum is seeking submissions for the session "Advancing farmland biodiversity monitoring: From indicators to action" . It could be a good opportunity for us to get AI based monitoring and indicators publicised.

David Rolnick (dsrolnick@gmail.com)

2025-12-03 11:21:56

Climate Change AI (CCAI) is looking for new core team members. CCAI empowers a global community of innovators, practitioners, and decision-makers to accelerate responsible climate action through the use of AI, by addressing critical gaps in expertise, education, coordination, and research-to-deployment infrastructure. Core team members are volunteers who are expected to dedicate 5-10 hours of time per week. Most core team members have a background in either machine learning or areas relevant to climate change (such as energy, transportation, climate science, etc.), and a strong interest in both topics.

This recruitment call includes roles related to communications, community and education. Applications will be assessed on a rolling basis, but candidates are expected to be able to start in January 2026.

Full details about the roles we are recruiting for and how to apply are available at:
https://www.climatechange.ai/contribute/core-team-application

We look forward to receiving your application!

🌍 Oisin Mac Aodha, Robin Zbinden, Sara Beery, Felipe Parodi, Aleksis Pirinen

Tanya Berger-Wolf (berger-wolf.1@osu.edu)

2025-12-03 12:00:01

When vital data disappear, science and society lose more than information; we lose our ability to research, respond, and recover. The Repository Crisis Scorecards (RCS) project asks a simple but urgent question: How resilient are our data repositories?

Through the RCS, we’re helping repositories, researchers, and communities understand where their strengths lie, where they’re vulnerable, and how to better prepare for disruption, whether from loss of funding, functionality, material, or personnel.

Why get involved?
🔒 Help protect your repository from data loss
👾 Assess risks to your organization’s critical information structure
💸 Stipends are available for completing the RCS or joining a focus group.

What can you do?
🤝 Sign up to let us know you’re interested: https://forms.gle/S6NncgvreHPRBiat6
🔄 Repost to help us spread the word.
🔗 Learn more about the project: esipfed.org/rcs

Google Docs

Interest Form: Repository Crisis Scorecards

Thanks for your interest in the Repository Crisis Scorecards (RCS) project, funded by the Sloan Foundation. This form is designed to gather general interest from individuals who want to stay informed and be contacted about opportunities to participate in project activities. Completing this form does not create any commitment! It simply helps us build a list of interested community members so we can reach out with relevant updates and engagement opportunities as they arise throughout the project. Please note that we are collecting and processing personal data collected through this form in accordance with ESIP’s Privacy Policy. If you have any questions about the project or the form, please email <a href="mailto:rcs@esipfed.org">rcs@esipfed.org</a>.

Original URL: https://forms.gle/S6NncgvreHPRBiat6

✅ Jason Holmberg (Wild Me), Sara Beery, Alan Stenhouse

Keenan Johnson (keenan.johnson@gmail.com)

2025-12-03 16:00:58

Hey everyone! Keenan Johnson here from Seattle, WA USA. I've always been a technologist working in different projects. Most recently I have been really active with the Seattle Aquarium's Coastal Climate Resilience program which is starting up to use robotic systems to quickly collect large observations of kelp forests to understand this important species. You can see more about that here if you are interested.

I also started a non-profit building low cost sensors for citizen science greenhouse gas measurements called Ribbit Network and I spend as much of my free time on that project as I can, which isn't directly related to conservation, but a lot of the technologies and tools for remote sensing are directly applicable.

Glad to join this group and help anyone and everyone I can! My linkedin is here if you care to connect that way.

😎 Jason Holmberg (Wild Me), Shir Bar, Hayley Smith, Sara Beery, Alexander Merdian-Tarko, Arthur Caillau

Keenan Johnson (keenan.johnson@gmail.com)

2025-12-04 15:30:45

Question / Reverse Call for Problems
Hey everyone, I am looking for problems I can help with! I'm a career engineer, and not an expert conservationist, so I've found that one of the ways I can support experts like yourselves is by doing things writing examples, templates and documentation etc for the different (usually confusing) software tools that are out there.

I have a little bit of free time at the moment, and so if there is a problem you have, an example of something, a template you need or anything of that sort, I'd be happy to help create it and post it in the open-source world for you and others! Just let me know if you have any ideas!

🎉 Jon Van Oast, Jason Holmberg (Wild Me), Sarah, Yi Jin Toh

❤️ Roshni Biswas, Shayan Azhar, Yi Jin Toh

Sky Sheng (skysheng7@gmail.com)

2025-12-06 15:06:44

🌻 Hi everyone, hope things are going well for you as we are close to the end of 2025! I wish to follow up about the UBC Master of Data Science Capstone project application. The deadline to submit a proposal, to have 3-4 data scientists work on your project for 2 month for free, is due December 15th, 2025. Application form is here: https://ubc.ca1.qualtrics.com/jfe/form/SV_6rqTW70L2PfkAt0

Please let me know if you have any questions!

Sky Sheng (https://aiforconservation.slack.com/team/U0684AG03EE)

📊 Got conservation data that needs analysis? UBC Master of Data Science students can help! Do you have datasets collecting dust, or a data-driven conservation idea that needs extra hands? UBC’s Master of Data Science program is looking for Capstone partners for May–June 2025. Please note that we welcome remote partners too! What we’re offering: Our MDS students spend the final 2 months of their program on real-world Capstone projects. Past partners have gotten reproducible pipelines, custom visualizations, interactive dashboards, and analysis tools—all well-documented for full reproducibility. Check out past projects: <a href="https://ubc-mds.github.io/capstone/past_projects/">https://ubc-mds.github.io/capstone/past_projects/</a> Why partner with us: • Fresh perspectives on your conservation data challenges • Rigorous analysis backed by faculty mentors • Deliverables you can actually use (pipelines, reports, dashboards, scripts) • Free data science support for research and projects that matters Timeline: Deadline for early feedback on proposals: October 15, 2025 Final deadline on proposals without feedback: November 30, 2025 Interested? Submit a proposal: <a href="https://ubc.ca1.qualtrics.com/jfe/form/SV_6rqTW70L2PfkAt0">https://ubc.ca1.qualtrics.com/jfe/form/SV_6rqTW70L2PfkAt0</a> Partner info: <a href="https://ubc-mds.github.io/capstone/partner_info/">https://ubc-mds.github.io/capstone/partner_info/</a> Full timeline: <a href="https://ubc-mds.github.io/capstone/timeline/">https://ubc-mds.github.io/capstone/timeline/</a> Questions? DM me or reach out to Maymay Milad at <a href="mailto:maymay@science.ubc.ca">maymay@science.ubc.ca</a> (This could be a great fit for camera trap data, habitat analysis, climate datasets, or any conservation challenge that needs data science expertise!)

ubc.ca1.qualtrics.com

Online Survey Software | Qualtrics Survey Solutions

Qualtrics sophisticated online survey software solutions make creating online surveys easy. Learn more about Research Suite and get a free account today.

Original URL: https://ubc.ca1.qualtrics.com/jfe/form/SV_6rqTW70L2PfkAt0

Keenan Johnson (keenan.johnson@gmail.com)

2025-12-07 16:13:25

Hey all! @Matt Weldy and I have been having some good conversation on how to spread up audio analysis tools for work in bird identification. This inspired me to try out and document the performance enhancements you might be able to see from switching a keras model to onnx (spoiler it's a lot!) which is a big win for processing time. If you do similar work it might be helpful or interesting to you, so check it out. https://github.com/breadboardfoundry/pnw-cnet-5-onnx

👍 Ankita Shukla, Sarah, Isaac Corley, Justin Kay, Navodita Mathur, Sara Beery, Xiangteng He, Boe Zhou

👏 Matt Weldy, Yi Jin Toh, Sam Lapp

😮 Angela Zhu

Matt Weldy (matthewjweldy@gmail.com)

2025-12-10 13:41:01

Just to follow up on this.

NWFP monitoring in the PNW collects approximately 2 million hours of audio a year, and data processing is always a big compute lift. If the speed gains in the test Keenan ran (using a published dataset) scale to the larger dataset processing, this change might save us weeks of compute time!

Keenan Johnson (keenan.johnson@gmail.com)

2025-12-10 16:37:25

I am very excited about the potential of this.

Isaac Corley (isaac.a.corley@gmail.com)

2025-12-10 20:10:50

Nice! If you’re concerned about runtime I’d recommend converting the ONNX file to TensorRT format with Ahead-of-Time (AOT) compilation quantized to FP16. Then you should be able to max out the inference batch size since it will use significantly less GPU memory

Keenan Johnson (keenan.johnson@gmail.com)

2025-12-10 20:25:52

Oh interesting and thanks for the suggestion @Isaac Corley . Are there any trade offs in doing so we should be aware of?

👍 Isaac Corley

Isaac Corley (isaac.a.corley@gmail.com)

2025-12-10 20:37:08

The main difference between Just-in-Time (JIT) and AOT) compilation is that JIT can be used on different devices while AOT is compiled for a specific device and version which sometimes isn't backwards compatible, e.g. cpu, cuda, mps. However AOT compiling on CUDA 12.X wouldn't work on CUDA 11 drivers. They are forwards compatible though.

Isaac Corley (isaac.a.corley@gmail.com)

2025-12-10 20:37:48

That's main downside but overall the benefits are better, near zero inference warmup and better optimizations since you're compiling for a specific hardware

Keenan Johnson (keenan.johnson@gmail.com)

2025-12-10 16:43:28

As I mentioned before, in my work on kelp conservation with the aquarium, we wanted to move them to a cloud GPU so they could use linux and stop fighting with their local IT department over windows etc. So I priced out the cloud GPU options for a big RAM GPU system for their image analysis and the results were crazy (spreadsheet attached). That just wouldn't work with their budget, and so I set up a GPU system (Nvidia RTX 6000 pro) as part of the company that I run and got the aquarium access at a third of the price of some others. It's been a real win as it's much cheaper than them acquiring the hardware, cheaper than the big cloud companies, runs on sustainable clean power here in the PNW, and we are able to be fully transparent on the hardware stack. and I'm wondering if others would be interested or if it would be a mutual win? Sorry if this is too salesy and really that's not my intention so feel free to ignore if so. I just feel like having access to the compute at reasonable prices should not be what we are all spending our brainpower thinking about.

Anyway if you are interested or think we could help out your org, let me know.

👍 Parker DePond, Sarah, Alexander Merdian-Tarko

👀 Steve Haddock

Keenan Johnson (keenan.johnson@gmail.com)

2025-12-10 16:44:31

We wrote up this little PDF on what we are doing for the aquarium if you want more info.

Steve Haddock (haddock@mbari.org)

2025-12-15 13:31:34

Checking with our IT dept... we have been in exactly the same situation, trying to set up an instance to run AlphaFold.

Keenan Johnson (keenan.johnson@gmail.com)

2025-12-15 13:39:31

Hey @Steve Haddock sorry to hear that and let me know if you want to chat about it!

🙌 Steve Haddock

Sheila Holmes (sheila.holmes@slu.se)

2025-12-11 06:21:42

PhD position in bioacoustics, statistics, and AI for forest biodiversity at the Swedish University of Agricultural Sciences in Umeå: https://www.slu.se/en/about-slu/work-at-slu/jobs-and-vacancies/doktorand5/

slu.se

PhD position in bioacoustics, statistics, and AI for forest biodiversity | slu.se

SLU - Science and Education for Sustainable Life

Original URL: https://www.slu.se/en/about-slu/work-at-slu/jobs-and-vacancies/doktorand5/

👍 Holger Klinck, Robin Zbinden, Fadel Seydou, Alexander Merdian-Tarko, Magali Frauendorf, Ciara Sypherd

Dylan Van Bramer (she/her) (dcv26@cornell.edu)

2025-12-11 12:25:39

Hi everyone, I’ve been in this community for a while now, and so I’m excited to have something to share now!

I’m a PhD student at the University of Toronto in the CS department working with the Toronto Climate Observatory, Earth Genome and Development Seed on a project studying engineering teams working on developing and deploying products with LLMs for environmental conservation and climate change solutions. In particular, we’re curious about the evaluation of user-centric tools and models that include LLMs either as the main driver of the system, through a user interface like a chat system, or as part of an agentic workflow. We are interested in how these evals are complicated by the integration of complex geospatial data and domain-specific environmental problems. We hope to identify challenges, emerging best practices, and opportunities for collaboration around these evaluations for environmental AI systems through interviews with team members.

If you work on developing AI tools for community partners, scientists, or other users and you’re interested in a ~45 minute chat to discuss your team’s practices and help shape some work around evaluation practices and standards in this area, please send me a message or directly book a time to meet with me! I’m happy to answer any questions via DM. Thank you so much. I'm really looking forward to collaborating with you all!

calendar.google.com

Environmental LLM Evals Interview - Dylan V

Original URL: https://calendar.app.google/HGWKRXNgzPsb6j5X8

🎉 Jon Van Oast, Sara Beery, Sarah, Youssef Bayouli, Andy Viet Huynh, Courtney Davis, Aleksis Pirinen, Paridhi Jain

👀 Ben Williams

Oisin Mac Aodha (macaodha@caltech.edu)

2025-12-11 12:37:15

Sounds very interesting. Perhaps not directly related to your task but @Filip Dorm, myself, and collaborators have a paper describing a benchmark for evaluating LLM's ecological knowledge:
https://www.biorxiv.org/content/10.1101/2025.02.10.637097v2.abstract

Dylan Van Bramer (she/her) (dcv26@cornell.edu)

2025-12-11 12:44:24

@Oisin Mac Aodha Thank you so much for sharing! I'm looking forward to reading this later today. Exploring the baseline ecological knowledge of a language model is absolutely an important piece that we hadn't considered yet.

👍 Oisin Mac Aodha

Sam Heinrich (skh83@cornell.edu)

2025-12-11 13:33:01

Hey everyone! The Cornell Lab of Ornithology is hiring a Machine Learning Engineer to work on the Merlin Bird ID app.
This is a high-impact role where you’ll join a small R&D team to help turn AI/ML research into production tools used by millions of people to identify and enjoy birds and support conservation.
Key focus:

Integrating vision and language models into mobile and web apps.

Building datasets from the Lab's citizen science data to train and evaluate models

Designing machine learning tools for public engagement and conservation

If you want to use AI to connect people with nature, check it out! Don't hesitate to reach out with questions - my email is skh83@cornell.edu
👉 Link to Apply: https://jobs.hr.cornell.edu/us/en/job/WDR-00051454

Cornell University

Machine Learning Software Engineer in Ithaca, New York, United States of America | Information Technology at Cornell University

Apply for Machine Learning Software Engineer job with Cornell University in Ithaca, New York, United States of America. Information Technology at Cornell University

Original URL: https://jobs.hr.cornell.edu/us/en/job/WDR-00051454

🙌 Justin Kay, Keenan Johnson, Sarah, Maddie Cusimano, Sara Beery, Robin Zbinden, Subhransu Maji, Xiangteng He, Alexander Merdian-Tarko, Andy Viet Huynh, Courtney Davis, Arthur Caillau, Boe Zhou, Sam Tymorek, Ciara Sypherd

🙌:skin_tone_3: Alan Stenhouse

❤️ Angela Zhu, Laura Chrobak

Parker DePond (depond.pl@gmail.com)

2025-12-11 13:33:28

Hi everyone! I just wanted to introduce myself to this group and say Hello!

I'm developing a near-infrared vein pattern recognition system for individual wildlife identification—essentially creating a "fingerprint" for animals without distinctive markings. The technology captures unique vascular patterns through fur and skin, enabling permanent, non-invasive ID without physical marking or relying on spot patterns. I've built and patented a prototype, and am now looking for graduate and PhD positions to continue exploring the project.

My background includes:

BS in Animal Ecology from Iowa State.

camera trap ML work at Wind Cave National Park (implementing algorithms for automated analysis)

Arctic shorebird research with the Wildlife Conservation Society in Alaska

Happy to share more details about the technology, discuss potential applications, or just chat about conservation tech in general. Feel free to DM me or drop a comment here. Thanks for having me in this community—really looking forward to learning from all of you!

LinkedIn: https://www.linkedin.com/in/parker-depond-56b2b025a/
Email: DePond.Pl@gmail.com

linkedin.com

Parker DePond - Appalachian Mountain Club | LinkedIn

I’m a wildlife biologist and natural resource interpreter passionate about connecting… · Experience: Appalachian Mountain Club · Education: Iowa State University · Location: Ames · 31 connections on LinkedIn. View Parker DePond’s profile on LinkedIn, a professional community of 1 billion members.

Original URL: https://www.linkedin.com/in/parker-depond-56b2b025a/

😎 Jason Holmberg (Wild Me), Ben Seleb, Shir Bar, Sarah, Sara Beery, Gracie Ermi, Andy Viet Huynh, Elizabeth Campolongo, Hayley Smith

👋 Jon Van Oast, Sarah, Navodita Mathur, Alexander Merdian-Tarko, Andy Viet Huynh, Boe Zhou, Anabel Bedlan

👋:skin_tone_5: Prabath Gunawardane

🤩 Gerard, Andy Viet Huynh, Paridhi Jain

Anna Willoughby (arwill19@gmail.com)

2025-12-12 15:01:36

Woah, super cool! I’m assuming you would need special camera and standard camera traps wouldn’t cut it? Wish I could retroactively do this on my study of ringtails

Parker DePond (depond.pl@gmail.com)

2025-12-12 15:09:15

Thanks @Anna Willoughby ! Surprisingly most CCD and CMOS camera sensors can detect NIR light but they have a filter to limit the range to the visible spectrum.

Chaim (c.elchik123@gmail.com)

2026-01-21 10:41:48

@Parker DePond at what max range does your method work? Most medical applications of vein reading rely on close distances.

Parker DePond (depond.pl@gmail.com)

2026-01-21 17:00:18

@Chaim It is a very close method of identification the scanner would have to be flush with with skin. The idea is that it would be used as a non invasive alternative marking for animals that you already have in hand.

Keenan Johnson (keenan.johnson@gmail.com)

2025-12-13 22:23:12

Anyone have any ideas for pieces of software that you wish existed but didn't? My team is looking to give back to the community a bit by spearheading development of a thing and collecting ideas of things that would be most helpful.

Douglas Mbura (douglas.mbura@gmail.com)

2025-12-15 05:51:06

@Keenan Johnson So many ideas from my team which works with bioacoustics and indigenous people. Will you build free and open source or proprietary software?

🎉 Jon Van Oast, Sarah

Keenan Johnson (keenan.johnson@gmail.com)

2025-12-15 06:01:03

Open source! I should have mentioned that above :)

❤️ Jon Van Oast, Elizabeth Campolongo, Sarah

Hugo Magaldi (magaldi.hugo@gmail.com)

2025-12-16 05:10:50

🚀 New Publication –Automated species recognition for tropical forests monitoring!

Hi all! Very happy to share the release of 𝗗𝗲𝗲𝗽𝗙𝗼𝗿𝗲𝘀𝘁𝗩𝗶𝘀𝗶𝗼𝗻, an open-source algorithm for camera trap photos and videos from African tropical forests. Team work at the French Museum of Natural History and the One Forest Vision initiative with my primatologist colleague Sabrina Krief and 16 co-authors.

📚 Research article accessible in open-access in the BES Ecological Solutions and Evidence journal
https://besjournals.onlinelibrary.wiley.com/doi/10.1002/2688-8319.70167

💻 Model weights and Python inference code available on the project GitHub page
https://github.com/MNHN-OFVI/DeepForestVision
DeepForestVision is also acccessible in the very handy AddaxAI interface from @Peter van Lunteren

Thanks also to @Dan Morris, @Sara Beery and the Microsoft team for making MegaDetector available for the detection step 🙏

We have been the using DeepForestVision in Kibale national park, Uganda for a year, with overall convincing performance, and identified some biaises/confusions that I will try to improve in an upcoming V2. Any feedback will be very useful!

I'll also be happy to demo the model if needed, with the hope that it can be useful to research and conservation projects 🙂

🙌 Youssef Bayouli, Peter van Lunteren, Robin Zbinden, Dan Morris, Yi Jin Toh, Hiroto, Sarah, Parker DePond, Owen Martin, Jason Holmberg (Wild Me), Sonny Burniston, Anabel Bedlan, Nanda Nanduri

🎉 Jenna Kline, Yi Jin Toh, Jason Holmberg (Wild Me), Aleksis Pirinen, Sonny Burniston

🙌:skin_tone_2: Cara Appel

❤️ Owen Martin, Jason Holmberg (Wild Me)

Carly Batist (carly.h.batist@gmail.com)

2025-12-16 15:16:01

Early registration is ending soon for the International Conservation Technology Conference!

https://www.linkedin.com/feed/update/urn:li:activity:7406785604660654080

linkedin.com

#ictc2026 #ictc #ictc2026 #wildlabs #conservationtech #conservation #naturetech #tech4wildlife | WILDLABS Community

The countdown is on for #ICTC2026 and tickets are going fast! ⏰🎉 In just two months, the global conservation technology community will come together in Lima, Peru, for the first-ever International Conservation Technology Conference! 🎟️ More than half of all early-bird tickets have already been claimed, so now is the time to secure your place! Early registration closes on January 16 (or earlier if tickets sell out) and prices may increase after that. 🗓️ Registration covers the core conference (Feb. 18-20), but stay tuned for upcoming information about how to attend the pre-conference workshops on Feb. 16-17! 👉 Register now to secure the lowest price: <https://lnkd.in/gU-Yqiiq> To make the conference as accessible as possible, we’re using a tiered pricing structure for high, middle, and low-income countries, as well as different rates based on organization type, early career status, and other factors. We can’t wait to welcome the conservation tech community to Lima in just 2 months!🎉 #ICTC #ICTC2026 #WILDLABS #ConservationTech #conservation #NatureTech #tech4wildlife

Original URL: https://www.linkedin.com/feed/update/urn:li:activity:7406785604660654080

👍 Holger Klinck, Sara Beery, Timm Haucke, Jason Holmberg (Wild Me), Will Armstrong, Robin Zbinden, Aleksis Pirinen, Sam Tymorek, Ed Miller

Carly Batist (carly.h.batist@gmail.com)

2025-12-19 09:46:44

And first look at the program is out now!
https://www.linkedin.com/posts/wildlabs-community_ictc-ictc2026-wildlabs-activity-740[…]m=member_desktop&rcm=ACoAABnhxIMBifnfoZoUyrfCOUUr5qiZMFhbyRo

linkedin.com

#ictc #ictc2026 #wildlabs #conservationtech #conservation #naturetech #tech4wildlife | WILDLABS Community

A first look at the ICTC 2026 program! We are excited to share an early preview of what’s to come at the International Conservation Technology Conference (ICTC) 2026, happening this February in Lima, Peru! Across five days of programming (Feb. 16–20), ICTC 2026 will bring together practitioners, technologists, researchers, Indigenous leaders, funders, and decision-makers to explore how conservation technology is being used today, and what’s needed to scale solutions responsibly and equitably. Check out some of the pre-conference workshops and conference sessions here ➡️ This is the one working <https://lnkd.in/em5qJiC5> Participants will: ✅ Build practical skills in conservation technologies 🌱 Share real-world lessons from the field 📊 Explore how data, AI, and innovation are driving impact ⚖️ Tackle questions of access, governance, and equity 🤝 Connect across disciplines, regions, and sectors The program will include sessions covering technologies such as camera traps, acoustic monitoring, eDNA, drones, and remote sensing, as well as broader themes like integrated data systems, AI-enabled decision-making, and rights-based, community-led approaches. This is just a first look at the program, with many more workshops, sessions, and speakers to be announced soon! If these sessions sound appealing, make sure to register for ICTC soon before early registration tickets run out ➡️ <https://lnkd.in/gU-Yqiiq> We can’t wait to gather the global conservation technology community in Lima this February to learn, connect, and move conservation forward together💚 #ICTC #ICTC2026 #WILDLABS #ConservationTech #conservation #NatureTech #tech4wildlife

Original URL: https://www.linkedin.com/posts/wildlabs-community_ictc-ictc2026-wildlabs-activity-7407527626648752128-GSCe?utm_source=share&utm_medium=member_desktop&rcm=ACoAABnhxIMBifnfoZoUyrfCOUUr5qiZMFhbyRo

🎉 Jon Van Oast, Prabath Gunawardane

Santiago Ruiz Guzman (sanruizguz@gmail.com)

2026-01-13 11:46:03

Pre-conference workshop list for the ICTC! https://whova.com/web/YBLOiSDZJ2t10k0fMnzvppJh%40t0sBJVXlWRFu5Kc4WY%3D/Pre-Conference%20Workshops/

whova.com

International Conservation Technology Conference hosted on Whova

February 18 – 20, 2026, Peru, Lima

Original URL: https://whova.com/web/YBLOiSDZJ2t10k0fMnzvppJh%40t0sBJVXlWRFu5Kc4WY%3D/Pre-Conference%20Workshops/

👏 Sara Beery, Sofía Miñano, Dan Morris, Carly Batist

Ben Weinstein (benweinstein2010@gmail.com)

2025-12-24 13:29:18

I am updating the general airborne bird detector for DeepForest with about 300,000 new annotations for a total of about a million across maybe 15 or 16 projects. Updated from the original one here https://esajournals.onlinelibrary.wiley.com/doi/abs/10.1002/eap.2694, adding in all that was on lila.science from @Rowan Converse and @Dan Morris and work from USGS and FWS (https://eros.usgs.gov/doi-remote-sensing-activities/2023/fws/applying-deep-learning-detect-and-classify-ocean-wildlife). If there are teams that want to submit a 1 or 2 test images to see if it generalizes (@Anna Vallery, @Wade Sedgwick) or others. It will be online in a couple weeks if everything looks okay. A good reminder to the community to be careful of weird edge cases, we found that the original bird detector produced predictions in pure white images under certain conditions. Adding blank images to harden against this was simple and effective.

eros.usgs.gov

Applying Deep Learning to Detect and Classify Ocean Wildlife | Land Imaging Report Site

The U.S. Fish and Wildlife Service (FWS) Division of Migratory Bird Management-- in partnership with the Bureau of Ocean Energy Management (BOEM), the U.S. Geological Survey (USGS), academic institutions, and private contractors-- is using airborne remote sensing technologies to improve aerial migratory bird surveys. The objective of this project is to improve the quality of wildlife population data by minimizing error and quantifying uncertainty.

Original URL: https://eros.usgs.gov/doi-remote-sensing-activities/2023/fws/applying-deep-learning-detect-and-classify-ocean-wildlife

🎉 Timm Haucke, Jose Ruiz-Munoz, Sara Beery, Mitchell Rogers, Dan Morris, gyeongsu, Jon Van Oast, Devi Ayyagari, Anabel Bedlan, Sarah, Alan Stenhouse

❤️ Owen Martin

🎁 Matt Weldy

Ben Weinstein (benweinstein2010@gmail.com)

2025-12-24 13:41:01

predictions didn't quite upload. here's another zero-shot project.

James Farrell (jamespfarrell@gmail.com)

2025-12-28 03:19:39

AI for public goods - Up to $10,000 per proposal.

This fast grants program offers up to $10,000 USD for each accepted proposal to support AI research and development in public goods decision making.

"Supporting AI research that advances public goods funding through improved decision-making, transparent allocation, and impact assessment."

Recerts

Recerts Journal of Mechanism Design for Public Goods

Original URL: https://www.recerts.org/ai4pg2025

Tiffany Vlaar (Tiffany.Vlaar@glasgow.ac.uk)

2025-12-29 12:59:16

Hi everyone, we have a PhD position on Scalable Deep Learning for Biodiversity Monitoring under Real-World Constraints available within ExaGEO at the University of Glasgow. Supervisor Team: Dr Tiffany Vlaar (primary), Prof. Rachel McCrea, Prof. Colin Torney, Dr. Thomas Morrison, and Dr. Paul Eizenhöfer. Team expertise spans deep learning, mathematics/statistics, ecology, and geoscience.
More information: https://www.exageo.org/phd-student-projects/
Applications close: 9 January 2026.

Lancaster University

Rachel McCrea - Lancaster University

Profile page for Rachel McCrea at Lancaster University

Original URL: https://www.lancaster.ac.uk/maths/people/rachel-mccrea

🎉 Joe Nash, Oisin Mac Aodha, Boe Zhou, Robin Zbinden, Ross Gardiner

🌲 Burak Ekim

Autumn Nguyen (ngoc54n@mtholyoke.edu)

2026-01-02 20:37:35

Research Presentation & Connection opportunities for early-career researchers

This virtual panel will be particularly helpful for
1) people who are interested in learning about research going on in those fields that are driven by budding, self-driven individuals, rather than elite universities and big techs.
2) those self-driven individuals who are hoping to practice research communication,
3) people who are hoping to recruit those hidden-gem students/collaborators.

It will be a 1 hour 15 mins session, starting at 5:30pm PST (6:30pm MST/7:30pm CT/8:30pm EST) on Tuesday Jan 13. However, we may be able to change it to another weekday evening, or a weekend daytime, if the majority of participants’ availability agrees — so let us know in the form below.

We currently have 1-2 more presenter slots for Jan 13! Nonetheless, even if you are not ready to present or cannot make it on Jan 13, let us know of your interest to present in the future in the form, and if there is enough interest, we may be able to organize future sessions!

👉 2 action items: 1) Please leave a comment or repost (if you’re able to) to help this reach more people. 2) fill out the interest form if you’re interested in joining (whether presenting or not): https://forms.gle/Mitu2UF1YyXztUL1A

You can see the meeting link after you click Submit.

Contact 🍁 autumn.yngoc@gmail.com for any inquiries.

Google Docs

Research presentation and connection

Theme: Science for the Real World. This is intentionally broad to include any scientific research (computer science, environmental engineering, social sciences, etc.) that aim for real world impact and/or deal with challenges in real world data. This will be particularly helpful for 1) people who are interested in learning about research going on in those fields that are driven by budding, self-driven individuals, rather than elite universities and big tech groups. 2) those self-driven individuals who are hoping to practice research communication, 3) people who are hoping to recruit those hidden-gem students/collaborators. It will be a 1 hour 15 mins session, starting at 5:30pm PST (6:30pm MST/7:30pm CT/8:30pm EST) on Tuesday Jan 13. However, we may be able to change it to another weekday evening, or a weekend daytime, if the majority of participants’ availability agrees — so let us know in the form below. We currently have 1-2 more presenter slots for Jan 13! Nonetheless, even if you are not ready to present or cannot make it on Jan 13, let us know of your interest to present in the future in the form, and if there is enough interest, we may be able to organize future sessions! You can see the meeting link after you click Submit. Contact 🍁 <a href="mailto:autumn.yngoc@gmail.com">autumn.yngoc@gmail.com</a> for any inquiries.

Original URL: https://forms.gle/Mitu2UF1YyXztUL1A

🙌 Justin Kay, Adrian Ibanez

🧡 Wenxin Yang

Autumn Nguyen (ngoc54n@mtholyoke.edu)

2026-01-02 20:38:21

(optional to read) Additional details that’ll also be sent out to registered participants:

Theme: Science for the Real World.

This is intentionally broad to include any scientific research (computer science, environmental engineering, social sciences, etc.) that aim for real world impact and/or deal with challenges in real world data.

2. This will be different from typical research presentations you have seen, where the speaker gives a 45-min monologue while 95% people in the room are lost, and there remains barely enough time for more than 1 question (and when that’s asked, the question is usually also so advanced that it doesn’t help lost folks at all). Here, the focus is on practicing, improving, and learning. And connecting as well (in the optional social time after the presentations)!

3. Each presenter will have 20 minutes for:

5 mins presentation (of one project, focusing on what you did and what you learned, rather than your group’s results)

10 mins for general Q&A and feedback (e.g. “can you explain what federated learning means in the first place again? —> helping participants learn new advanced concepts as well as helping the presenter to practice explaining technical concepts)

5 mins for in-depth Q&A

4. 1 hour 15 min session includes: first two presentations, one 5-min coffee/break, the third presentation, 10 mins buffer time and mingle time

5. Presentation chances will be prioritized for people who are early in their research career and/or whose life circumstances don’t allow them much chances to go to research conferences.

Xiangteng He (xiangteng.he@ubc.ca)

2026-01-13 14:11:12

📣 FGVC13 Workshop @ CVPR 2026 — Call for Papers
FGVC13 has three tracks (Proceedings / Non-archival / Nectar). Topics include fine-grained categorization, SSL, multimodal learning, human-in-the-loop, benchmarks, and real-world applications (biodiversity/medical/etc.).
⏰ All deadlines: 11:59pm PT

Proceedings (8 pages, in CVPR workshop proceedings): Feb 27, 2026

Non-archival (4-page extended abstract): Apr 3, 2026

Nectar (already-published work): Apr 30, 2026

🔗 Submission info: https://sites.google.com/view/fgvc13/home

sites.google.com

FGVC13 Workshop

Announcements This year, FGVC13 has two paper tracks, and a nectar track. More information is under Call for Papers. Proceedings 8-pages: Submission deadline: Feb 27, 2026 Non-archival 4-pages: Submission deadline: April 03, 2026 Nectar track: Present already published work as a poster:

Original URL: https://sites.google.com/view/fgvc13/home

❤️ Lukas Picek, Subhransu Maji, Sara Beery, Riccardo de Lutio, murilo, Malte Pedersen, Gustavo Perez, Robin Zbinden, Sarah, Nico Lang, Victor Boussange

🎯 Lukas Picek, Kevin Barnard, Oisin Mac Aodha, Robin Zbinden, Sarah, Nico Lang, Sepand Dyanatkar

👍 Aleksis Pirinen

Nanda Nanduri (nnanduri3@gmail.com)

2026-01-14 14:29:07

Hi everyone!
I’m a final-year Computer Science student from Botswana, where wildlife is very much part of everyday life. I’m genuinely grateful to be part of this community and to learn from so many people doing impactful work in AI for Conservation.
I’m currently brainstorming ideas for my final year project, and I’d really appreciate any guidance, suggestions, or problem areas where AI(especially computer vision or machine learning) could make a meaningful contribution.
Thank you in advance for sharing your experience and insights — I truly appreciate the knowledge in this community.

❤️ Mitchell Rogers, Sarah, Dan Morris, Irina Tolkova, Alexander Merdian-Tarko, Andy Huynh, Filippo Varini, Cathy Richards

Dan Morris (agentmorris@gmail.com)

2026-01-14 17:37:50

I hope this is the first of many posts advertising workshops or conference sessions at the ICTC... shamelessly self-plagiarizing this post to/from other platforms.

For folks attending the International Conservation Technology Conference in Peru in February, if you are interested in working on your AI coding skills in a room with like-minded conservation technology practitioners, consider joining our "vibe coding party" pre-conference workshop on Monday February 16th, from 2pm-6pm local time. Registration link is here.

"Coding" is a bit of a misnomer: this is as much for non-coders as it is for coders, we just want to help everyone take one step along their journey toward doing stuff - especially building stuff - with AI. We will follow the general structure of previous vibe coding parties: before the workshop, we'll encourage participants to think through things they might want to build (and to share their ideas with the group), and we'll give pointers re: tools we might recommend, but mostly the focus is semi-structured, hopefully-very-social "practice time" during the event.

Those of you who know me know that I never go anywhere, so I will be co-organizing virtually, but @Lauren Harrell and @Timm Haucke will be leading on site. Priority will be given to in-person participants, but if you are interested in participating remotely, we may have room for some virtual participants, so consider filling out the form even if you won't be there in person.

And as excited as I am about having lots of conservation tech folks in the same room for this, don't wait for this event to practice your AI-building-stuff skills: consider organizing your own (virtual) vibe coding parties for your peer groups; take a look at the suggested structure that has worked well so far. And let me know if you're organizing a vibe coding party I can crash!

Case in point: I was at a delightful vibe coding party last week (along with some other folks from this Slack, I think) where some people used AI to write code related to conservation, but some people didn't write code at all: some of the ecologists on the call who were less jazzed about writing code discovered NotebookLM, and found a whole new way they can use AI, which is just as fantastic as learning to use a coding agent. Whatever step people can take along their AI journey is what we hope to achieve at these events, including the event at the ICTC.

Personally I made random stuff about guitar chords. Everyone's journey takes them different places.

🙌 Carly Batist, Nanda Nanduri, Jenna Kline, Andy Huynh, Aleena Munshi, Kishore Panaganti, Aakash Gupta, Sofía Miñano

❤️ Kishore Panaganti

🙌:skin_tone_3: Alan Stenhouse

Jenna Kline (jennamkline@gmail.com)

2026-01-14 17:41:11

`🔔 Registration Open: ICTC 2026 Pre-Conference Workshop — Edge AI for Conservation

If you’re attending ICTC 2026 in Lima, consider joining us before the main conference for a full-day, hands-on workshop on Edge AI for Conservation: Deploying Intelligence Where It Matters Most.

Edge AI—running intelligence directly on devices in the field rather than in the cloud—is enabling faster, more autonomous conservation responses in remote and connectivity-limited environments. This workshop focuses on practical tools, deployment strategies, and real-world lessons from wildlife monitoring systems using camera traps, bioacoustics, and autonomous sensing.

🗓️ Date: Monday, February 16, 2026 (pre-conference)
📍 Location: Lima, Peru

This workshop will also serve as the inaugural in-person gathering of the WILDLABS Edge Computing community, featuring applied cyberinfrastructure from the NSF ICICLE Institute and opportunities for hands-on learning and networking.
👉 Workshop details & registration:
https://jennamk14.github.io/edgeAI-ICTC-2025/index.html

Space is limited, and workshops are expected to fill quickly so early registration is encouraged.

If I wasn't organizing this workshop, I would def check out @Dan Morris’s vibe coding party! 😄
Check out the other pre-conference workshops here: https://whova.com/web/YBLOiSDZJ2t10k0fMnzvppJh%40t0sBJVXlWRFu5Kc4WY%3D/Pre-Conference%20Workshops/

whova.com

International Conservation Technology Conference hosted on Whova

February 18 – 20, 2026, Peru, Lima

Original URL: https://whova.com/web/YBLOiSDZJ2t10k0fMnzvppJh%40t0sBJVXlWRFu5Kc4WY%3D/Pre-Conference%20Workshops/

🙌 Youssef Bayouli, Carly Batist, Dan Morris, Nanda Nanduri, Timm Haucke, Andy Huynh, Elizabeth Campolongo

🎉 Ed Miller, Sofía Miñano

🙌:skin_tone_3: Alan Stenhouse

Robert Long (Robert.Long@zoo.org)

2026-01-16 13:03:23

Hi folks. I’ll be at the International Conservation Technology Conference in Peru in February. Looking to connect with anyone attending! My interests include noninvasive methods for surveying species and populations--camera trapping, acoustic monitoring, individual ID, wildlife population monitoring, and networked sensors. I study wolverines, lynx, and a suite of other northwest carnivores. I’m also initiating a new Conservation Technologies Lab at Woodland Park Zoo with the aim of jumpstarting and developing tools and methods for noninvasive wildlife studies. I’d love to chat with anyone about ideas, funding, and structure of this new endeavor.

😎 Jason Holmberg (Wild Me), Dan Morris, Sarah, Timm Haucke, Sara Beery, Youssef Bayouli, Sam Tymorek, Karthik Ashok, Carly Batist, Kyra Swanson, Joe Nangle

🙌 Andy Huynh, Timm Haucke, Santiago Ruiz Guzman, Carly Batist

🐾 Irina Tolkova

Keenan Johnson (keenan.johnson@gmail.com)

2026-01-16 13:12:47

Hey @Robert Long ! I live just down the street from the Woodland park zoo, so great to know you at starting such an initiative! I won't be at the conference in Peru, but happy to connect if I can ever help you out locally!

Robert Long (Robert.Long@zoo.org)

2026-01-16 14:22:54

Excellent! I’ll reach out directly, as I’d love to hear more about your work.

Kyra Swanson (tswanson@sdzwa.org)

2026-01-20 18:49:29

Hi! Myself and Ian Ingram of the SDZWA Conservation Tech Lab will be present at the conference, happy to chat with you!

😎 Jason Holmberg (Wild Me)

Robert Long (Robert.Long@zoo.org)

2026-02-08 17:24:06

Hi Kyra…I’d love to connect! I’ll reach out directly.

Aleena Munshi (aleenamunshi001@g.ucla.edu)

2026-01-18 01:20:42

Hi everyone! Just found out about the ICTC. I'm a current student at UCLA studying applied math + conservation biology, and I'd love to attend the conference! Is there any financial support for student/grants I could apply for to make transportation + hotel more affordable? Not sure if the conference is aimed for such early-stage conservationists, but I've spent most my time here working in environmental data science and would love to expand my network.

Autumn Nguyen (ngoc54n@mtholyoke.edu)

2026-01-19 20:17:12

same question here!

Jes Lefcourt (jeslefcourt@gmail.com)

2026-01-20 11:28:21

Flagging for @Talia Speaker

Carly Batist (carly.h.batist@gmail.com)

2026-01-20 12:29:51

Hi folks, answering on behalf of the ICTC team -

We did have travel grants for presenters but unfortunately the call for sessions and apply for those grants has already passed. There are many low-cost lodging options around the convention center (see our Logistics page). I know some folks have found affordable airbnbs nearby as well. We also have reduced registration for early-career attendees. I would recommend also looking into options that your university may have to support conference attendance.

Aleena Munshi (aleenamunshi001@g.ucla.edu)

2026-01-20 15:34:28

Is it too late to buy tickets? I checked the website and it said registration has ended.

Carly Batist (carly.h.batist@gmail.com)

2026-01-20 15:47:13

Early registration has ended, but the second round is opening soon! Just dealing with getting the ticketing system updated as the prices will increase for this round. Website is updated to reflect this now too!

❤️ Aleena Munshi

Isaac Corley (isaac.a.corley@gmail.com)

2026-01-21 12:21:33

New survey of Earth Embeddings https://arxiv.org/abs/2601.13134

arXiv.org

Earth Embeddings as Products: Taxonomy, Ecosystem, and Standardized Access

Geospatial Foundation Models (GFMs) provide powerful representations, but high compute costs hinder their widespread use. Pre-computed embedding data products offer a practical "frozen" alternative, yet they currently exist in a fragmented ecosystem of incompatible formats and resolutions. This lack of standardization creates an engineering bottleneck that prevents meaningful model comparison and reproducibility. We formalize this landscape through a three-layer taxonomy: Data, Tools, and Value. We survey existing products to identify interoperability barriers. To bridge this gap, we extend TorchGeo with a unified API that standardizes the loading and querying of diverse embedding products. By treating embeddings as first-class geospatial datasets, we decouple downstream analysis from model-specific engineering, providing a roadmap for more transparent and accessible Earth observation workflows.

Original URL: https://arxiv.org/abs/2601.13134

💚 Burak Ekim, Andy Viet Huynh, Esther Rolf, Timm Haucke, Robin Zbinden, Andy Huynh, Seven Song, Alan Stenhouse

Ben Weinstein (benweinstein2010@gmail.com)

2026-01-21 16:57:52

Is there anyone in the community working on drone-based object detection and classification of things that are outside of biodiversity monitoring? I'm trying to gauge how much overlap there is from the tools we are writing in deepforest and the challenges faced by other members of the machine learning for airborne observation community. Perhaps anyone here in civil engineering, geography, meteorology, humanities, archeology, etc that are using airborne tools and want more automation, give me a DM.

Sara Beery (sara.beery@gmail.com)

2026-01-21 17:31:58

@Hannah Kerner @Caleb Robinson ?

Caleb Robinson (calebrob6@gmail.com)

2026-01-21 17:45:20

I've heard of using high-res overhead pointing down drone imagery for fossil identification, finding geoglyphs, refugee camp mapping (solar panels, electricity poles, toilets, structures)

Caleb Robinson (calebrob6@gmail.com)

2026-01-21 17:47:58

over the past few weeks I've been playing with SAM3 + non overhead drone imagery to zero shot count sea lions (and seals) and its crazy good (although you need to do something complicated like SAM3(image, prompt="sea lion") - SAM3(image, prompt="bird"))

Devis Tuia (devis.tuia@epfl.ch)

2026-01-22 07:44:12

interesting prompt chain 😉

👍 Sara Beery

Devi Ayyagari (akdevi.92@gmail.com)

2026-01-22 08:05:53

I have come across some work using drones for early detection of disease in agriculture.(perhaps this is very similar to biodiversity monitoring?)
https://www.nature.com/articles/s41598-025-32384-1

Nature

AI-driven drone technology and computer vision for early detection of crop disease in large agricultural areas

Scientific Reports - AI-driven drone technology and computer vision for early detection of crop disease in large agricultural areas

Original URL: https://www.nature.com/articles/s41598-025-32384-1

Rohan Sa (rohan.sawahn@gmail.com)

2026-01-22 09:03:18

Did some work with SearchWing previously who build drones for search and rescue missions. They are working quite a lot on object detection & classification on edge devices onboard of drones. If of interest, we published some work on fast on-board object detection in high-altitude drone imagery.

Caleb Robinson (calebrob6@gmail.com)

2026-01-22 10:30:06

interesting prompt chain

It sometimes thinks birds are sea lions, but seems very confident sea lions aren't birds 🙂. Comparing to human counts its MAE is 1.2 sea lions (so far). It is crazy to me that you can (essentially) "okay remove this class of false positives" and it just happens...

🤩 Devis Tuia, Sara Beery

Elizabeth Bondi-Kelly (ecbk@umich.edu)

2026-01-22 11:07:36

I think this is a relevant grant call in MI, will ping if we submit! https://www.michiganbusiness.org/ofme/aam-statewide-mobility-challenges/

Michigan Economic Development Corporation (MEDC)

AAM Statewide Mobility Challenges | Michigan Business

Michigan plays a leading role in building a strong domestic drone sector by leveraging our statewide infrastructure to build, test, commercialize, and scale AAM technologies.

Original URL: https://www.michiganbusiness.org/ofme/aam-statewide-mobility-challenges/

Ben Weinstein (benweinstein2010@gmail.com)

2026-01-28 12:17:33

Sorry, this is a great thread, I was out of the office for a couple days. Back now. We are prepping an infrastructure grant for DeepForest and along with things discussed with this group already (@Caleb Robinson, tighter integration with TorchGeo, @Sara Beerya megadetector-like airborne generic biodiversity model (thanks @Dan Morris), we are building out tools and agentic connections to other drone processing libraries and just trying to knit the community together. I sometimes feel like this is all still too in house for our community. As @Caleb Robinson said, SAM3 starting to make alot of headway on zero-shot tasks. In our MillionTrees benchmark (still to be released), last week was the first time I saw SAM3 prompt outperform a custom finetuned model. We will see if that holds. If you are an archeologist, or a geographer or a urban planner, it feels like you would need all the same tools for applying ML models to drone data. Support for reading in files, scaling up prediction, multiple annotation geometries, annotation platforms, active learning, visualization of results. Where I potentially see differences, but looking for feedback, is the type of tasks and the challenges in those tasks. Biodiversity monitoring is pretty heavy on detection and classification, with somewhat less on semantic segmentation. Remote sensing more on segmentation, a bit less on detection. I can imagine archeology, civil engineering being alot more like remote sensing, and search and rescue being along more like biodiversity monitoring, with the additional of more video tracking (@Benjamin Hoffman). The model challenges are different too. We still see a real drop in generalization in zero-shot drone imagery cross resolution. Even small changes in resolution can still throw a model badly for the similar scenes with similar target objects. I'm still mulling all of this over, but if there are others here interested in exploring how we reach out of this community to others, i've been writing a small broader impact section to NSF to try articulate why that would reduce duplication, increase efficiency of development and broaden access to ML tools.

👍 Sara Beery, Sofía Miñano

👍:skin_tone_3: Alan Stenhouse

Dan Morris (agentmorris@gmail.com)

2026-01-28 16:17:35

Only a medium-sized distraction from Ben's main purpose on this thread, but not totally a distraction, because I'm following Caleb's lead re: the increasing set of cases where zero-shot localization tasks are possible with VLMs, either to replace object detectors, or - if the only models that can do zero-shot work are still to expensive - to replace the need for human annotators to label images. I.e., in some cases SAM3 might replace models we would have trained previously, but in some cases non-free frontier models - or very computationally expensive open-weights models - might generate the labels, then AI can write the code to generate a bespoke model from those labels.

E.g. when Caleb sent me the seals example from this post earlier on this thread, I also fed one of those images to Nano Banana and asked it to generate a new image with dots on the seals, and it was also very good at this task (see image). This case is less interesting for Nano Banana, because as per Caleb's slides, SAM3 is also good at this, and SAM3 is free. But I did something similar for a set of drone images of elephants where off-the-shelf models for drone images did not do well, and SAM3 was a catastrophe, but Nano Banana crushed it (via a prompt like "please generate a new image with red boxes on the elephants"). NB is too expensive to run on every frame of a collection of drone videos, but not too expensive to use to generate labels.

tl;dr: we have a very big menu of tools to choose from before we train bespoke models, so a lot of the tasks for which it was just not economically feasible to train models two years ago are now at least within striking distance, both because foundation models can often solve at least some parts of the problem, and because agentic coding tools reduce the need for significant engineering support.

👀 Sofía Miñano

👍:skin_tone_3: Alan Stenhouse

Caleb Robinson (calebrob6@gmail.com)

2026-02-02 10:58:40

@Ben Weinstein @everyone, a couple of ideas coming from this little sea lion project:

Is there a way to distill a few concepts from SAM3 into a YOLO model that can be run in browser without loosing to much in terms of generalization? SAM3 definitely can't be run in browser, and I'm not sure what would happen if I scraped a ton of pictures of sea lions, used SAM3 to label them, trained a YOLO model on those labels, then compared YOLO vs. SAM3 on new images. Is there something from model distillation literature that would let me improve YOLO performance (maybe not only matching output logits but matching intermediate feature activations somehow?)

The question above stems from the more general problem: OK -- SAM3 (or other crazy huge SOTA model) seems to work pretty well and we've done a few other things to get a workflow set up with partner X, but running the model takes a GPU and the ability to setup a conda env + run a python script, how do we take Microsoft AI for Good out of the loop so they can do their thing sustainably long term?

I think I might just be looking for huggingface spaces (haven't tried to get it running there and not sure how expensive it would be for them to fund their own, "run SAM3 for like 30 minutes every other week" environment)

Ben Weinstein (benweinstein2010@gmail.com)

2026-02-02 11:42:31

This is a really good question, on my to do board is to https://docs.autodistill.com/ SAM3 to see what happens. This is part of the broader conversation for the entire community about how much we want to utilize these large foundation models. 4 things on my mind, 1) In general, I still see bespoke targeted models outperforming foundation models for biodiversity tasks. 2) The foundation models require the step of hugginface authentication, and I don't know how much to care about that. I think others in the community care more about open-source purity than i do, 3) related, what if any legal restrictions, for example commercial applications, are impacted by distilling data from SAM3 and similiar foundation models, as a whole I don't think we have a clear answer for how models and data inherit legal license structures from their component pretraining workflows. and then 4) after all that, SAM3 is still really heavy.

docs.autodistill.com

Home - Autodistill

Distill large foundational models into smaller, domain-specific models for deployment

Original URL: https://docs.autodistill.com/

👍 Sara Beery, Devi Ayyagari

Carly Batist (carly.h.batist@gmail.com)

2026-01-26 12:56:16

📸 Join the 2026 WILDLABS #Tech4Wildlife Challenge next week (Feb 2-6)!

For the 10th year in a row, WILDLABS invites the community to share photos or videos of how they’re engaging with technology for wildlife conservation. Participate to connect with the community, vote for your favorites, and have your work featured by Ladera Sur and at the 2026 International Conservation Technology Conference (ICTC)!

https://wildlabs.net/article/join-2026-tech4wildlife-photo-challenge-feb-2-6

❤️ Parker DePond, Jon Van Oast, Ted Schmitt, Sarah, Prabath Gunawardane, Dan Stowell, Malika Nisal Ratnayake, Nathan Fox, Sofía Miñano

Grace Lindsay (gracewlindsay@gmail.com)

2026-01-26 13:38:38

Hi all, if you apply for or receive NSF funding, please consider giving feedback on their plans here: https://www.nsf.gov/od/updates/nsf-seeks-public-input-on-its-fy-2026-2030-nsf-strategic-plan (the form closes at the end of the day on the 27th)

NSF - U.S. National Science Foundation

NSF seeks public input on its Fiscal Year (FY) 2026–2030 NSF Strategic Plan

Original URL: https://www.nsf.gov/od/updates/nsf-seeks-public-input-on-its-fy-2026-2030-nsf-strategic-plan

Raj (therajsmile@gmail.com)

2026-01-27 03:49:19

Hello all, I do have a question regarding the CC-BY-SA license of datasets. Can I use it for my own research and publish a small conference proceedings paper (of course all the due credits will be given) in work-in-progress category. It is kind of performance evolution of my chosen tool on various datasets under this license, any inputs are greatly appreciated. Thanks in advance.

Lukáš Adam (lukas.adam.cr@gmail.com)

2026-01-27 04:05:36

Hi, sure. You can do evaluation of your method on datasets with almost any license. CC-BY-SA is one of the most permissible licenses, which allows you even to re-upload the dataset (under giving credit and certain under conditions).

➕ Lukas Picek

Raj (therajsmile@gmail.com)

2026-01-27 04:24:27

Thanks a lot @Lukáš Adam for your valuable input. Now, I understand that, it is okay to publish with this license with all due credits given. Thanks a lot once again.

☺️ Vincent Lostanlen

Greg Newman (gregorynewman7@gmail.com)

2026-01-29 05:24:14

Hey all, I hope it is ok that I post this here. Stellenbosch University has a postdoctoral position open for Computer Vision in Wildlife Monitoring. The position is fully funded, its a great team, and the research they are doing will have immediate impact on conservation teams all over Africa. I’ve attached the description, but if you have any questions also feel free to reach out to me.

🙌 Robin Zbinden, Fadel Seydou, Jason Holmberg (Wild Me), Sarah, Sara Beery, Andy Huynh, Ciara Sypherd

🙌:skin_tone_3: Alan Stenhouse

Fadel Seydou (fadel.seydou@gmail.com)

2026-01-29 08:14:38

Hi Greg,
I've been for working on aerial wildlife counting in South Africa for the past 3 years and I'd love to connect with the lab. The flyer mentions "camera traps", are they working on aerial census?

🧡 Sara Beery

Greg Newman (gregorynewman7@gmail.com)

2026-01-29 08:33:50

Hey Fadel, they would definitely be open to it

🙌 Fadel Seydou

Fadel Seydou (fadel.seydou@gmail.com)

2026-01-31 06:23:40

Hi all! I’m looking to connect with conservation NGOs that have AI needs but lack the hands to execute them.
We have a team of AI volunteers (through a partnership with the African Institute of Mathematical Sciences in Cameroon) ready to support conservation missions in Africa and beyond. If you have a project that needs technical support, I’d love to chat and see how we can provide the talent to get it done. 🚀
Feel free to DM me or send an email at fadel.seydou@delcaux.com

❤️ Sarah, Sara Beery, Jon Van Oast, Ciara Sypherd, Alan Stenhouse

🙌 Malte Pedersen, Alexander Merdian-Tarko, Devi Ayyagari, Sara Beery, Erin W

Jorge Lizarazo (jorge.lizarazo.b@gmail.com)

2026-02-02 14:18:15

Hi everyone! I’m a PhD student at Western University working on a conservation tech project using AI and video analysis to monitor Bank Swallows (Riparia riparia), a threatened bird in Canada. We’ve already developed a deep learning model that detects nesting behavior from temperature sensors, and we’re now building a video-based system to automatically detect active burrows from short recordings. We’re looking for an industry or NGO partner to join the project through Mitacs Accelerate (a matched-funding program). If your org works in conservation, ecology, or sustainability, this could be a great fit. Please feel free to DM me if you're interested!

👍 Josh Seltzer, Alexander Merdian-Tarko

Tyler Marino (tylermarino8@gmail.com)

2026-02-02 15:31:50

Hello all!

I am beginning to work on a project that utilizes 2 orthogonal X-ray videos to create a 3D model of various bird species in order to study mechanics of birdsongs. The plan is to to use DeepLabCut for reconstruction which leverages key markers to help understand 3 dimensional space better. This requires marker tracking across frames to coordinate location which right now takes a lot of human annotation hours.

I am hoping to get some advice on semi or self supervised approaches to marker tracking in rhe general sense and especially if you have done so in the X-ray setting?

Also if you have experience or can guide me in the direction of some work on 3D reconstruction using orthogonal X-rays!

A comment, DM or an email would be great! tylermarino8@gmail.com

Thanks for your help! I hope I can create some nice visualizations for you all to see!

Tyler Marino

❤️ Sara Vanaki

Sara Vanaki (sv67@rice.edu)

2026-02-02 22:31:25

This paper may be helpful, I used it when I was learning about how DeepLabCut works!

Tyler Marino (tylermarino8@gmail.com)

2026-02-03 19:22:44

Thank you @Sara Vanaki

Sara Vanaki (sv67@rice.edu)

2026-02-02 22:31:46

[ MESSAGE TEXT EMPTY ]

😎 Jason Holmberg (Wild Me)

👏 Tyler Marino

Sara Vanaki (sv67@rice.edu)

2026-02-02 22:32:08

@Tyler Marino ^^

Ben Weinstein (benweinstein2010@gmail.com)

2026-02-03 17:04:39

Every year my lab hosts google summer of code students and we've developed a decent track record of successful projects. This year my theme is "recovering historical image annotations", like for example a research team in 2010 that annotated all their (thousands) of airborne flights in paint and dotted over the birds, corrupting the pixels and took a screenshot. Can we recover the locations and classes (different colors) and make it machine learning model ready. If there are other projects that people know about with medium level difficulty forensic work to unlock historical imagery, let me know. Extra points for airborne, but not required. These are pretty high level students, but they only have the summer.

❤️ Jennifer, Sara Beery, Jose Ruiz-Munoz, Jason Holmberg (Wild Me), Shir Bar, Louis Freeland-Haynes, Owen Martin, Holly Houliston, Timm Haucke, Oc, Sarah

👍 Sam Lapp, Alexander Merdian-Tarko

Hugo (hma@reallyarobot.com)

2026-02-05 08:33:46

Hey hey, just wanted to share some of mine and my supervisors/collaborators most recent research in zero-shot clustering to species level clusters of trail-camera images, using ViT models if anyone finds it interesting 😊

I am missing to release the code (will be added soon), but there is no magical chunk of code needed to get started, just follow the methods explained in the paper.

Preprint: https://arxiv.org/abs/2602.03894
Interactive UI: https://hugomarkoff.github.io/animal_visual_transformer/

arXiv.org

Vision Transformers for Zero-Shot Clustering of Animal Images: A Comparative Benchmarking Study

Manual labeling of animal images remains a significant bottleneck in ecological research, limiting the scale and efficiency of biodiversity monitoring efforts. This study investigates whether state-of-the-art Vision Transformer (ViT) foundation models can reduce thousands of unlabeled animal images directly to species-level clusters. We present a comprehensive benchmarking framework evaluating five ViT models combined with five dimensionality reduction techniques and four clustering algorithms, two supervised and two unsupervised, across 60 species (30 mammals and 30 birds), with each test using a random subset of 200 validated images per species. We investigate when clustering succeeds at species-level, where it fails, and whether clustering within the species-level reveals ecologically meaningful patterns such as sex, age, or phenotypic variation. Our results demonstrate near-perfect species-level clustering (V-measure: 0.958) using DINOv3 embeddings with t-SNE and supervised hierarchical clustering methods. Unsupervised approaches achieve competitive performance (0.943) while requiring no prior species knowledge, rejecting only 1.14% of images as outliers requiring expert review. We further demonstrate robustness to realistic long-tailed distributions of species and show that intentional over-clustering can reliably extract intra-specific variation including age classes, sexual dimorphism, and pelage differences. We introduce an open-source benchmarking toolkit and provide recommendations for ecologists to select appropriate methods for sorting their specific taxonomic groups and data.

Original URL: https://arxiv.org/abs/2602.03894

❤️ Sara Beery, Devi Ayyagari, Kevin Barnard, gyeongsu, Sepand Dyanatkar, Dasha Simons

🙌 Malte Pedersen, Alexander Merdian-Tarko

Filippo Varini (fppvrn@gmail.com)

2026-02-05 09:16:52

congrats @Hugo, super interesting!

Devi Ayyagari (akdevi.92@gmail.com)

2026-02-05 12:09:58

This is great @Hugo. Congratulations!

Ben Weinstein (benweinstein2010@gmail.com)

2026-02-05 17:29:16

Super interesting, would love to see it in lower resolution datasets. Do you have an opinion on its potential for novel object detection? Biodiversity surveys are often dominated by a small number of classes that are frequent that are of low interest to the user, and then a small number of instances that are of high interest to the user. The differences may be very fine-grained, I'd like to use these kind of clustering tools to highlight potential important images for human review. I can share some images if that's of interest.

Steve Haddock (haddock@mbari.org)

2026-02-06 00:08:12

@Danelle Cline 👀

Hugo (hma@reallyarobot.com)

2026-02-06 03:45:35

Thanks @Filippo Varini and @Devi Ayyagari!

@Ben Weinstein thanks, I am trying to understand if I get your question correctly. Are you speaking of a long tailed distribution where some animals are common and some rare? We test this in our paper under section 4.7 Performance on unevenly distributed data - "The extreme scenario samples randomly between 20 and the maximum available images per species, resulting in ranges of 48–4,145 for Aves and 29–6,431 for Mammalia." It has quite a few subsections discussing how to prioritize the small represented clusters and when they are either merged or put at "outliers" – where outliers would need expert attention.

Market (marketzimova@gmail.com)

2026-02-06 08:23:39

Excited to share that we are hiring a computer vision postdoc to help us analyze coat color and camouflage from remote camera trap images collected around the world.
This is a fully funded, three-year NSF-supported position based at the North Carolina Museum of Natural Sciences/NC State University in Raleigh. Applications are due 3/5, but we will begin reviewing them as they are received. There will also be opportunities to get involved in additional projects within the PIs’ wildlife ecology labs. More details in the full ad here.

😎 Sara Beery, gyeongsu, Sarah

❤️ Hugo, Tessa Rhinehart

Jan Huus (jhuus1@gmail.com)

2026-02-06 08:48:46

I am thrilled to announce the official release of BriteKit, a Python toolkit for developing bioacoustic recognizers. There is lots of documentation in the github repo, so check there for details. I’m currently using it to develop the next release of HawkEars, which is about a month from completion.

🙌 Hugo, Dan Morris, Elly Knight, Mike Gallimore, Sarah, Aditya S.K

👍 Holger Klinck

🎉 Jon Van Oast, Tessa Rhinehart, Kevin Kelly, Oc, Elizabeth Campolongo

🙌:skin_tone_3: Alan Stenhouse

Meghan (meghan.balk@gmail.com)

2026-02-06 12:27:24

this may be of interest to some here:
The TDWG 2026 Call for Sessions is now open! Deadline is 25 February 2026, 23:59 UTC-10.

We invite you to submit a proposal for an organized session at TDWG 2026 in Oslo, Norway, 21-25 September. Submissions should reflect the work that TDWG does and the ways that collections and standards contribute to our understanding and documentation of biodiversity.

We strongly encourage you to consider the conference theme: Research and Robot-ready Biodiversity Data Standards.

Organized session types include symposia with either long or lightning talks, discussions, workshops and others.
Every session must include at least two organizers who will be present at the in-person meeting to run the session. Every session must have an abstract that will be published on the TDWG 2026 website. ALL session organizers, presenters, and panelists in the session must be registered for the conference.

Visit the TDWG 2026 conference website for more information and detailed instructions.

tdwg.org

TDWG 2026 - Call for Sessions

Call for organized sessions at TDWG 2026 in Oslo, Norway

Original URL: https://www.tdwg.org/conferences/2026/call-for-sessions/

tdwg.org

TDWG 2026

TDWG Annual Conference - A hybrid conference in Oslo, Norway, 21-25 September

Original URL: https://www.tdwg.org/conferences/2026/

tdwg.org

Instructions for Session Proposals

Instructions for submitting a proposal for an organized session

Original URL: https://www.tdwg.org/conferences/2026/session-instructions/

🎉 Jon Van Oast

Arron Watson (axw213@student.bham.ac.uk)

2026-02-06 12:31:52

Does anyone have any advice on Bayesian online short courses or similar ( already undertaking a PhD), I am keen to explore the use of Bayesian methods in machine learning and AI as my research currently has a lot of different datasets with dimensions, types etc. I have been looking at hybrid models (process and data driven) but it sounds like Bayesian AI methods may be something for me to explore. I have enquired to UKCEH who have a bayesian course for environmental sciences but they don't currently have dates for a course.

Any advice is welcomed

Leonardo Viotti (leonardoviotti@gmail.com)

2026-02-06 12:44:26

Not sure if it falls under "short", but I really like Richard McElreath's Statistical Rethinking. He has a book, but also lectures on Youtube.

https://youtube.com/playlist?list=PLDcUM9US4XdPMtSV81e1R_4B6NugQBvTP&si=J7mm1ZRPiuU8LFC6

YouTube

Statistical Rethinking 2026 Beginner Section A

Beginner section for 2026 course. Full course information here <https://github.com/rmcelreath/stat_rethinking_2026>

Original URL: https://youtube.com/playlist?list=PLDcUM9US4XdPMtSV81e1R_4B6NugQBvTP&si=J7mm1ZRPiuU8LFC6

❤️ Aakash Gupta, Arron Watson

👍 Anthony Miyaguchi, Louis Freeland-Haynes, Evan Eskew

Aakash Gupta (aakash@thinkevolveconsulting.com)

2026-02-06 13:17:34

Love this , started it recently. Dont get much time to complete it. But taking it slowly -- like aged wine !

Aakash Gupta (aakash@thinkevolveconsulting.com)

2026-02-06 13:18:06

He has divided the 2026 course into a beginner and advanced section which are being run in parallel.

👍 Arron Watson

Evan Eskew (eveskew@gmail.com)

2026-02-09 10:46:34

Coming here to echo Statistical Rethinking. The book, multiple years of video resources online, plus folks have translated model code in the book into different languages (e.g., https://github.com/aloctavodia/Statistical-Rethinking-with-Python-and-PyMC3). So all in all a great learning ecosystem.

👍 Arron Watson

Arron Watson (axw213@student.bham.ac.uk)

2026-02-09 11:23:43

Thanks for all these, I didn't expect to get so much feedback so quickly. I was working away last few days so shall digest the messages and come back to you all

Arron Watson (axw213@student.bham.ac.uk)

2026-02-12 06:41:29

i am watching the lectures today for the course, thanks will let you know how i get on 🙂

Arron Watson (axw213@student.bham.ac.uk)

2026-02-19 05:42:14

i have done chapters A and really enjoyed it, would you recommend getting the book as well? i have been looking at the code examples to use on github but not tried any yet.

this is how this whole thing came about, my PhD funder can pay for a course in my second year and i was looking at this - Bayesian methods for ecological and environmental modelling | UK Centre for Ecology & Hydrology

ceh.ac.uk

Bayesian methods for ecological and environmental modelling

This training course will run over 5 days. It will be a mixture of presentations and practical exercises. You will get a solid grounding in Bayesian methods that you can use with any kind of model and data set to compare models, estimate parameters, analyse uncertainties and communicate results. £ 799 students (Early Bird Rate, then £849) £1049 non-students (Early Bird Rate, then £1199) This includes the cost for refreshments, but excludes the cost for evening meals and accommodation. Travel & accommodation: We will provide accommodation, where required. We will reimburse you for your travel costs as outlined in our reimbursement policy. We expect you to book your travel as soon as we have confirmed your place to keep the cost as low as possible. need to replace with update: We will cover: Day 1: (half-day): Bayesian basics & overview of the course Day 2: Hierarchical modelling, Spatio-temporal modelling Day 3: Comprehensive uncertainty quantification, Combining disparate datasets Day 4: Model comparison, Communication of results Day 5: (half-day): Software choices, Graphical modelling, Discussion of the future of Bayesian methodology

Original URL: https://www.ceh.ac.uk/training/bayesian-methods-ecological-and-environmental-modelling

Evan Eskew (eveskew@gmail.com)

2026-02-19 09:09:35

I am not actually sure how closely the video materials now match with the most recent edition of the book. Perhaps he addresses that in the lectures somewhere? He seems to have done substantial updates to the course material over time, which is a great thing. But as a result, the course material may now veer away from the book slightly? Regardless, the book is a fantastic resource on its own. Think it depends on how you'd like to learn. I'm sure some would prefer to sit with the text and code examples in physical book form rather than scan through videos on YouTube to revisit key material.

Arron Watson (axw213@student.bham.ac.uk)

2026-03-09 15:26:57

Thanks for this Evan, he said about updating the code on the website regularly. The book can be picked up fairly cheap so i think i shall invest, i have really enjoyed the sessions so far of module A

Juli Carrillo (juli.carrillo@ubc.ca)

2026-02-09 12:43:14

Funded postdoctoral position in Vancouver, Canada! We want to fill this position asap, so reach out soon if interested

😎 Sara Beery, Michael Bunsen

👀 Alan Stenhouse

Juli Carrillo (juli.carrillo@ubc.ca)

2026-02-09 12:47:15

Links embedded in doc

Josh Seltzer (jyseltz@gmail.com)

2026-02-11 12:30:00

One last reminder not to miss out on the WILDLABS AI for Conservation Matchmaking event - tomorrow @ 12pm ET!

It was genuinely so much fun meeting people at our last networking event, and everyone is invited regardless of your experience so I recommend signing up 🎉https://luma.com/2rx368k1

Viktor Domazetoski (https://aiforconservation.slack.com/team/U042XFTSXQQ)

Looking to meet new collaborators at the intersection of AI and conservation? Join us for the AI for Conservation Matchmaking event, bringing together computer scientists and conservation practitioners from around the world! Scheduled just before ICTC, this event is also the perfect opportunity to start building connections ahead of the conference. Whether you’re analyzing camera trap data, developing machine learning models, or seeking partners to tackle real-world conservation challenges with AI, this is your chance to connect with like-minded people. All backgrounds and experience levels are welcome. We’re hosting this special Valentine’s-themed “matchmaking” event because the best research collaborations are truly a match made in heaven. Complete a short survey in advance, and we’ll pair you with potential collaborators or like-minded people who share your interests and offer complementary expertise. When: Thursday, February 12 at 9AM Pacific Time (5 PM GMT) Register Here: <a href="https://luma.com/2rx368k1">https://luma.com/2rx368k1</a>

luma.com

AI for Conservation Match-Making · Zoom · Luma

Join us for a virtual networking event bringing together computer scientists and conservation practitioners. Whether you're working with camera trap data,…

Original URL: https://luma.com/2rx368k1

❤️ Jennifer, Oisin Mac Aodha, Robin Zbinden, Michael Bunsen

Fadel Seydou (fadel.seydou@gmail.com)

2026-02-14 07:38:18

Hi @Josh Seltzer, I missed the previous matchmaking :( Is there another one coming up soon?

Josh Seltzer (jyseltz@gmail.com)

2026-02-14 18:52:12

Hey @Fadel Seydou we don't have anything scheduled just yet but will be sharing more events here in the coming weeks

👍 Fadel Seydou

Shane Josias (josias@sun.ac.za)

2026-02-12 07:26:49

Hi all, just a quick reminder about the postdoctoral position in Computer Vision for Wildlife Monitoring at Stellenbosch University that Greg shared earlier.

I am one of the researchers leading the project. The role focuses on modern machine learning and computer vision methods applied to large, real-world wildlife image datasets, in collaboration with conservation partners across Africa.

The position is fully funded, and the successful candidate would join us in the beautiful town of Stellenbosch, South Africa. The deadline is 18 February. If you are interested, or know someone who might be a good fit, please feel free to reach out to me directly. I am happy to chat and answer any questions.

Fadel Seydou (fadel.seydou@gmail.com)

2026-02-14 07:42:10

Hi @Shane Josias,
I've been for working on aerial wildlife counting in South Africa for the past 3 years and I'd love to connect with your lab. Here is a recent publication I co-authored. By any chance, are you working on aerial census as well? I'd love to have a chat :)

David Bennett (dsbennett2023@gmail.com)

2026-02-12 08:00:30

Hi everyone,

Would this be of interest to any of your masters or PhD students? While it is not technically AI focused, it is connected to most of our study topics.

https://landchange.imk-ifu.kit.edu/summer-school-2026

Sara Beery (sara.beery@gmail.com)

2026-02-12 09:06:56

The graduate-level deep learning course I teach at MIT is now publicly and freely available through MIT OpenCourseWare! https://ocw.mit.edu/courses/6-7960-deep-learning-fall-2024/

Definitely not an intro class, but if you're looking to dive deeper into deep learning maybe you'll find this helpful ❤️

MIT OpenCourseWare

Deep Learning | Electrical Engineering and Computer Science | MIT OpenCourseWare

This course covers the fundamentals of deep learning, including both theory and applications. Topics include neural net architectures (MLPs, CNNs, RNNs, graph nets, transformers), geometry and invariances in deep learning, backpropagation and automatic differentiation, learning theory and generalization in high dimensions, and applications to computer vision, natural language processing, and robotics.

Original URL: https://ocw.mit.edu/courses/6-7960-deep-learning-fall-2024/

🙌 murilo, Shir Bar, Nico Lang, Michelle Voong, gyeongsu, Justin Kay, Jose Ruiz-Munoz, Gustavo Perez, Subhransu Maji, Juli Carrillo, Jennifer, Don Cosseboom, Rupa Kurinchi-Vendhan, Jesse Leung, Madison Van Horn, Abigail Grassick, Mamoon, Atul Ingle, Aaron Serianni, Sarah, Mitchell Rogers, Xiangteng He, Irina Tolkova, Declan, Boyu Zhang, Tim Zhou, Emilio Luz-Ricca, Aleksis Pirinen, Angela Zhu, Soledad Rivas

💯 murilo, Don Cosseboom, Shayan Azhar, Soledad Rivas

❤️ Joe Nash, Shir Bar, Brett Ford, Alba Márquez-Rodríguez, Yuyan Chen, Viktor Domazetoski, Juli Carrillo, Don Cosseboom, Michael Wethington, Blair Costelloe, Jon Van Oast, Hayley Smith, Yseult Hb, Irina Tolkova, Shayan Azhar, Arron Watson, Morgan Langley, Alan Stenhouse, Justine Boulent, Abigail Oppong, Soledad Rivas, Aamir

Atul Ingle (ingle@uwalumni.com)

2026-02-12 11:58:48

Which workshops are conservation folks attending at cvpr2026? CV4Science? EarthVision? https://cvpr.thecvf.com/Conferences/2026/Workshops

Sara Beery (sara.beery@gmail.com)

2026-02-12 12:05:48

Those, plus FGVC, CV4Animals, Auto-annotation with expert created guidelines (relevant to field guides!), monitoring the world through an imperfect lens, maritime CV

🙏 Atul Ingle, Jason Holmberg (Wild Me), Mitchell Rogers, Joe Nash

Mitchell Rogers (mitchell.rogers@auckland.ac.nz)

2026-02-12 15:41:26

Speaking of CVPR2026 workshops...

We are pleased to announce CV4Animals! 🎉

The 6th Workshop on Computer Vision for Animal Behavior Tracking and Modeling (CV4Animals) has been accepted to CVPR 2026 in Denver, Colorado as a half-day workshop (3rd or 4th of June TBD)!
CALL FOR PAPERS: https://www.cv4animals.com/

Tracks for unpublished and published work

Deadline: Midnight 17th of March AOE

If you have presented some work at CV4Animals before, you may recognize some of the animals on the poster! :giraffe_face:🐦🐘🐧🐳🐻‍❄️

🌟 Atul Ingle, Aleena Munshi, gyeongsu, Yseult Hb, Robin Zbinden, Tuan-Anh VU, Timm Haucke, Hugo, Thijs van der Plas, Sarah

🎉 Jon Van Oast, Alex Chan, Tuan-Anh VU, Jenna Kline, Timm Haucke, Simone Franceschini, Jaycee Choi, Elizabeth Campolongo, Dasha Simons, Sarah, Aleksis Pirinen

🙌 Matthias Zuerl, Nico Lang, Malte Pedersen, Timm Haucke, Alexander Merdian-Tarko, Thijs van der Plas

😎 Sara Beery, Timm Haucke

Atul Ingle (ingle@uwalumni.com)

2026-03-13 20:15:24

I'm not seeing a place to upload supplementary zip files on openreview.

Mitchell Rogers (mitchell.rogers@auckland.ac.nz)

2026-03-15 21:58:31

Hi Atul, we are just going to add this now. Also, we will be extending the deadline by a week if you need more time.

Atul Ingle (ingle@uwalumni.com)

2026-03-24 14:57:47

I don't see a place to upload supplement

Atul Ingle (ingle@uwalumni.com)

2026-03-24 16:19:10

Never mind. I see it now!

🙌 Mitchell Rogers

Mitchell Rogers (mitchell.rogers@auckland.ac.nz)

2026-03-24 16:29:33

Yeah, you should be able to edit your submission to add supplementary material after the initial submission

Atul Ingle (ingle@uwalumni.com)

2026-03-24 16:39:16

Thank you!

Aleena Munshi (aleenamunshi001@g.ucla.edu)

2026-02-14 03:51:14

Hi everyone! Currently looking for any opportunities for aspiring conservation tech scientists, aimed towards undergrads!! Does anyone know of a lab/program looking for students this spring/summer?!! Can be really anything from bioacoustics to SDMs to animal behavior!

❤️ Joe Nash, Sarah, Jon Van Oast

Ben Weinstein (benweinstein2010@gmail.com)

2026-02-14 13:11:35

On the front page of NPR, our very own, @Sara Beery.

🙌 Rupa Kurinchi-Vendhan, Shir Bar, Dan Morris, Yuyan Chen, Jose Ruiz-Munoz, Joe Nash, Juli Carrillo, murilo, Eric Snyder, Ben Weinstein, Yseult Hb, Jason Holmberg (Wild Me), Vincent Lostanlen, Madison Van Horn, Tyler Marino, Mitchell Rogers, Nico Lang, Lin Lin, Ethan White, Oisin Mac Aodha, Benjamin Kellenberger, Aaron Serianni, Robin Zbinden, Justin Kay, Jenna Kline, Jennifer, Dylan Van Bramer (she/her), Rachelle Ketelhohn, Esther Rolf, Andy Huynh, Alexander Merdian-Tarko, Matthias Zuerl, Sara Vanaki, Martin Holub, Timm Haucke, Sarah, Levi Cai, Ben Lopata, Devi Ayyagari, Julia Chae, Gracie Ermi, Abigail O'Neill, Yumi Arimitsu, Neehar Kondapaneni, Lauren Harrell, Declan, Shayan Azhar, Kyra Swanson, Michael Bunsen, Elizabeth Campolongo

🫣 Sara Beery

😎 Jason Holmberg (Wild Me)

🤩 Jon Van Oast, Shayan Azhar, Michael Bunsen, Alice Heiman

🥺 Shayan Azhar

Sara Beery (sara.beery@gmail.com)

2026-02-18 07:12:29

If you're at ICTC today, don't miss @Timm Haucke talking about our project towards improving occupancy modeling with images on the ground, and how we can use this framework to not only improve predictive accuracy but also improve transparency and explanability - building an interpretable understanding of which habitat features are most important for a species.

Limatambo 2 at 2:40! See you there 🤩

Joe Nash (joe@jna.sh)

2026-02-18 07:33:08

Whilst we’re sharing ICTC sessions, we have a participatory session on ML evaluation metrics and their interpretation for conservation use cases, today at 4.20-5pm in Cusco 1, part of the AI in the Wild block. Would love to hear people’s experiences and best practices if you can make it!

😎 Sara Beery

Lilli (lilli@mbari.org)

2026-02-18 20:55:42

A new FathomNet Kaggle competition is live! Please share this with your data science and machine learning networks to spread the word!

kaggle.com

FathomNetCLEF2026 @ LifeCLEF & CVPR-FGVC

Positive-Unlabeled Object Detection in Marine Images

Original URL: https://www.kaggle.com/competitions/fathomnet-2026

❤️ James Liu, Jose Ruiz-Munoz, Xiangteng He, Robin Zbinden, Cameron Trotter, Oisin Mac Aodha, Morgan Langley, Sara Beery, murilo, Kevin Barnard, Spencer Bialek, Sarah

🙌 Sonny Burniston, Sara Beery, murilo, Nico Lang, Elizabeth Campolongo, Alexander Merdian-Tarko, Sarah

🐟 Brian Geuther, Sarah

Lukas Picek (lukaspicek@gmail.com)

2026-02-19 07:33:48

📣 Call for participation: AnimalCLEF 2026

You’re given labeled images of known individuals (Train) and unlabeled images (Test).
The test set includes a mix of known individuals (seen in the database) and new individuals you need to discover via clustering.

Species in this challenge:

Loggerhead sea turtle (Zakynthos, Greece)

Texas horned lizard (Texas, USA)

Eurasian lynx (Czechia)

Salamanders (Czechia)

Key dates (CET):

Competition started Feb 1, 2026

Final deadline May 7, 2026

🔗 Join here: https://www.kaggle.com/competitions/animal-clef-2026/

👏 Oisin Mac Aodha, murilo, Jason Holmberg (Wild Me), Xiangteng He, Kevin Barnard, Sonny Burniston, Mitchell Rogers, Amee Assad

🙌 Kostas Papafitsoros, Shir Bar, Robin Zbinden, murilo, Nico Lang, Elizabeth Campolongo, Jason Holmberg (Wild Me), Malte Pedersen, Xiangteng He, Sonny Burniston, Alexander Merdian-Tarko, Dan Morris, Aleksis Pirinen

🐢 Mitchell Rogers

Sara Beery (sara.beery@gmail.com)

2026-02-19 13:19:28

AI for Conservation Lunch at ICTC!! Thanks @Timm Haucke for organizing 🤩

❤️ Suzanne Stathatos, Timm Haucke, Xiangteng He, Shir Bar, Louis Freeland-Haynes, Joe Nash, Jason Holmberg (Wild Me), Matt Hron, Lukas Picek, Ted Schmitt, Justin Kay, Robin Zbinden, Mitchell Rogers, murilo, Arjun Subramonian (they/them), Youssef Bayouli, Elizabeth Campolongo, Prabath Gunawardane, Ethan White, Mamoon, Hammed Akande, Shayan Azhar, Alexander Merdian-Tarko, Hiroto, Cameron Trotter, Tanya Berger-Wolf, Owen Martin, Lauren Harrell, Boe Zhou, Kishore Panaganti, Julia Chae, Andrea Zampetti, Otto Brookes, Sam Tymorek, Carly Batist, Sarah, Aleksis Pirinen, Md Adith Mollah

🎉 Dan Morris, Shayan Azhar, Jenna Kline, Lauren Harrell, Julia Chae, Hannah Murray, Carly Batist, Sarah, Md Adith Mollah

🙌 Robin Sandfort, Shayan Azhar, Lauren Harrell, Julia Chae, Sarah

Ben Weinstein (benweinstein2010@gmail.com)

2026-02-23 15:21:27

As part of my work with Fish and Wildlife and Bureau of Ocean Energy Management, we are building active learning systems for large scale review and model training for airborne biodiversity detection in marine areas. We are starting to see alot of progress is being able to quickly look through millions of images and identify target species. I was doing a speed run on improving dolphin classification, and found some interesting images covering other large marine animals. For example, images that look alot like basking sharks (Massachusetts) and tiger sharks (North Carolina) Are there any shark experts out there that would benefit from us releasing taxa specific detection models? For example, a 'seal' model a 'shark model', or better to combine it all into single detection model 'Object of interest', with subsequent classification on each box? This is part a broader discussion i'm having with @Serge Wich, @Dan Morris and others on a global airborne detector. These are from piloted aircraft so they are suitable for detection, and maybe a general size measurement. I'm not sure how many people are working in this area for offshore marine predators and would like to discuss the best way for us to release models through deepforest. I'm hoping that long term these observations will connect to fathomnet, but decision is not mine. In general, are these one off observations ecologically valuable for others, should we make the effort to get them into GBIF?

Steve Haddock (haddock@mbari.org)

2026-02-23 15:30:42

@Danelle Cline

Shir Bar (shirbar@tauex.tau.ac.il)

2026-02-23 15:36:39

This is super cool!!!
@Alexandra DiGiacomo this might be relevant to you and your sharks?
And not sure if this would be relevant for satellite imagery as well (I guess not?) but if yes then @Holly Houliston with whales.

👍 Alexandra DiGiacomo

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2026-02-23 15:43:19

We have aerial right whales (2x species), belugas, and bowhead whales. Would need permission to share but happy to ask. Belugas are already NOAA data.

Ben Weinstein (benweinstein2010@gmail.com)

2026-02-23 15:45:44

Drone or piloted aircraft? Definitely would like to think about how we put all of this together, both from the data side, but also, just how we release a model. Generalization among resolutions, i think i've posted here before, is still really hard compared to my hopes. But yes, i'd love to integrate that into our release, provided any ESA needs are met for right whales.

Danelle Cline (dcline@mbari.org)

2026-02-23 15:55:47

We have drone data taken around 60 meters with a variety of species, including sharks. It would be great to have a common data repository to create more general models. @Duane Edgington can tell you more about the detection model we have which may be of use to this community.

👍 Ben Weinstein, Alexandra DiGiacomo, Ciara Sypherd

Ben Weinstein (benweinstein2010@gmail.com)

2026-02-23 16:11:07

Do these data live in fathomnet archive? Are there other airborne datasets in there, just guessing by the mbari name .png

Danelle Cline (dcline@mbari.org)

2026-02-23 16:14:12

Our data is available at https://drone.mbari.org/ @Steve Haddock can confirm if sharing is ok. One needs an account to access it.

👍 Ben Weinstein

Alexandra DiGiacomo (alexandra.digiacomo@stanford.edu)

2026-02-23 18:52:17

Happy to chat more about this! I largely work with single-species data (white sharks) for which we have a ~2000 img dataset that will be released with an upcoming publication. A general detection model with species-specific classification initially seems more versatile than taxa-specific models, leaving it open to the possibility for incorporating biotic/prey context (e.g. marine mammals).

Steve Haddock (haddock@mbari.org)

2026-02-23 19:03:05

@Danelle Cline giving Ben a login to the system is totally fine with me.

🆗 Danelle Cline, Ben Weinstein

Ben Weinstein (benweinstein2010@gmail.com)

2026-02-23 20:05:11

Great I've got a spreadsheet of about 400 or so projects (not just marine) and writing this into next round of development.

👍 Michael Bunsen

🔥 Kalindi Fonda

Ben Weinstein (benweinstein2010@gmail.com)

2026-02-26 12:54:59

@Danelle Cline are you experienced in airborne shark ID, or can recommend anyone who can give our annotators some tips in both the gulf of mexico and atlantic. we are seeing atleast 5 species, just by size there are some very large individuals. Blue sharks, basking sharks, great whites?, tiger sharks, spinner sharks.

Ben Weinstein (benweinstein2010@gmail.com)

2026-02-26 12:55:46

(plus common dolphins this morning)

Steve Haddock (haddock@mbari.org)

2026-02-26 13:35:54

@Ben Weinstein I would reach out to Francesco Ferretti. https://fishwild.vt.edu/faculty/ferretti.html

Alexandra DiGiacomo (alexandra.digiacomo@stanford.edu)

2026-02-26 13:48:16

@Ben Weinstein We (Barb Block's group at Stanford) could definitely assist with aerial ID across pelagic species in those regions. Happy to provide drone-based experience and biological/morphological context to support classification.

👍 Steve Haddock

Danelle Cline (dcline@mbari.org)

2026-02-26 17:20:09

Sorry, @Ben Weinstein I'm not a shark expert but it looks like you have some good leads here.

👍 Ben Weinstein

Laura Chrobak (lchrobak@mbari.org)

2026-03-16 14:01:21

@Ben Weinstein Really excited to hear you are working on scaling review for biodiversity detection! FathomNet has also been approaching this challenge. I presented on FathomNet’s behalf at OSM on the topic,“Scaling Marine Image Annotations with AI-enabled, Human-In-the-Loop Pipelines”. Per this work we have developed a bulk review UI for the FathomNet database specifically for facilitating this kind of workflow, and I’d be happy to chat more with you about it if your interested and or if you would like to leverage it. Additionally, our team would be glad to assist with submitting your data to the FathomNet database - feel free to reach out to Kevin or me directly if so!

Brett Buzzanga (brett@conservationta.org)

2026-02-24 17:28:53

@Brett Buzzanga has joined the channel

Yu-Cheng Wei (yuchengwei42@gmail.com)

2026-02-24 21:22:01

@Yu-Cheng Wei has joined the channel

Bastiaan den Braber (bastiaan@zambezi-zero.com)

2026-02-24 21:28:29

@Bastiaan den Braber has joined the channel

Aleena Munshi (aleenamunshi001@g.ucla.edu)

2026-02-25 02:25:34

Hi everyone! I’m working on a large-scale bioacoustics project with terabytes of passive acoustic recordings from the Amazon in Peru. We’re planning to run BirdNET across the full dataset on an HPC cluster, but we're concerned about the verification step. Even with a relatively low false positive rate, the total number of detections could be enormous.
For those working at similar scales, how are you handling verification? Are you validating a fixed percentage of detections, stratifying by confidence scores, using embedding-based filtering, or relying on other workflows? I’d really appreciate hearing what approaches have been effective.

👀 Joe Nash, Timm Haucke, Ben Weinstein, Karthik Ashok

😎 Jon Van Oast

👍 Sam Lapp

Joe Nash (joe@jna.sh)

2026-02-25 03:19:13

Very cool! Do you have any existing labelled data at all for the species/region @Aleena Munshi ?

Justin Kitzes (justin.kitzes@pitt.edu)

2026-02-25 08:42:55

All great questions, and I think you're thinking the right way here - I'll ask someone from my lab to get in touch to help with advice

❤️ Joe Nash, Sara Beery, Aleena Munshi

Aleena Munshi (aleenamunshi001@g.ucla.edu)

2026-02-25 15:47:20

@Joe Nash All of the data that we have collected is completely unlabelled, so we were hoping to see what BirdNet produced in a first pass!

👍 Joe Nash

Carly Batist (carly.h.batist@gmail.com)

2026-02-26 14:44:32

I would suggest starting by using a small subset of the dataset just to see how BirdNet does and to be able to test different approaches (e.g., by confidence scores) and verification processes. It will save a lot of time and compute to be able to iterate on a smaller subset, and it gives you an initial idea of how well the model seems to be doing!

In my own experience, I verified all detections but that was for a a single-species model, so it wasn't that much. It depends to some extent as well on what the downstream task is and what the constraints of that are. E.g., are you doing occupancy modelling, SDMs, etc.? The type of analysis you're using the detections in will determine how flexible you can be with potential false positives

✅ Aleena Munshi

Aleena Munshi (aleenamunshi001@g.ucla.edu)

2026-02-27 14:35:03

@Carly Batist Thank you so much! We're likely doing occupancy modelling, so working with a subset at first seems to be a good idea.

Sam Lapp (sam.lapp@pitt.edu)

2026-02-28 07:49:55

Yep I'm happy to follow up to talk about the approach, we do this quite a lot now! Basically for occupancy modeling you need to eliminate false positives, but you only need one confirmed detection per site (spatial replicate) per “visit” (temporal replicate). Therefore for each species you want to review the highest N scoring audio clips per site per visit, and once you confirm the species is present at that site- visit you dont need to annotate any more clips from that site-visit. This results in a detection history with imperfect detection but no false positives, which is suitable for occupancy modeling. You choose your temporal visit replicate design and N, how many clips to review per site-visit.
We have an open source desktop gui called Dipper for making this review process efficient. Feel free to reach out via email or slack to discuss further

❤️ Joe Nash, Aleena Munshi

👍 Carly Batist

Justin Kitzes (justin.kitzes@pitt.edu)

2026-02-28 09:36:48

Should have also mentioned Lydia's paper, which tries many options available in principle for this problem, the one Sam described is our preferred one at this point https://www.nature.com/articles/s41598-025-95207-3

Nature

A comparison of statistical methods for deriving occupancy estimates from machine learning outputs

Scientific Reports - A comparison of statistical methods for deriving occupancy estimates from machine learning outputs

Original URL: https://www.nature.com/articles/s41598-025-95207-3

❤️ Aleena Munshi

👍 Carly Batist, Abram Fleishman

Rachel Walsh (rachel.walsh@floridamuseum.ufl.edu)

2026-02-25 15:30:04

@Rachel Walsh has joined the channel

Mohamed Elabbas (mohammedelabbas@outlook.com)

2026-02-25 19:11:36

@Mohamed Elabbas has joined the channel

Trevor Cobb (trevorstormcobb@gmail.com)

2026-02-25 22:21:50

@Trevor Cobb has joined the channel

Kate Reed (k.reed@cornell.edu)

2026-02-26 11:24:10

@Kate Reed has joined the channel

Niki Amini-Naieni (niki@adainnovations.ai)

2026-02-26 12:12:58

Hi I am looking for audio datasets which have clearly repeating sounds with the number of repetitions labeled. I am wondering if anyone has any recommendations.

Michael Burke (michael.burke@thelandapp.com)

2026-02-28 19:05:35

@Michael Burke has joined the channel

Mark Thomas (mark.j.thomas8@gmail.com)

2026-03-02 09:39:48

@Mark Thomas has joined the channel

Taneea S Agrawaal (taneea.agrawaal@gmail.com)

2026-03-02 12:43:29

@Taneea S Agrawaal has joined the channel

Rodrigo Oyanedel (royanedel@marfutura.org)

2026-03-02 12:44:42

@Rodrigo Oyanedel has joined the channel

Anthea (u20472006@tuks.co.za)

2026-03-03 02:51:33

@Anthea has joined the channel

Paul Allin (allinpaul@gmail.com)

2026-03-03 04:47:25

Hi all, just wanted to share a publication (http://doi.org/10.1002/wlb3.01523) from my PhD using AI to detect and identify large wild mammals in African savanna from nadir (drone 2.3cm GSD) imagery. The data are all open source, including all annotations and code we created. I hope that this can be useful and very open to comments and collaborations.

🙌 Viktor Domazetoski, Jason Holmberg (Wild Me), Alexander Merdian-Tarko, Eric Snyder, Jenna Kline, Trevor Cobb

😎 Jason Holmberg (Wild Me), Lucia Gordon

Paul Allin (allinpaul@gmail.com)

2026-03-03 05:15:38

Feel free to contact me or @Fadel Seydou

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2026-03-03 10:08:42

Very cool!

Daniel Haider (leinad.rediah@gmail.com)

2026-03-03 05:46:43

@Daniel Haider has joined the channel

Ronan Wallace (ronanlwallace@gmail.com)

2026-03-03 16:01:00

@Ronan Wallace has joined the channel

Ida Heijmink (iheijmink@gmail.com)

2026-03-03 16:50:17

@Ida Heijmink has joined the channel

Mariah Meek (mariah@icatch.app)

2026-03-03 17:00:04

@Mariah Meek has joined the channel

Moeri Okuda (moeriena.abc@gmail.com)

2026-03-03 17:07:48

@Moeri Okuda has joined the channel

Kathy Lin (klin020709@gmail.com)

2026-03-03 19:38:25

Hello,

My name is Kathy, and I’m a rising senior in high school from Cambridge, MA, interested in applying AI and machine learning to climate and environmental challenges.

I developed a drone-based AI system to inspect wind turbines for surface damage and am currently building an edge-deployable vision-language object detection model to support post-disaster search and rescue. Last summer, I interned at the Princeton Visual AI Lab researching knowledge distillation for model compression.

I program in Python, Java, and C++, qualified for USACO Gold, and have extensive experience with PyTorch and TensorFlow. I’ve completed Calculus BC and am currently taking Multivariable Calculus & Linear Algebra, and AP Physics C.

I’m seeking an unpaid summer internship in computer vision or machine learning and would be grateful for the opportunity to connect. Thank you for your time and consideration!

Kathy

👏 Hammed Akande, Robin Zbinden, Aly Mtsumi, Alexander Merdian-Tarko

Kevin Barnard (kbarnard@mbari.org)

2026-03-04 09:40:10

Hi @Kathy Lin, MBARI is currently accepting applications for the summer internship program until March 9. Details and project descriptions are available here: https://www.mbari.org/about/careers/internships/

MBARI

Internships • MBARI

Applications for the 2026 MBARI Summer Internship program are NOW OPEN. This 10-week internship allows college students and educators to work on a research or outreach project at MBARI.

Est. reading time

16 minutes

Original URL: https://www.mbari.org/about/careers/internships/

❤️ Aleena Munshi, Kathy Lin

Sara Estlander (sara@fosonline.org)

2026-03-04 04:40:52

@Sara Estlander has joined the channel

Frank Fundel (frank.fundel@outlook.com)

2026-03-04 05:08:16

@Frank Fundel has joined the channel

McKalee Suzanne Steen (mckalee_steen@berkeley.edu)

2026-03-04 18:49:07

@McKalee Suzanne Steen has joined the channel

stefano puliti (stefano.puliti@nibio.no)

2026-03-06 06:56:21

PhD opportunity to join #NextBON, a newly funded EU project aimed at developing a biodiversity monitoring network for Europe.

The candidate to lead the work on improving biodiversity computer vision based on nature images 📸 🐞🌱🐠

The position is based in Zurich at EcoVisionUZH supervised @Jan Dirk Wegner and @Damien Robert and will be working closely with us NIBIO and several NFIs and monitoring networks across Europe

Link to job posting 👇
https://www.linkedin.com/jobs/view/4379638175

linkedin.com

University of Zurich hiring Doctoral Candidate in computer vision and machine learning for developing novel deep learning methods for plant and animal species identification from camera-trap and crowd-sourced imagery in Zurich, Zurich, Switzerland | LinkedIn

Posted 7:32:22 AM. The EcoVision Lab in the Department of Mathematical Modeling and Machine Learning (DM3L) at…See this and similar jobs on LinkedIn.

Original URL: https://www.linkedin.com/jobs/view/4379638175

🎉 Cameron Trotter, Oisin Mac Aodha, Sara Beery, Malte Pedersen, Aleksis Pirinen, Jagoda, Md Adith Mollah

Francisco Gonzalez (fjgonzalez@google.com)

2026-03-06 12:02:36

@Francisco Gonzalez has joined the channel

Ben Weinstein (benweinstein2010@gmail.com)

2026-03-06 13:48:49

I'm playing with the idea of hosting a google summer of code project in which we have to re-align polygons of annotations made in flight by observers moving at high speeds. There are multiple sources of spatial error, leading to shifts compared to other remote sensing products. The goal is a model that uses the image features to re-align to the remote sensing imagery. In this case talking about dead trees. Two ways i'm thinking about it, 1) train an affine-shift model based on image features (hard!), 2) use the specific task of this project to identify dead trees and then shift the polygon overtop (probably brittle). My interest here is in recovering decades of data from an airborne survey of western forests. I think there are many projects out there like this, in which data were collected without the future knowledge of AI tools, and moving them towards 'AI ready'. If people know of analogous projects or ideas about annotation matching. I am very aware of the circularity of using a model to align annotations to train a model for classification, I'm told that the general features (clumps of dead trees) can help define the shifting, whereas the specific features (pattern of dying, color, tree species) point to specific pathologies (bark beetles versus algids versus drought). I like these kinds of projects because they 1) get teams that don't see themselves as AI users involved, 2) Have the high payoff of huge data catalogs unlocked, 3) are at the bounds of what we know is possible/reasonable in our community. I'm also considering just throwing out the annotation entirely and using it as just a weak signal for weak labeling approaches!

❤️ Owen Martin, Sara Beery, Md Adith Mollah, Venkanna Babu Guthula

🌲 Dan Morris, Md Adith Mollah

Benjamin Hoffman (benjaminsshoffman@gmail.com)

2026-03-06 13:55:43

This is a big leap, but we ran a challenge at last year’s BioDCASE where participants had to align audio recordings made by spatially separated, desynchronised audio recorders. There was affine shift, plus time warping. It was very hard to beat the baseline of doing nothing.

The best solutions did something like: propose multiple possible alignments, score possible alignments based on patch-level features, choose the alignment with the best score. People submitted supervised models (train on the dataset of interest with some known alignments) plus unsupervised models.

👍 Ben Weinstein

❤️ Owen Martin

Ben Weinstein (benweinstein2010@gmail.com)

2026-03-06 13:57:01

I love that doing nothing is a good alternative, I actually find that super interesting. Something we don't talk about much.

Amanda Swayze-Rebeaud (alswayze@alumni.ncsu.edu)

2026-03-06 16:58:08

@Amanda Swayze-Rebeaud has joined the channel

Prabath Gunawardane (prabathg@gmail.com)

2026-03-06 22:10:36

I'm excited to share our new contribution to lila.science : https://lila.science/datasets/felidae-conservation-fund

This dataset contains 357k+ images collected across 216 camera traps in the San Francisco Bay Area, and was collected by Felidae Conservation Fund as part of our research for the Bay Area Puma Project and Bay Area Bobcat Project, Please let me know if you have any questions about the dataset 🐈 🙂

Thank you @Dan Morris for hosting and all the help getting the dataset ready!

LILA BC

Felidae Conservation Fund 2020-2025 - LILA BC

Overview This dataset contains 357k+ images collected across 216 camera traps in the San Francisco Bay Area and surroundings. This data was collected by Felidae Conservation Fund as part of their research for the Bay Area Puma Project and Bay Area Bobcat Project. There are 66 category labels, most of which are species. However, there...

Written by

lilawp

Est. reading time

2 minutes

Original URL: https://lila.science/datasets/felidae-conservation-fund

🐱 Dan Morris, Timm Haucke, Yseult Hb, Mitchell Rogers, Sara Beery, Kyra Swanson, Elizabeth Campolongo, Md Adith Mollah, Marion Richardot, Francisco Gonzalez

❤️ Hugo, Aakash Gupta, Md Adith Mollah, Marion Richardot

Maria Camila (lposada@cicese.edu.mx)

2026-03-08 17:50:15

@Maria Camila has joined the channel

Philippe Questel (pquestel@g.ucla.edu)

2026-03-09 03:54:04

@Philippe Questel has joined the channel

Greg White (greg.white@rspb.org.uk)

2026-03-09 10:47:15

@Greg White has joined the channel

Robert Pless (pless@email.gwu.edu)

2026-03-09 10:53:47

Hey everyone,
I joined a while ago and have been lurking and enjoying seeing everything!

I saw the dataset that @Prabath Gunawardane posted a few days ago (~350K images from ~200 camera traps). I've been working on text/image search for a while so my student Shrenik wired this dataset up with CLIP text search, allowing you to search with any natural language. With the warning that the first this this page does is download a ~750 megabyte feature index, You can try it here: https://shrnik.github.io/felidae-image-search/

I've attached top search images for:
dry, dusty trail
wet, muddy trail
bird flying
sunset
trash
foggy
grazing

👏 Xiangteng He, Filippo Varini, Ciara Sypherd

❤️ Prabath Gunawardane

Sara Beery (sara.beery@gmail.com)

2026-03-09 11:42:48

Nice! We've done some similar analysis on Snapshot USA. One thing we've seen is that the autocorrelation in camera trap data ends up skewing the search quite a bit (lots of very similar returns for a query coming from, eg, one site). Have you played around with filtering the search at all?

Sara Beery (sara.beery@gmail.com)

2026-03-09 11:43:39

We also found that SigLIP seems to work much better than CLIP for camera trap data if the queries are more behavioral, for example!

Prabath Gunawardane (prabathg@gmail.com)

2026-03-09 12:39:56

This is awesome, plus I'm so happy to see the dataset being used already 😁. I'll play with it and also share with our ecologists and be back in touch!!

Robert Pless (pless@email.gwu.edu)

2026-03-09 10:54:33

[ MESSAGE TEXT EMPTY ]

Robert Pless (pless@email.gwu.edu)

2026-03-09 10:55:15

These are somewhat cherry picked examples; CLIP isn't so good at understanding dynamic behaviors, but I thought the ability to explore free form search of the data might be compelling for some of you! Happy to explore collaborations to make this free form search better for particular application domains...

❤️ Prabath Gunawardane, Marion Richardot

SHRENIK BORAD (s.borad@gwmail.gwu.edu)

2026-03-09 11:13:32

@SHRENIK BORAD has joined the channel

Sayan Patra (sayan.patra@gwmail.gwu.edu)

2026-03-09 15:36:40

@Sayan Patra has joined the channel

Taylor Gill (taylorgill249@gmail.com)

2026-03-09 16:17:40

@Taylor Gill has joined the channel

Carly Batist (cbatist@gradcenter.cuny.edu)

2026-03-09 19:40:47

💰 Funding opp! 💰

Applications are open for the 2026 WILDLABS Awards! With the support of Arm, this year’s Awards will distribute $350,000 to advance conservation technology across up to 15 projects. There are two grant levels of $10,000 and $50,000, designed to support projects at both early and more advanced stages of conservation technology deployment and development.

Expressions of interest are due by March 18!

Details around eligibility, requirements, and timelines all here: https://wildlabs.net/funding-opportunity/wildlabs-awards-2026-express-your-interest-now

🎉 Jon Van Oast, Bernie Boscoe, Dan Morris, Eric Snyder, Kalindi Fonda, Sarah, Spencer Bialek, Laura Berman

Filippo Varini (fppvrn@gmail.com)

2026-03-10 05:50:30

Hello everyone.
@Kalindi Fonda and I are hosting a Vibe Coding Party on March 17th from 4pm to 7pm UK time (online) 😁 .

A Vibe Coding Party is a way for us to learn vibe coding tools. We will gather online use these tools to build a personal project. We will have the opportunity to learn from each other by sharing insights and receiving support from other vibe coders.

You can find more about the event here.

Everyone is welcome to join. Even if you have never programmed before, consider joining! These tools give us unprecedented opportunities to build websites/visualisations/games without needing to write code.

If you are interested in joining, please sign up by filling this form https://forms.gle/P6ZTTx41HD8wRa5s6

Thanks @Dan Morris for kickstarting these events and inspiring us to host more 🔥

👍 Sara Estlander, Carly Batist, Louis Fiorina, Dan Morris, Justine Boulent, Molly Blank, Mikel, Maurits

🌊 Kalindi Fonda, Alexander Merdian-Tarko

❤️ Jennifer, Prabath Gunawardane, Devi Ayyagari, Elizabeth Campolongo, Hayley Smith

🧡 Abhi M

Sara Beery (sara.beery@gmail.com)

2026-03-10 17:38:45

The call for applications for CV4Ecology 2027 instructors is up!!

Help spread the word:

https://x.com/cv4ecology/status/2031482823773757594

https://www.linkedin.com/in/sara-beery-1ba89166/

https://bsky.app/profile/sarameghanbeery.bsky.social/post/3mgqdechiec26

X (formerly Twitter)

CV4Ecology Workshop (@cv4ecology) on X

We're looking for our next cohort of amazing instructors for #CV4Ecology!! CV4E is an intensive workshop designed to empower ecologists to efficiently process their existing data, design new studies around CV, and scale their research to larger datasets.

Original URL: https://x.com/cv4ecology/status/2031482823773757594

linkedin.com

Sara Beery - Massachusetts Institute of Technology | LinkedIn

I&#39;m an assistant professor at MIT EECS&#39; Faculty of AI and Decision Making and CSAIL, and… · Experience: Massachusetts Institute of Technology · Education: Caltech · Location: Pasadena · 500+ connections on LinkedIn. View Sara Beery’s profile on LinkedIn, a professional community of 1 billion members.

Original URL: https://www.linkedin.com/in/sara-beery-1ba89166/

bluesky

Sara Beery (https://bsky.app/profile/sarameghanbeery.bsky.social)

Original URL: https://bsky.app/profile/sarameghanbeery.bsky.social/post/3mgqdechiec26

🎉 Jon Van Oast, Kalindi Fonda, Dan Morris, Brian Geuther, Madison Van Horn, Shir Bar, Peter Bull, Keenan Johnson, Prabath Gunawardane, Jose Ruiz-Munoz, Irina Tolkova, Laura Chrobak, Julia Chae, Md Adith Mollah, Filippo Varini, Marion Richardot

Sara Beery (sara.beery@gmail.com)

2026-03-10 17:46:12

And check out the feature on the program in Smithsonian magazine!!

https://www.smithsonianmag.com/blogs/national-zoo-conservation-biology-institute/2026/03[…]ists-learn-ai-skills-at-the-national-zoos-science-campus/

💪 Kalindi Fonda, Shir Bar

Kalindi Fonda (kalindi.fonda@gmail.com)

2026-03-10 17:47:36

Haha, you all look way colder than we were in LA 😅

Dan Stowell (dan.stowell@naturalis.nl)

2026-03-11 03:39:01

Pleased to announce the BioDCASE Challenge 2026 - the Bioacoustics Data Challenge! Now with 6 different tasks to choose from, each an open-data computational challenge of its own: https://biodcase.github.io/challenge2026/

biodcase.github.io

BioDCASE Challenge - BioDCASE

Tasks We are excited to announce the 2026 BioDCASE task...

Original URL: https://biodcase.github.io/challenge2026/

❤️ Joe Nash, Oisin Mac Aodha, Sara Beery, Céline Angonin, Rupa Kurinchi-Vendhan, Jagoda, Eric Snyder, Jason Holmberg (Wild Me), Julia Wilkins, Ragib Nihal, Aleena Munshi, Irina Tolkova, Vincent Lostanlen, Greg White, Md Adith Mollah, Spencer Bialek

🎉 Ben McEwen, Sara Beery, Céline Angonin, Rupa Kurinchi-Vendhan, Jose Ruiz-Munoz, Jason Holmberg (Wild Me), Julia Wilkins, Burooj Ghani, Vincent Lostanlen

Daniel Thornton (daniel.thornton@wsu.edu)

2026-03-11 23:43:18

@Daniel Thornton has joined the channel

Thongthornpatch Chamadol (thongthornpatch@gmail.com)

2026-03-12 04:16:40

@Thongthornpatch Chamadol has joined the channel

Matteo Contini (lombardata@gmail.com)

2026-03-12 04:34:12

@Matteo Contini has joined the channel

Tedi Yankov (teodor.yankov@new.ox.ac.uk)

2026-03-12 04:45:29

@Tedi Yankov has joined the channel

Md Adith Mollah (adibhasan35@gmail.com)

2026-03-12 04:47:53

@Md Adith Mollah has joined the channel

Mary Feliz (mkfeliz@mit.edu)

2026-03-12 06:43:07

@Mary Feliz has joined the channel

Salma Daadouch (Salma.Daadouch@vetmeduni.ac.at)

2026-03-12 07:02:45

@Salma Daadouch has joined the channel

Filippo Varini (fppvrn@gmail.com)

2026-03-12 08:58:19

Hi!
I wanted to ask some guidance to folks who have trained AI species classifiers from images/videos for partner ecologists/organisations (both land and marine).

My partner organisation has TONS of videos. They'd like AI models to detect and classify their species. The problem is that they have NO AI-ready annotations and only 30% of the data is annotated with MaxN.

Based on your experience, what's the best approach to follow to gather the right data to train species classifiers, assuming we have a preliminary detector of animals, like the Megadetector?

I had two ideas in mind, but I would love to hear your thoughts if you faced similar challenges.

On one hand, we could explain them how to annotate the data for AI models (generate csv or put the crops in folders), hand off the task to them, while assisting throughout. On the other hand, we could create a software pipeline to speed up the annotation process (something like preliminary detector -> crops -> UI for classification -> UI button to send the classified images to us). That could maybe speed up the process, but would be more time consuming?

What do you advise based on your expertise? What tools are out there that you think I should consider (the domain is underwater video, but anything from land ecology would be interesting).

Thank you so much! Would be great to take inspiration from the community instead of re-inventing the wheel 🙂

Sara Beery (sara.beery@gmail.com)

2026-03-12 09:12:35

I think taking an active or iterative approach is really helpful, don't try to label everything and then train. Embed everything with a good feature encoder (Dinov2/3 is great for this), start with a few labels, train a linear or a simple classification head, check a few results, iterate. Optimal criteria for what to check can vary wildly based on the data and the goal.

👍 Malte Pedersen, Sarah, Filippo Varini, Devi Ayyagari, Marion Richardot, Laura Chrobak

Malte Pedersen (mape@create.aau.dk)

2026-03-12 09:37:52

Depending on the exact task and environment, it may also be a problem to figure out the "true ground truth" and when to annotate/not annotate an object in marine videos. E.g., how certain do you need to be before labeling an object? Especially in underwater videos the objects can sometimes be incredibly difficult to classify even when they are relatively close to the camera due to turbidity, natural disguise, etc. If it is by any means possible, I suggest having a "representative" test set which is fully labeled by multiple people, to get an understanding of the uncertainty involved in the annotation process. The labeling uncertainty not only affects training, but also the evaluation. Knowing something about the human label-uncertainty also makes the discussion about "when is the model good enough" more informed.

👍 Sarah, Kalindi Fonda, Filippo Varini, Sara Beery, Devi Ayyagari

Ben Weinstein (benweinstein2010@gmail.com)

2026-03-12 09:49:45

Definitely agree with @Sara Beery . Pick a couple dominant classes train a model, choose an active learning strategy, for example lowest confidence score, and iterate. I often separate detection and classification models because you go through cycles of needing to focus on one or the other. Your goal should be rapid iteration, training on new annotations every few days. Defining at the start what the organization goal is, not just a "classifier" but what species needs a specific accuracy, or is it just screening away the most common classes so a human can look at anything else. Planning on human review of inference helps build trust.

👍 Filippo Varini, Marion Richardot, Sara Beery, Elizabeth Campolongo, Devi Ayyagari, Danelle Cline

Filippo Varini (fppvrn@gmail.com)

2026-03-12 10:22:42

Thanks everyone!

Devi Ayyagari (akdevi.92@gmail.com)

2026-03-12 10:37:54

Agree with all the comments so far. I have also had some success with clustering DINOv2/v3 embeddings to determine how to choose the images to annotate/review. The workflow that I am playing with looks something like this
preliminary detector -> crops -> embeddings -> clustering -> choosing data to annotate from different clusters based on use case -> UI for review.
Also, considering phenotypic similarity rather than strict taxonomic grouping when clustering species has been helpful for me.

👍 Marion Richardot, Sara Beery, Filippo Varini

Tedi Yankov (teodor.yankov@new.ox.ac.uk)

2026-03-12 13:41:46

One approach that might be good to explore is using CLIP! In case you haven't encountered it before - it's a dual-encoder model pre-trained on a massive amount of text-image pairs from the web. It learns a shared embedding space and it does zero-shot classification by embedding a new image and finding the text class closest to the newly embedded image in the multimodal embedding space. Useful for when you have no labels!

I've used it on a dataset of a bunch of frames of sharks in the wild and supplied only with 10 annotated examples as in-context learning (ie few-shot CLIP classification) it already got pretty good results on classifying between my 4 target species.

This might be a useful tutorial: https://medium.com/@myscale/zero-shot-classification-with-openais-clip-6c50c25cf094

Medium

Zero-shot Classification with OpenAI’s CLIP

Traditional image classification models, such as Convolutional Neural Networks (CNNs), have been the cornerstone of computer vision tasks…

Reading time

8 min read

Original URL: https://medium.com/@myscale/zero-shot-classification-with-openais-clip-6c50c25cf094

🙌 Filippo Varini

Michael Yair (m1cha3l.ya1r@gmail.com)

2026-03-16 07:41:32

First, our system has different goals, it is to serve several groups of researches.
We label only the relevant to a certain research, but we started from the same statement "to label all the images".
It is a compromise (cost, time, available resources) that might not be relevant to you, but might inspire to find your path.
I did as you suggested, first train a generic model (based on detectron2, binary to the class taxonomy level) to identify the relevant subject in the image and box them.
Also tailored a labeling solution to the group of users mode of operation - the labeling persons prefer to have a folder full of the cropped images, create on their PC a folder per species and drop each image into the relevant folder.
from there to the system is a simple pipeline, the complicated side is the polling between mismatches (each user might label differently).
What I'm trying to say is simply ask your customers - what is their mode of operation and seek together how to optimize it.
It is important to mention the level of classification, if it is in the level for generic common names (cat, dog, whale...) or should be bound to a taxonomy database. In my case it is bound to a local taxonomy database, while some prefer to identify to the family/genus level some very strict and prefer species level - which complicate a bit the database structure and the labeling.

🙌 Filippo Varini

Danelle Cline (dcline@mbari.org)

2026-03-16 15:04:28

@Filippo Varini We have found across domains from drone to midwater to plankton imagery, separating the detection from the classification yields the best performance. I agree with most of the advise above, and would also add to not forget to label artifacts which may be confused with your classes.

In case it is useful, we built a tool to help with that initial two-step processing called sdcat - this uses megadetectors, saliency detection, and transformer models like DINO. If you have already done the detection, format your data into a compatible csv and run the clustering step. We have found that smaller block sizes, e.g. 8 versus 14 or 16 offer better performance if your targets have low pixel resolution.

A trained classifier will work best on all your common species, but your outliers/anomalies are best found using a vector search approach using e.g. cosine distance.

Good luck!

👍 Sara Beery

🙌 Filippo Varini

Elizabeth Campolongo (e.campolongo479@gmail.com)

2026-03-17 09:46:43

If you're looking for a classifier following detection, you could try BioCLIP; we have a package (pybioclip) to make the implementation easier. The docs also have some how-tos for no-fine-tune improvements.

🙌 Filippo Varini

Elizabeth Campolongo (e.campolongo479@gmail.com)

2026-03-17 09:52:25

@Maria Napolitani, you worked on something similar, right?

🙌 Filippo Varini

Filippo Varini (fppvrn@gmail.com)

2026-03-27 06:50:00

Thank you so much!

Andy Delvaux (andy.delvaux@telenet.be)

2026-03-12 09:25:08

@Andy Delvaux has joined the channel

Justin Kay (justinkay92@gmail.com)

2026-03-12 10:28:34

Hi all, I am looking for recommendations for constructing a distance-to-road geospatial layer -- does anyone have any workflows for this they recommend?

Michael Catchen (mdcatchen@gmail.com)

2026-03-12 10:58:49

Getting roads as a polygon from OpenStreetMap using osmdata in R, rasterizing it with terra, and using terra::distance is pretty straightforward. There’s probably more efficient methods if the extent is giant or resolution you want is really high

👍 Justin Kay, Matt Weldy

👍:skin_tone_2: Cara Appel

Justin Kay (justinkay92@gmail.com)

2026-03-12 11:22:57

Thanks @Michael Catchen I'll give that a try

Matt Weldy (matthewjweldy@gmail.com)

2026-03-12 12:12:13

If your looking for a distance to road raster layer I often find terra::distance crashes when my raster is too big. Depending on the use case, I will sometimes aggregate the raster to a coarse resolution, calculate distances at that resolution and then map the coarse distances back to the finescale raster. You can do that by just assigning the coarse values to overlapping fine scale cells, or by using a moving window calculation that takes the average of all the neighboring cells

👆:skin_tone_2: Cara Appel

👍 Justin Kay

Justin Kay (justinkay92@gmail.com)

2026-03-12 17:31:40

Awesome thanks @Matt Weldy

Thijs van der Plas (vdplasthijs@gmail.com)

2026-03-13 04:45:56

Alternatively if you don't want to download a raster layer; I use google earth engine to calculate this using the online GRIP4 dataset (via the python API). Happy to share the code if that's of interest!

Justin Kay (justinkay92@gmail.com)

2026-03-13 09:46:44

I do want rasters, but actually it would be great to just slot this into the rest of my GEE pipeline. Are you just using these feature collections @Thijs van der Plas?

gee-community-catalog.org

Global Roads Inventory Project - awesome-gee-community-catalog

Community Datasets in Google Earth Engine

Original URL: https://gee-community-catalog.org/projects/grip/

Thijs van der Plas (vdplasthijs@gmail.com)

2026-03-13 11:02:49

Yep those!

Thijs van der Plas (vdplasthijs@gmail.com)

2026-03-13 11:11:18

If it's useful, you can adapt/adjust this function that calculates the distance to road for every pixel in an area of interest (AOI) polygon, and then gets the mean/max of all pixels in the polgyon. (But the road doesn't need to be confined to that polygon, it uses the entire GRIP4 data collection up to some radius_max away from the polygon). You can stack multiple GRIP4 collections if needed. For me the calculation takes somewhere between 0.1-1 sec per AOI. Hope that helps! (Source code, I've just extended the docstring).

def get_distance_to_road_within_aoi(aoi, cell_size=30, radius_max=5000):
    """Calculates for each pixel in AOI the distance to the nearest road within radius_max and
    returns the maximum distance inside the AOI.
    
    Parameters:
    - aoi: Earth Engine Geometry defining the area of interest.
    - cell_size: The resolution at which to calculate the distance (in meters).
    - radius_max: The maximum search radius for roads (in meters). Pixels farther than this distance will be assigned a distance of radius_max.
    
    Returns:
    A dictionary with:
    - 'maxdist_road': The maximum distance to the nearest road within the AOI, capped at radius_max.
    - 'meandist_road': The mean distance to the nearest road within the AOI, capped at radius_max.
    """
    roads = ee.FeatureCollection("projects/sat-io/open-datasets/GRIP4/Europe")
    distance = roads.distance(searchRadius=radius_max, maxError=50)
    distance_masked = distance.clip(aoi).rename("distance")
    max_distance = distance_masked.reduceRegion(
        reducer=ee.Reducer.max(), geometry=aoi, scale=cell_size, maxPixels=1e9
    )
    mean_distance = distance_masked.reduceRegion(
        reducer=ee.Reducer.mean(), geometry=aoi, scale=cell_size, maxPixels=1e9
    )
    return {
        "maxdist_road": int(max_distance.get("distance").getInfo() or radius_max),
        "meandist_road": int(mean_distance.get("distance").getInfo() or radius_max),
    }

🙌 Sam Lapp

Justin Kay (justinkay92@gmail.com)

2026-03-13 11:31:25

awesome thanks @Thijs van der Plas!

👍 Thijs van der Plas

Sam Lapp (sam.lapp@pitt.edu)

2026-03-17 09:22:24

this is the kind of valuable info that will be sad to lose to 90 day slack history limit… @Sara Beery do you think there’s a bit of funding somewhere that would pay for longer slack memory, or does anyone else (@Dan Morris perhaps?) have a recommendation for how to capture/retain slack expert knowledge outside of slack?

Sara Beery (sara.beery@gmail.com)

2026-03-17 10:03:36

I've looked into this many times over the years. The biggest issue is that based on my prior search it would be a pretty significant cost, even if I started an NGO to get an NGO discount. Currently we're hosting an archived version that is searchable, @Timm Haucke this is a good reminder to update that archive! And maybe we should link it more visibly again in the main channel.

👍 Timm Haucke

Sara Beery (sara.beery@gmail.com)

2026-03-17 10:04:00

And if anyone does have $$ to host this as a paid slack, I'm very open to that!

Sam Lapp (sam.lapp@pitt.edu)

2026-03-17 10:04:12

ooh good to know! archive sounds great

Timm Haucke (timm@haucke.xyz)

2026-03-17 10:04:34

Will try to update the archive this afternoon!

😎 Sara Beery

🙌 Jennifer

Timm Haucke (timm@haucke.xyz)

2026-03-17 21:46:54

Also @Justin Kay, let me know if you run into rate limits on GEE. I do have a workflow for doing this exact thing using just OSM data and QuackOSM

Timm Haucke (timm@haucke.xyz)

2026-03-18 09:21:55

Our archive of old messages is updated, as always available at https://beerylab.csail.mit.edu/AIforConservationArchive/

👏 Sam Lapp, Rupa Kurinchi-Vendhan, Julia Wilkins, Thijs van der Plas, Owen Martin, Shir Bar, Declan, Robin Zbinden, Juan Sebastián Cañas, Elizabeth Campolongo, Malte Pedersen, Boe Zhou, Alexander Merdian-Tarko, Carly Batist, Roshni Biswas, Hayley Smith, Aleksis Pirinen, Sepand Dyanatkar

👏:skin_tone_3: Alan Stenhouse

Daniel Thornton (daniel.thornton@wsu.edu)

2026-03-12 11:35:58

Hi all – I am new here – a professor from Washington State University…long time camera trapper but only recently really delving into AI applications. I am happy to introduce a new dataset on LILA (https://lila.science/datasets/wsu-lynx/). 1.2 million camera trap images from 637 cameras in the mountains of northcentral Washington (collected in 2016/2017). Focused at high elevations as part of a long-term Canada lynx study. Thanks to Dan Morris for helping to put this all together. We (me, Dan Morris and others) recently used this dataset in a just accepted paper in Journal of Applied Ecology to compare multi-species occupancy models derived from AI-based and expert-based species identifications (to spoil the paper, the models were quite similar). This image dataset is part of the open access release of data and code used to perform the occupancy model comparison.

LILA BC

WSU Lynx - LILA BC

Overview This data set contains 280,485 sequences of camera trap images from 637 locations in Washington, USA, totaling 1,205,453 images. Labels are provided for 26 categories. The most common categories are “empty” (215,659 sequences), “domestic cattle” (17,929 sequences) and “odocoileus hemionus” (mule deer) (10,756 sequences). Images of humans and vehicles have been removed. Images were...

Written by

lilawp

Est. reading time

2 minutes

Original URL: https://lila.science/datasets/wsu-lynx/

🎉 Peter van Lunteren, Jason Holmberg (Wild Me), Marion Richardot, Cara Appel, Sara Beery, Dan Morris, Malte Pedersen, Xiangteng He, Filippo Varini, Elizabeth Campolongo, S M Rayeed, Shir Bar, Prabath Gunawardane, Matt Weldy, Timm Haucke, Kyra Swanson, Angela Zhu

😎 Jason Holmberg (Wild Me), Marion Richardot, Sara Beery, Timm Haucke

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2026-03-12 11:50:44

We have some new re-ID algorithms too if we can continue to support you with re-ID of individual Canada lynx in Whiskerbook.org.

🎉 Jon Van Oast

Dan Morris (agentmorris@gmail.com)

2026-03-12 13:46:14

Note to self: figure out why Slack uses that Oryx as the thumbnail for all links to LILA. In case you were wondering, no, there are not oryx in Washington. (Although working with Dan T's data did teach me that there are some camels that visit camera traps in WA!)

🐫 Cara Appel, Jason Holmberg (Wild Me), Shir Bar, Mitchell Rogers

😄 Hayley Smith

Cara Appel (appelc@oregonstate.edu)

2026-03-12 14:57:48

Very cool, thanks for sharing! @Daniel Thornton I was really interested in your talk on this project at TWS and look forward to seeing the paper

Cara Appel (appelc@oregonstate.edu)

2026-03-12 14:57:54

(also camels, what?!)

Daniel Thornton (daniel.thornton@wsu.edu)

2026-03-12 15:23:49

The megafauna of Washington....

😂 Mitchell Rogers, Yves Bas, Anna Willoughby

Cara Appel (appelc@oregonstate.edu)

2026-03-12 15:27:40

Amazing

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2026-03-12 15:29:18

My dream to bring the okapi to PNW forests is alive but yet unrealized.

Dan Morris (agentmorris@gmail.com)

2026-03-12 15:29:25

If you're wondering whose camels they are:

https://sctimeswa.com/a-nomad-like-journey-to-valley-wa/

🐐 Cara Appel

Matt Weldy (matthewjweldy@gmail.com)

2026-03-12 15:38:51

Careful with your assumptions Dan 😀 Portland to Vancouver WA is well within the dispersal range of an oryx

😏 Dan Morris

🤦‍♀️ Cara Appel

Siyuan Li (sli20@ualberta.ca)

2026-03-12 15:36:18

@Siyuan Li has joined the channel

Aditya S.K (skaditya2601@gmail.com)

2026-03-12 17:20:42

@Aditya S.K has joined the channel

Aleena Munshi (aleenamunshi001@g.ucla.edu)

2026-03-12 18:29:10

I built a SDM comparison app a while ago and might add more models (have some free time), if anyone is interested in this sort of thing. Not an expert whatsoever, but it was a fun project. https://aleenamunshi.shinyapps.io/sdm_comparison_app/

🎉 Sara Beery, Timm Haucke, Elena Sierra, Robin Zbinden, Dan Morris

Lucas Mateo Waldburger (lwaldburger@berkeley.edu)

2026-03-12 19:09:20

@Lucas Mateo Waldburger has joined the channel

Jack (jackp@sentala.org)

2026-03-13 00:34:13

@Jack has joined the channel

Khushbu Sainani (khushbusainani20@gmail.com)

2026-03-13 04:12:25

@Khushbu Sainani has joined the channel

Holger Klinck (hk829@cornell.edu)

2026-03-13 10:10:12

The BirdCLEF 2026 competition is live: https://www.kaggle.com/competitions/birdclef-2026

👏 Sara Beery, Lukas Picek, Timm Haucke, Dan Morris, Naveen Dhar, Irina Tolkova, Thijs van der Plas, Juan Sebastián Cañas, Sam Lapp

Edwin Arkel Rios (edwinarkelrios.ee08@nycu.edu.tw)

2026-03-15 06:56:21

@Edwin Arkel Rios has joined the channel

Mahny Barazandehtar (mahnybarazandehtar1998@gmail.com)

2026-03-15 08:43:48

@Mahny Barazandehtar has joined the channel

Maurits (mauritsvzb@gmail.com)

2026-03-16 13:00:28

@Maurits has joined the channel

Caroline Chan (cmchan@mit.edu)

2026-03-17 10:08:07

@Caroline Chan has joined the channel

Jack O'Hara (jack@centaurlabs.com)

2026-03-17 20:38:06

Hey everyone! I'm working on classifying + bounding a complex grayscale sonar imagery dataset. Any marine scientists/archaeologists, hydrographic surveyors, or experienced sonar operators in this group? Or anyone who might be friends with one? Would love to chat!

Sara Beery (sara.beery@gmail.com)

2026-03-17 20:41:28

@Justin Kay @Michael Hobley

Michael Hobley (mahobley@gmail.com)

2026-03-17 22:28:52

Very down to chat

Jack O'Hara (jack@centaurlabs.com)

2026-03-18 13:27:59

Thanks @Sara Beery!

🫡 Sara Beery

Timm Haucke (timm@haucke.xyz)

2026-03-18 09:31:52

set the channel topic: Contact info to join: aiforconservation@gmail.com | Archive: http://beerylab.csail.mit.edu/AIforConservationArchive/

Sara Beery (sara.beery@gmail.com)

2026-03-18 10:29:00

Fun task for a future vibe coding party: take the archive and make it searchable!

👍 Holger Klinck, Shir Bar, Jason Holmberg (Wild Me), Yseult Hb, Elizabeth Campolongo, Cheng Zhang, Timm Haucke, Carly Batist, Aleksis Pirinen

🤓 Kalindi Fonda

Timm Haucke (timm@haucke.xyz)

2026-03-18 14:33:10

For reference if anybody actually decides to look into this. We use the Slack export feature to get the attached archive (mostly JSON files) and then slack-export-viewer to convert that into static HTML files.

👍 Jon Van Oast, Sara Beery

Timm Haucke (timm@haucke.xyz)

2026-03-18 14:34:45

Bonus points for using slackdump instead, which doesn't seem to require admin privileges

Sara Beery (sara.beery@gmail.com)

2026-03-18 10:29:27

@Dan Morris

🔥 Filippo Varini

Casey Youngflesh (caseyyoungflesh@gmail.com)

2026-03-18 13:38:05

Hi, folks! We are conducting a brief voluntary survey of the ecology/evolution (and allied fields) research communities to examine how researchers use and perceive Generative AI (GenAI) tools in research practice and would love your participation.

The survey takes < 7 minutes to complete and has been approved by the Clemson University Institutional Review board (IRB2026-0127). Results from this study will provide a clearer picture of current practices and attitudes within our community and will be published and shared in aggregate form. Responses are completely anonymous. Researchers in ecology, evolutionary biology, or allied fields that are at least 18 years of age (any career stage) are eligible.

Survey link: https://clemson.ca1.qualtrics.com/jfe/form/SV_0AIiyOikmB1l8G2

Your perspective is valuable regardless of whether you frequently use GenAI tools or have never used them. We would also greatly appreciate it if you could share this widely. If you have any questions, I would be happy to answer them. Thank you for considering!

clemson.ca1.qualtrics.com

Qualtrics Survey | Qualtrics Experience Management

The most powerful, simple and trusted way to gather experience data. Start your journey to experience management and try a free account today.

Original URL: https://clemson.ca1.qualtrics.com/jfe/form/SV_0AIiyOikmB1l8G2

👍 Carly Batist

Ian McFadden (ian.mcfadden@wsl.ch)

2026-03-19 08:58:31

Hi everyone!

In collaboration with Iran Roman and the QMUL Centre for Multimodal AI, the McFadden Computational Ecology Lab is offering a 3-year, fully-funded PhD studentship at the interface of ecological theory, AI and global biodiversity mapping.

Students should ideally have a strong computer science or mathematical background and be interested in applying their skills to environmental problems, and we welcome applicants from diverse backgrounds and career trajectories.

Please submit your application by the 18th of April using the link below, and if you have any questions please reach out to both of us directly.

https://www.qmul.ac.uk/sbbs/postgraduate/phd-programmes/projects/display-title-1751303-en.html

📣 Oisin Mac Aodha, Viktor Domazetoski, Sara Beery, Xiangteng He, Jason Holmberg (Wild Me), Jennifer, Sarah, Robin Zbinden, Frank Fundel, Timm Haucke, Aleksis Pirinen, Alan Stenhouse

👀 Julia Wilkins, Jason Holmberg (Wild Me), Burak Ekim

🙌 Aleksis Pirinen

Sara Beery (sara.beery@gmail.com)

2026-03-19 09:29:11

Very cool!

🙌 Ian McFadden

Ian McFadden (ian.mcfadden@wsl.ch)

2026-03-19 08:59:38

Image credits: Apple Emoji, Tim Laman, Aidan Hersh, OpenAI: Sora, Carlos de-Oliveira-Nogueira, Duke Lemur Center, Mike Carlo, Netflix: Our Planet, Winfried Wisniewski, Angela Albi

Jennifer (jzhuge@alumni.cmu.edu)

2026-03-20 10:22:12

hey all, does this slack require an invite? I can't remember. Would anyone be able to invite my friend Ebuka, a PhD at Mines looking for internship oppurtunities this summer? ceokwor@mines.edu

Sam Lapp (sam.lapp@pitt.edu)

2026-03-20 11:04:30

nope! you can click “Ai for Conservation” then get an invite link (expires pretty quickly)
Here’s one I just made: https://join.slack.com/t/aiforconservation/shared_invite/zt-3qomqqkz8-Iy89yMq6DqG020Vb6pueKg

Slack

Join AI for Conservation on Slack

Slack is a new way to communicate with your team. It’s faster, better organized, and more secure than email.

Original URL: https://join.slack.com/t/aiforconservation/shared_invite/zt-3qomqqkz8-Iy89yMq6DqG020Vb6pueKg

💯 Jon Van Oast, Jennifer

Jennifer (jzhuge@alumni.cmu.edu)

2026-03-21 11:10:03

thank youuu

Cara Appel (appelc@oregonstate.edu)

2026-03-23 18:57:09

Hi all, I am happy to share a new camera-trap dataset on LILA called Oregon Critters. This dataset contains 99,909 images from sites with paired ground- and trail-facing cameras to detect both small and large animals in western Oregon. We provide bounding-box annotations for 45 taxonomic categories including mammals and birds. Thank you of course to @Dan Morris for his help getting this ready for publication!
https://lila.science/datasets/oregon-critters/

LILA BC

Oregon Critters - LILA BC

Overview This dataset contains 99,909 images from camera traps in western Oregon, USA. These cameras were primarily deployed by the U.S. Forest Service as part of biodiversity monitoring under the Northwest Forest Plan Effectiveness Monitoring Program. Additional data from external projects in western Oregon are also included. These studies were conducted in the H. J....

Written by

lilawp

Est. reading time

3 minutes

Original URL: https://lila.science/datasets/oregon-critters/

🎉 Timm Haucke, Dan Morris, Elizabeth Campolongo, Jon Van Oast, Shir Bar, Seven Song, Sarah, Yunxuan Chai (wsyxbcl), Jose Ruiz-Munoz, Aleena Munshi, Magali Frauendorf, Boe Zhou, Alexander Merdian-Tarko, Owen Martin, Ciara Sypherd, Sara Beery, Declan, Viktor Domazetoski, Sam Lapp, Prabath Gunawardane, Hiroto, Aleksis Pirinen, Arthur Caillau, Hugo, Michael Procko, Miguel Silva

🙌 David Green, Justin Kay, Kaitlyn Gaynor, Dante Wasmuht, Alba Márquez-Rodríguez, Tedi Yankov, Ed Miller, Hiroto, Aleksis Pirinen

👀 Angela Zhu, Hiroto

David Green (davidsethgreen@gmail.com)

2026-03-23 19:20:45

Amazing work @Cara Appel! This is such a great dataset!

Cara Appel (appelc@oregonstate.edu)

2026-03-23 19:23:58

hi @David Green!! thank you!

Arthur Caillau (arthur@caillau.me)

2026-04-21 11:13:34

Amazing work! Thanks a lot!
I am curious how you guys curated your dataset. We have been working on Biowatch to make it as painless as possible to curate large camtrap data. Would love to learn more about your process.

Cara Appel (appelc@oregonstate.edu)

2026-04-21 12:36:15

hi @Arthur Caillau, I'd be happy to answer any questions!

Taylor Gill (taylorgill249@gmail.com)

2026-03-24 13:52:31

Hi all!

My name is Taylor Gill. I am a graduate student in Data Science at the University of Miami, concentrating in remote sensing, with a background in marine biology, coral reef ecology, and environmental data science.

I'm working on a project applying ML-based object detection to satellite, camera trap, or drone imagery for non-invasive wildlife monitoring and I'd love to make it useful to someone here rather than just an academic exercise. The project also includes a GIS component, so spatial analysis and maps of species distribution, habitat use, and detection hotspots are part of the output as well.

If you're working with imagery or occurrence data that has known animal presence and would find it useful to have detections extracted or mapped spatially, I'd love to connect. I'm flexible on species and region, and happy to share all outputs back with you.

Feel free to reply here or DM me.

Thanks,
Taylor

❤️ Ana Tarano, Aleksis Pirinen

Holly (holly@earthspecies.org)

2026-03-26 06:15:29

Hi everyone, Earth Species Project is hiring for a Director of Development! 🚀

We’re looking for someone who can build and lead a fundraising strategy that works across major gifts, foundations, and new funding opportunities.

You’d be joining a small, ambitious team, partnering closely with leadership and researchers, and helping translate complex science into compelling stories to inspire people to support our work.

If you’ve raised at scale, enjoy building relationships from the ground up, and care about nature and science, we’d love to hear from you!

Learn more and apply here: https://earthspecies.org/2026/03/24/were-hiring-director-of-development/

Earth Species Project

We’re Hiring! Director of Development

About Earth Species Project Earth Species Project (ESP) is a non-profit using frontier AI to decode animal communication. We believe the exponential progress we’re seeing in AI offers new ways of l…

Original URL: https://earthspecies.org/2026/03/24/were-hiring-director-of-development/

👏 Ciara Sypherd, Ana Tarano, Aleksis Pirinen

Luca (luca.pegoraro90@gmail.com)

2026-03-30 02:16:18

Hey folks, we just announced the third edition of InsectAI's ACCESS!

Please spread the word, and mind the eligibility criteria 🙂

Luca (https://aiforconservation.slack.com/team/U03AB9UG56F)

📢 ACCESS 2026 Summer School - Applications Open! We're excited to announce the 2026 edition of ACCESS (Aarhus Computational Entomology Summer School) 🦟 🗓️ When: Sep 14-18, 2026 📍 Where: Kaløvigcenter, near Aarhus, Denmark 💰 Funding: Up to 15 fully-funded spots + some self-funded slots ⏰ Deadline: April 30th Perfect for grad students/postdocs in biology or data science who want to work on automated insect monitoring, AI pipelines, and field hardware. You'll build a team project with international mentors and build skills at the cutting edge of biodiversity monitoring in scenic surroundings 🌲 Interested? For more details and how to apply visit the course website 🔗<a href="https://darsa.info/ACCESS-2026/"> </a> Know someone who'd be interested? Please share! 📧

InsectAI ACCESS-2026

Evaluation procedure | InsectAI ACCESS-2026

Eligibility Participation is open to any individual affiliated with a recognised legal entity anywhere in the world (see Section 4.1 of the COST annotated rules). While participation is globally open, eligibility for financial support (i.

Original URL: https://darsa.info/ACCESS-2026/evaluation/

🐛 Kalindi Fonda, Andrea Loya Castillo, Elizabeth Campolongo, Aleksis Pirinen, Ross Gardiner

🐝 Helena Russello, Bernie Boscoe, Aleksis Pirinen

Aleksis Pirinen (aleksis.pirinen@ri.se)

2026-04-07 14:16:52

Cool! We can also share it in climate ai nordics

https://climateainordics.com/

climateainordics.com

Climate AI Nordics

Original URL: https://climateainordics.com/

Aleksis Pirinen (aleksis.pirinen@ri.se)

2026-04-08 01:51:14

also, @Luca, is there some linkedin-post etc one can re-circulate?

Manuela Ramos (mramoso@eafit.edu.co)

2026-03-30 16:49:08

Hi all! 👋
Does anyone have heard about an opportunity for forest conservation bonds and payments for environmental services, for landowners / local communities that protect and conserve forest areas, particularly in the tropics?

Kalindi Fonda (kalindi.fonda@gmail.com)

2026-03-30 19:48:22

This is probably not what you were thinking of, but I found it interesting, basically money to buy land: https://www.iucn.nl/en/news/call-for-proposals-for-ngos-and-csos-protecting-nature-under-threat

IUCN NL

Call for proposals for NGOs and CSOs protecting nature under threat | IUCN NL

The IUCN NL Land Acquisition Fund calls on local conservation NGOs and CSOs worldwide to submit proposals to acquire threatened nature, create safe reserves, and connect habitats for endangered species. The fund supports land purchase, long-term lease of land, and other effective area-based conservation measures (OECMs) in Africa, Asia, Latin…

Original URL: https://www.iucn.nl/en/news/call-for-proposals-for-ngos-and-csos-protecting-nature-under-threat

❤️ Manuela Ramos

Manuela Ramos (mramoso@eafit.edu.co)

2026-03-30 22:17:01

That's exactly what I was thinking of. Thank you very much!! 🎉🤗

🌱 Kalindi Fonda

💪 Kalindi Fonda

Christine Muthee (xtinemuthee@gmail.com)

2026-04-02 01:12:57

Idk if this is rellevant, but I know of a project in Africa called Restore Local where local communities that are actively using their lands for restoration initiatives are financed for the same . https://www.wri.org/initiatives/restore-local

World Resources Institute

Restore Local

Nearly 65% of Africa’s farmland is degraded, by climate change, unsustainable farming practices and deforestation. This drives water scarcity and expands deserts. The livelihoods of the 60% of Africans are at risk who depend on the land for food.Yet across the continent, "restoration champions" are leading a powerful movement—restoring soil, boosting biodiversity, and addressing climate challenges. With growing support, this movement is being ignited on a scale the world has never seen before.

Original URL: https://www.wri.org/initiatives/restore-local

Kalindi Fonda (kalindi.fonda@gmail.com)

2026-04-20 01:08:02

https://www.linkedin.com/posts/global-south-opportunities_forestconservation-climateaction-environmentalfunding-activity-7451613547534618624-iJjn?utm_medium=ios_app&rcm=ACoAAAL-Q7wBf8QWEmhlgs-iJ7Zx167yrcBE7FM&utm_source=social_share_send&utm_campaign=copy_link

Saw this on linkedin today

#forestconservation #climateaction #environmentalfunding #sustainability #biodiversity #conservationgrants #globalsouthopportunities #natureprotection #climatefinance #greenprojects | Global South Opportunities

🌳 Forest Conservation Fund (FCF) 2026 ✅ Eligibility ◦ Minimum 500 hectares of native forest ◦ Legal ownership or management rights required ◦ Must have (or apply for) conservation status ◦ Forest under active deforestation pressure 🌱 Target Threats ◦ Palm oil, cocoa & coffee farming ◦ Timber extraction & mining ◦ Livestock & soy production ◦ Rubber & pulp industries 💡 Funded Activities ◦ Community livelihood development ◦ Anti-poaching & ranger training ◦ Forest restoration projects ◦ Conservation licensing & land protection ◦ Capacity building for local teams 📅 Funding Cycle ◦ Deadlines: June, September, December ◦ Grants: Up to 5 years ◦ First disbursement released after approval 🔗 Learn more: <https://lnkd.in/d7bzRwVD> #ForestConservation #ClimateAction #EnvironmentalFunding #Sustainability #Biodiversity #ConservationGrants #GlobalSouthOpportunities #NatureProtection #ClimateFinance #GreenProjects

Original URL: https://www.linkedin.com/posts/global-south-opportunities_forestconservation-climateaction-environmentalfunding-activity-7451613547534618624-iJjn?utm_medium=ios_app&rcm=ACoAAAL-Q7wBf8QWEmhlgs-iJ7Zx167yrcBE7FM&utm_source=social_share_send&utm_campaign=copy_link

Ben Weinstein (benweinstein2010@gmail.com)

2026-04-04 23:24:42

DeepMeerkat 3.0! Frame detection for video processing. Input long running videos and get frames with biodiversity detected. Two modes, classic background subtraction and running megadetector on every frame. Nice in app review and filtering of frames. Part of this broader conversation about vibe coding, DeepMeerkat 2.0 was made a full 9 years ago! Its been downloaded over 50,000 times, including once while I was making 3.0. I pointed cursor to the code base with the prompt ""https://github.com/bw4sz/DeepMeerkat I am the developer of DeepMeerkat. https://benweinstein.weebly.com/deepmeerkat.html. This software is quite old and in badly need of an update. Review the features and UI and create a new python-based toolset that will use the same ideas and update DeepMeerkat. Installer for Mac, Linux and Windows. The tests should be concise. Propose two modes, the current opencv computer vision background subtraction (though you are welcome to update it to something more modern) and the megadetector https://github.com/agentmorris/MegaDetector that predicts biodiversity images on a per object basis. This DeepMeerkat 3.0 with all the modern tools and support to make it a nice desktop app."

I was able to vibe code a full version update in an hour or so. The first version of MotionMeerkat in 2014 took me 11 months to make as a first year PhD! 3.0 is genuinely better than the last version. I wrote 0 lines of code. Not one. I'm not sure how to feel about this, but part of the broader conversation about what agents are good for. Refactoring code and adding new features for old software feels like a real win for scientific reproducibility. I would never have taken the time to update this. https://bw4sz.github.io/deepmeerkat.html, can be installed

pip install deepmeerkat[ui]
deepmeerkat-gui

I'll keep playing with this. Welcome feedback. Works very well (its basically @Dan Morris megadetector with some add-ins for video review). Looking forward to another decade of Meerkat.

Dr. Ben Weinstein

DeepMeerkat

This is archived! Go to <https://bw4sz.github.io/> for the newest version! Machine learning for video-based biodiversity surveys.

Original URL: https://benweinstein.weebly.com/deepmeerkat.html

bw4sz.github.io

DeepMeerkat 3.0 — Ecological video review

DeepMeerkat 3.0: desktop and CLI tools for camera-trap and ecological video using MegaDetector and classic motion detection.

Original URL: https://bw4sz.github.io/deepmeerkat.html

🙌 Molly Blank, Jennifer, Alexander Merdian-Tarko, Shir Bar, Dan Morris, Sara Beery, Timm Haucke, Jose Ruiz-Munoz, Yseult Hb, Fadel Seydou, Filippo Varini, Joe Nangle, Dasha Simons, Joshua Wilson, Sofía Miñano

🙌:skin_tone_3: Alan Stenhouse

Dan Morris (agentmorris@gmail.com)

2026-04-05 13:18:31

Amazing! Another paradox here is that the closer this type of project gets to requiring only a single prompt, the less necessary it is to actually execute that prompt and make the tool, unless the execution of the prompt requires non-publicly-available resources (private data, big GPUs, etc.). E.g. maybe the next person who wants a DeepMeerkat-like interface just runs your prompt, but adds "and make it orange" or "and it needs to run on an Apple IIe" or "and call it DeepPrairieDog" or whatever.

That said, we're not quite so close to one-shotting everything that I think prompts will entirely replace the need to instantiate and release software, but as you demonstrate here, we're not that far off.

As for me, I've been vibe-coding high-social-impact projects like a Chrome extension to remember playback positions in nfl.com video streams:

https://chromewebstore.google.com/detail/nfl-playback-position-sav/mkfajpdocmiaehhmoglmanbphpeogedo

https://github.com/agentmorris/nfl-playback-automation

What a world!

Ben Weinstein (benweinstein2010@gmail.com)

2026-04-05 15:03:41

@Dan Morris do you have a general sense for what kind of conditions mega detector should not be used for. I mean within the realm of things that people would record using videos, have you seen it used on underwater videos? Should I try to get the fathom net detector connected to it for underwater. I think that would be pretty easy.

Dan Morris (agentmorris@gmail.com)

2026-04-05 15:21:33

MegaDetector is for terrestrial camera trap images/videos with relatively typical camera trap placements, anything outside of that is using it "off label". It will typically not do anything useful on underwater videos or drone videos, or even on downward-facing small animal camera traps.

But for underwater videos where the focal species are fish, consider the Community Fish Detector!

https://github.com/WildHackers/community-fish-detector

Ben Weinstein (benweinstein2010@gmail.com)

2026-04-05 18:51:42

fish model added in 3.1.1

🐟 Dan Morris, Alan Stenhouse

Dan Morris (agentmorris@gmail.com)

2026-04-05 19:08:34

Are you still at 0% code viewed? I.e., you just asked AI to add the CFD?

Ben Weinstein (benweinstein2010@gmail.com)

2026-04-09 19:07:48

Yup I'm full on for this experiment

🙌:skin_tone_3: Alan Stenhouse

Ben Weinstein (benweinstein2010@gmail.com)

2026-04-09 17:11:19

@Dan Morris did you invest anymore time in your 'hero images' idea of finding the most charismatic images in a set for presentations. I'd like to run with this with our very large marine airborne surveys for an upcoming presentation. I looked in the slack archives, but I recall you had gotten a little ways into this.

😍 Jon Van Oast, Kalindi Fonda

Dan Morris (agentmorris@gmail.com)

2026-04-09 18:05:03

I did not invest a lot of time into it, but FWIW, the code is here:

https://github.com/agentmorris/hero-images

Sample results page here:

https://lila.science/public/misc/hero-image-samples/snapshot-serengeti-by_score/

You can see just by model names that this was a few weeks ago, i.e. a zillion LLM-years. There were two parts, neither of which is likely relevant to you, for different reasons...

The first part is where I used MD/SpeciesNet results and some simple heuristics to get from single-digit-millions of images to single-digit-thousands that might be candidates. The models (MD/SN) were specific to camera traps, and the heuristics were specific to the distribution of objects in camera trap images, so this likely wouldn't apply to your airborne surveys. But the general approach to this stage might apply: (1) run a model that does reasonable things, (2) write down some heuristics that you would use as a human (empty images aren't good, multi-species images are fun, just one tiny object in the corner is bad but a single big object in the middle is OK), and (3) let an agent write code to pick candidates. In your case this might be as simple as thresholding for a minimum number of detected objects.

The second part is the part where VLMs rate images, and the question of whether this useful is apropos of our other discussion... the only thing that is even possibly worth re-using is the prompt, and I'm not sure whether a prompt that refers to models from three months ago is still useful (plus the prompt describes some things about what makes a good camera trap image, which is not the same as what makes a good aerial image). But FWIW, the prompt (and some of my responses to follow-up questions) is here:

https://github.com/agentmorris/hero-images/blob/main/prompts.txt

Maybe what it means to share software in 2026 is to send each other prompts and say "take what you can from that, then make it your own, and make it relevant to present-day models".

Ben Weinstein (benweinstein2010@gmail.com)

2026-04-09 23:48:08

i didn't literally follow the code, but I did in spirit and got this awesome pod of humpbacks and common dolphins

🐬 Dan Morris

🌠 Anna Willoughby

Joshua Wilson (joshua.wilson1@uq.edu.au)

2026-04-09 22:31:22

Hi everyone, I’m Joshua Wilson, a researcher at the University of Queensland in Australia, working on the use of drones and computer vision for bird conservation.

I’m excited to share a new dataset now available on LILA: Big Bird. The dataset accompanies our recent publication and includes 4,284 RGB UAV images collected across 11 countries, with 49,990 annotations (41,337 bounding boxes and 8,653 polygons) covering 102 bird species. The most represented species include Australian pelican (5,109 annotations), Chinstrap penguin (5,083), and Chilean flamingo (3,741).

Huge thanks to the researchers who contributed to the dataset (you'll find them listed on the dataset page), and to @Dan Morris for reaching out and helping get this dataset hosted on LILA!

❤️ Jon Van Oast, Jianyang Gu, SHRENIK BORAD, Xiangteng He, Aleena Munshi, Viktor Domazetoski, Sarah, Sonny Burniston, Yseult Hb, Ellie Bowler, Joe Nash, Aleksis Pirinen, Timm Haucke, Mitchell Rogers, Thijs van der Plas, Elizabeth Campolongo, Aamir, Arthur Caillau, Alan Stenhouse

🐦 Dan Morris, Yseult Hb, Timm Haucke, Kalindi Fonda, Yunxuan Chai (wsyxbcl), Arthur Caillau

Ben Weinstein (benweinstein2010@gmail.com)

2026-04-09 23:47:12

hi @Joshua Wilson! I wrapped this release into our next bird detector, it looks great!

Joshua Wilson (joshua.wilson1@uq.edu.au)

2026-04-16 20:04:01

Thanks, Ben! Love your work in this area.

Arthur Caillau (arthur@caillau.me)

2026-04-21 09:02:25

Awesome work! We are also working on a generic bird detector and a platform to let anybody generate insights from their drone flights. Would love to connect you with our team to see how we could leverage your work

Ben Weinstein (benweinstein2010@gmail.com)

2026-04-21 13:11:53

sure, happy be DM or on git repo https://deepforest.readthedocs.io/en/v1.3.3/prebuilt.html#bird-detection new model is just waiting on another manuscrupt for data release, welcome to try out from the PR https://huggingface.co/weecology/deepforest-bird/discussions/5

huggingface.co

weecology/deepforest-bird · Update model weights

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Original URL: https://huggingface.co/weecology/deepforest-bird/discussions/5

Alison Ketz (alison.ketz@gmail.com)

2026-04-10 10:40:06

Hi y'all!
I'm hiring a data scientist to join my team at Funga (https://www.funga.earth/), to support our work analyzing forest productivity and experimental data. This is a rare opportunity for an industry role in ecology. It's a fascinating ecological tech stack, and a fabulous group to work with.

https://job-boards.greenhouse.io/funga/jobs/5155669008

This is an onsite role in Austin, TX, USA. Please send the word!

Funga PBC

Home | Funga PBC

Original URL: https://www.funga.earth/

job-boards.greenhouse.io

Data Scientist

Austin, TX

Original URL: https://job-boards.greenhouse.io/funga/jobs/5155669008

😎 Jon Van Oast, Ciara Sypherd

🍄 Brian Geuther, Aly Mtsumi, Elizabeth Campolongo

Holly (holly@earthspecies.org)

2026-04-13 09:55:39

Earth Species Project wants your feedback! 🧠

We’ve been observing a shift in animal communication research, as large-scale behavioral and bioacoustic datasets meet ongoing advances in machine learning. We’ve been using the term Animal Language Processing (ALP) to describe this convergence. It's an approach that uses frontier AI to study non-human communication at a scale and resolution that wasn’t possible before.

This is a working definition and we’re sharing it as an invitation for the community to help critique, refine, and improve it together.

If you have 2–3 minutes, we’d really value your perspective:

Learn more about ALP

Share feedback directly

Thank you 🙏

Earth Species Project

Animal Language Processing: An AI Convergence In Animal Communication

We are starting to see the study of animal communication shift from narrowly defined, species-specific, and hypothesis-limited analyses toward data-driven discovery at scale. We refer to this conv…

Original URL: https://earthspecies.org/2026/04/02/animal-language-processing-an-ai-convergence-in-animal-communication/

🙌 Robin Zbinden, Julia Wilkins, Trevor Cobb, Mitchell Rogers, Dasha Simons

🥳 Kalindi Fonda

Matt Weldy (matthewjweldy@gmail.com)

2026-04-16 11:31:27

In a meeting this morning, we surfaced a slightly surprising amount of variation in some common terms used in camera trapping and acoustic monitoring and I wanted to take a poll to get a general idea about this communities preference.

We are looking for a term to define the total deployment of camera traps or ARUs during a sampling effort. The sampling effort, could be something like all the ARUs deployed for a week, or all the camera traps deployed in year.

We've used deployment quite a bit in reference to a single ARU, but less so in reference to a total data collection effort.

Here are the terms that were on the table. Please vote with the emoji or suggest other terms in reply.
Thanks!
🪙Round
🏕️Campaign
🎩Phase
🦠Deployment
🔵Collection
🎸Installation
📣Survey
💪Effort
♦️Collection Round
♠️Deployment Round
🚲Monitoring Cycle

🦠 Elly Knight, Jenna Kline, Kevin Kelly, Cara Appel, Pat Lyon, Mahny Barazandehtar, Joe Nash, Sam Lapp, Juan Sebastián Cañas, Abram Fleishman

⛺ Jenna Kline, Ștefan Istrate, Dan Morris, Laura Berman, Yseult Hb, Blair Costelloe, Yves Bas, Angela Zhu, Sara Beery

📣 Kalindi Fonda, Angela Zhu

😵‍💫 Kalindi Fonda

♠️ Miguel Silva

Kalindi Fonda (kalindi.fonda@gmail.com)

2026-04-16 16:51:14

The Department of Conservation in New Zealand, does a survey every year (or two years) at a each of the marine protected areas. each MPA has a Survey and multiple sites. Each combo of survey+site is a deployment, and there is usually one drop per deployment (only if the drop was invalid, we have a repetition)

👍 Matt Weldy, Kaspar Soltero

Sam Lapp (sam.lapp@pitt.edu)

2026-04-16 18:11:15

Kitzes lab does “deployments” of 50-500 ARUs one or multiple time per year, at one “site” (eg a state park), typically one ARU per “point”, where each deployment produces one “dataset” of recordings from a specific time period and set of points. However this use of site is definitely unusually/ambiguous, as is deployment (eg HopLite package uses deployment to refer to a single ARU deployed and retrieved, project to refer to a set of deployments)

Sam Lapp (sam.lapp@pitt.edu)

2026-04-16 18:12:59

Survey sounds to me like point count survey or transect survey which is on a short time and spatial scale (but then again, BBS is a survey too and it has annual nationwide scale)

Sam Lapp (sam.lapp@pitt.edu)

2026-04-16 18:52:16

Mission 🕵️‍♂️ 😆

😂 Sara Beery

Matt Weldy (matthewjweldy@gmail.com)

2026-04-16 18:53:45

I'm honestly a little surprised that campaign and deployment are so close! I wonder if the term preference varies by region or method.

Yves Bas (yves.bas@gmail.com)

2026-04-17 03:11:27

I think campaign refer more often to longer deployments than a week (e.g. year-round) but I prefer it to deployment because deployment refers more often to a single sensor so not exactly what's asked here.

Joe Nash (joe@jna.sh)

2026-04-17 10:45:43

Proposal to start using the collective noun for moths for AudioMoth deployments, I quite fancy putting out an “eclipse” or a “whisper” of ARUs

💯 Matt Weldy, Sam Lapp, Sara Beery

Aamir (muhammad.aamir@cs.ox.ac.uk)

2026-04-20 10:33:45

Hi Everyone,

Calling all ecology researchers working with camera trap imagery!
Researchers at the University of Oxford are developing a new opensource software tool called MorphoCam to support workflows for camera trap distance sampling and morphometric measurements. Help shape the tool by taking part in this short 15 minute survey to share your experience and preferences: MorphoCam: Adding depth to camera trap analyses – Fill in form

Your suggestions and support in this survey is highly appreciated and valued.

Thanks!

😎 Timm Haucke, Sara Beery, Elizabeth Campolongo, Karina Torres

📣 Timm Haucke

Arthur Caillau (arthur@caillau.me)

2026-04-20 10:50:19

Hi everyone 👋 — wanted to share something we've been building that feels right at home in this community: Biowatch, a free and open-source desktop app for wildlife researchers and conservationists working with camera trap data.

The idea is simple: everything runs locally on your machine. Species identification models, image browsing, spatial maps, temporal activity analysis — no uploads, no accounts, no tracking. It's CamtrapDP-compatible for clean import/export with GBIF, and works on Windows, macOS, and Linux. We'd really love feedback from folks here — especially on which models you'd like to see integrated next, and what analyses would actually save you time in the field. Download + docs: https://www.earthtoolsmaker.org/tools/biowatch/

earthtoolsmaker.org

Biowatch | EarthToolsMaker

Open-source desktop app for analyzing camera trap data. Runs 100% locally — your sensitive wildlife data never leaves your computer.

Original URL: https://www.earthtoolsmaker.org/tools/biowatch/

🎉 Jason Holmberg (Wild Me), Sonny Burniston, Stephanie Roilo, Fadel Seydou, Robin Zbinden, Alexander Merdian-Tarko, Sara Beery, Dan Morris, Kalindi Fonda, Aleksis Pirinen, Elizabeth Campolongo, Oliver Stringham, Karina Torres, Yunxuan Chai (wsyxbcl)

👍 Aamir

stephen (leard@stlzoo.org)

2026-04-23 10:21:34

Hi everyone, 👋 I'm Stephen Leard.
I'm a solutions architect and data engineer in the Georgia Tech OMSCS program and I work on data platforms for business intelligence and conservation technology at the Saint Louis Zoo.

I'm currently developing a few things around -
1) generative spatial pathing and raster analytics for protected area management and conservation protection,
2) streamlined deployments for converged FOSS data platforms, and
3) MLOps solutions for more accessible in situ (ish) pose estimation workflows (like, an automated loop from BORIS to DLC to ONNX for RTSP, etc.)
On this last one, @Dan Morris shared this great paper a couple weeks ago and I am still learning from it - https://motion-forecasting.github.io/

I recently completed my instructor training with the Carpentries (https://carpentries.org/), and I think they're cool. I'm hoping to volunteer with them pretty regularly, particularly with workshops for Wildlabs (https://wildlabs.net/en/discussion/carpentries-workshop-interest) and the AZA (https://www.aza.org/).
I also think it would be really worthwhile to develop new lessons with the Carpentries (https://carpentries-incubator.org/) around modern dev-workflows (LLMs, harnesses, spec-driven-development, etc.), and I think this group would be a great source of subject matter expertise and experience for building those resources.

I'm excited and grateful to join this group and looking forward to learning from y'all.

👋 Dan Morris, murilo, Sara Beery, Hammed Akande, Jon Van Oast, Arthur Caillau, Mitchell Rogers, Elizabeth Campolongo, Ben

murilo (murilogustineli@gatech.edu)

2026-04-23 10:49:10

hi Stephen! I'm also in the GT OMSCS program :)

I co-lead DS@GT ARC at Georgia Tech - we're a student run group focusing on matching learning research through competitive challenges. I've been mostly involved with ML research towards biodiversity conservation and ecology (images and audio). Reach out if you'd like to know more about us 😄

DS@GT ARC

DS@GT Applied Research Competitions

Learn about DS@GT Applied Research Competitions, a Georgia Tech student research group focused on machine learning, information retrieval, and competitive research challenges.

Original URL: http://dsgt-arc.org/

🌱 Kalindi Fonda

🙌 stephen

stephen (leard@stlzoo.org)

2026-05-01 10:05:54

Very cool, thanks @murilo!

DS@GT looks awesome, I'm going to check this out further, looking at the CLEF site now 👀 👍

🙌 murilo

Isaac Corley (isaac.a.corley@gmail.com)

2026-04-23 12:23:25

New global agricultural field boundaries available for 2024 and 2025 https://taylorgeospatial.org/agricultural-field-boundaries-mapped-globally-for-the-first-time/

Taylor Geospatial: For the Digital Public Good - We aim to democratize the power of GeoAI through global partnerships while strengthening innovation capacity in the St. Louis region.

Agricultural Field Boundaries, Mapped Globally for the First Time - Taylor Geospatial: For the Digital Public Good

For the first time, every agricultural field on Earth has a boundary on the map. Taylor Geospatial funded and co-developed this work with Microsoft AI for Good Lab — one of the most ambitious GeoAI efforts we know of — because we believe GeoAI should work everywhere, not just in the data-rich regions where labeled

Original URL: https://taylorgeospatial.org/agricultural-field-boundaries-mapped-globally-for-the-first-time/

🙌 Molly Blank, Alexander Merdian-Tarko, Sonny Burniston

😲 Mitchell Rogers

Nora Gourmelon (nora.gourmelon@fau.de)

2026-04-24 11:19:25

Hi everyone, we have a new paper out in IEEE TPAMI:
“Comparison Study: Glacier Calving Front Delineation in Synthetic Aperture Radar Images With Deep Learning”
https://ieeexplore.ieee.org/abstract/document/11488532
If you’re working on deep learning for Earth observation or SAR imagery, this comparison study might be useful for you.

👏 Robin Zbinden, Hiroto, Siyuan Li, Alexander Merdian-Tarko, Aleksis Pirinen

Oisin Mac Aodha (macaodha@caltech.edu)

2026-04-24 11:30:09

Hi everyone! This year the UK AI Research Symposium (UKAIRS) will have a track on "Sustainability & AI". The conference will be held in Edinburgh in Nov 2026, and the deadline for submissions is June 8th.

Relevant topics for the Sustainability & AI track include:
** AI and Sustainable Computing
** AI for Climate
** AI for Energy
** AI for Environment & Nature
** AI for Food Systems
** AI for Sustainable Transportation & Logistics
** AI for Circular Economy
** Edge AI
** Sustainability of AI
** AI for Sustainable Development

Note there are some restrictions on submissions (need to be eligible for UKRI (i.e., UK Research Funding)). You can find more info about the conference on the webpage:
https://www.ukairs.ac.uk/i/CallforSubmissions

👏 Malte Pedersen, Ross Gardiner, Filip Dorm, Tedi Yankov, Mitchell Rogers, Nico Lang, Greg White, Subhransu Maji, Xiangteng He, Aleksis Pirinen, Anthea

Mikel (mikel@earthgenome.org)

2026-04-29 12:03:45

This might be of interest to some folks, Earth Index just rolled over to Open Access:

The new, free Open tier gives anyone no cost full planet-wide access, plus uploads, downloads, and the full feature set. The Full tier adds higher usage quotas, Deep Search, and API access, and is also no cost for most grassroots and high-impact environmental users.
We’re also releasing our most capable Earth Index embeddings yet, covering both 2024 and 2025 imagery.

https://www.linkedin.com/feed/update/urn:li:activity:7454911211932119040

👏 Ted Schmitt, Dylan Van Bramer (she/her), Aleena Munshi

👏:skin_tone_4: Oc

Aleksis Pirinen (aleksis.pirinen@ri.se)

2026-05-05 09:35:22

Hi all!

The registration is open to the 2nd Nordic Workshop on AI for Climate -- to be held in Copenhagen, June 26th! 🙂 Topics include AI for climate mitigation & adaptation, environmental monitoring, conservation, sustainable AI, ...

https://climateainordics.com/events/2026-nordic-workshop

The event is organized by Climate AI Nordics (https://climateainordics.com/)

Hope many of you want to join us in Copenhagen!

Link to post in case you wish to re-circulate: https://www.linkedin.com/posts/climate-ai-nordics_registration-is-now-open-for-the-2026-activity-7450573059104698368-ADFw

Climate AI Nordics

2026 Nordic Workshop on AI for Climate

Registration and pricing

Original URL: https://climateainordics.com/events/2026-nordic-workshop

climateainordics.com

Climate AI Nordics

Original URL: https://climateainordics.com/

:loudspeaker: Registration is now open for the 2026 Nordic Workshop on AI for Climate :bookmark_tabs: Registration: https://lnkd.in/dr9K3nrY :spiral_calendar_pad: Date & location: June 26th, 2026 at University of Copenhagen (Københavns… | Climate AI Nordics

📢 Registration is now open for the 2026 Nordic Workshop on AI for Climate 📑 Registration: <https://lnkd.in/dr9K3nrY> 🗓️ Date & location: June 26th, 2026 at University of Copenhagen (Københavns Universitet)! 💻 Workshop webpage with more info: <https://lnkd.in/daDsmc4R> Additional information on additional speakers & overall schedule coming soon! Workshop partners & sponsors: Pioneer Centre for AI (P1), Climes - Swedish centre for impacts of climate extremes, RISE Research Institutes of Sweden

Original URL: https://www.linkedin.com/posts/climate-ai-nordics_registration-is-now-open-for-the-2026-activity-7450573059104698368-ADFw

🙌 Oisin Mac Aodha, Nico Lang, Malte Pedersen, Jennifer

🎉 Jon Van Oast, Filip Dorm, Thijs van der Plas, stefano puliti

stefano puliti (stefano.puliti@nibio.no)

2026-05-07 03:00:51

Feel like moving to the land of the Vikings :flag_no: to work on an exciting project aimed at developing a EU-wide biodiversity monitoring network?

Then check out this 3-years Postdoc position:

https://www.linkedin.com/posts/stefano-puliti-854b7424_horizoneurope-ai-biodiversity-share-7458045582604058624-a83d?utm_source=share&utm_medium=member_ios&rcm=ACoAAAUn2sABvkN71_GkuJHYSyqwV11nnLUTrBU

#horizoneurope #ai #biodiversity | Stefano Puliti

📢 Postdoc Alert – Deep Learning for Image-Based Biodiversity Surveys 🌍🌲🐛 🌿 🪻🧬🤖 We’re looking for a talented researcher in computer vision and deep learning 🧠 passionate about transforming how biodiversity is monitored across Europe :flag_eu: This 3-year postdoc focuses on developing next-generation AI methods for image-based 📸 biodiversity surveys — moving beyond individual species detection toward a more holistic understanding and quantification of species diversity 🌿, and habitat condition 🪾 across natural ecosystems. As part of this position, you'll get the chance to play with some unique systematic, country-scale image datasets from National monitoring programs. The position is based at NIBIO Norwegian Institute of Bioeconomy Research (Norway) and is part of the new #HorizonEurope Next-BON project, which aims to build a European-wide biodiversity monitoring network using large-scale and systematic ecological datasets and remote sensing 📡 The successful candidate will collaborate closely with leading computer vision researchers, including Jan Dirk Wegner and Damien ROBERT at the EcoVisionUZH Lab at the University of Zurich :flag_ch:, with opportunities for research stays and active international collaboration. Apply 👉 <https://lnkd.in/eb9KTmpu> 🗓️🚨 Application deadline: May 27, 2026 If you’re excited about #AI and want to contribute to shaping #biodiversity monitoring at a continental scale, we’d love to hear from you 🌱 📣 Please share in your networks and help us find the right candidate for this exciting opportunity! Annemarie Bastrup-Birk, Iciar Alberdi, Sabina Burrascano, Mari Myllymäki, Gherardo Chirici, Ruben Valbuena, Mikko Peltoniemi, Gert-Jan Nabuurs

Original URL: https://www.linkedin.com/posts/stefano-puliti-854b7424_horizoneurope-ai-biodiversity-share-7458045582604058624-a83d?utm_source=share&utm_medium=member_ios&rcm=ACoAAAUn2sABvkN71_GkuJHYSyqwV11nnLUTrBU

#horizoneurope #ai #biodiversity | Stefano Puliti

🤩 Yseult Hb, Sara Beery, Robin Zbinden

🙌 Malte Pedersen, Alexander Merdian-Tarko

🙌:skin_tone_3: Alan Stenhouse

Sara Beery (sara.beery@gmail.com)

2026-05-07 09:36:46

We are VERY excited to announce the call for applications has just been released for the 5th CV4Ecology Short Course!! This two-and-a-half-week intensive program trains ecologists and conservation practitioners to develop their own AI tools for their own data. Tell your friends! Apply now!

https://cv4ecology.csail.mit.edu/call_for_applications.html

When: January 13-15, 2027(remote); January 18-29, 2027 (in person)
Where: Smithsonian Conservation Biology Institute
Deadline to apply: June 13, 2026

🙌 Rupa Kurinchi-Vendhan, Shir Bar, Julia Chae, Timm Haucke, Burak Ekim, murilo, Yseult Hb, Sarah, Dan Morris, Jason Holmberg (Wild Me), Srikumar Sastry, Atul Ingle, Eric Snyder, Tiziana Gelmi Candusso, Brian Geuther, Mitchell Rogers, Gustavo Perez, Emily Croasdale, Jose Ruiz-Munoz, Kelly Kapsar, Caleb Robinson, Jacqueline Joya, Navodita Mathur, Ghazi Randhawa, Aleksis Pirinen, Hiroto

💚 Esther Rolf, murilo, Oisin Mac Aodha, Sarah, Jason Holmberg (Wild Me), Alexander Merdian-Tarko, Tanya Berger-Wolf, Tiziana Gelmi Candusso, Irina Tolkova, Mitchell Rogers, Gustavo Perez, Ghazi Randhawa, Aleksis Pirinen, Elena Sierra

🎉 Lukas Picek, murilo, Jason Holmberg (Wild Me), Xiangteng He, Jacqueline Joya, Ghazi Randhawa, Aleksis Pirinen, Elena Sierra

🙌:skin_tone_4: Sarthak Arora

Sara Beery (sara.beery@gmail.com)

2026-05-07 09:41:45

If you want to help spread the word:

https://www.linkedin.com/posts/sara-beery-1ba89166_we-are-very-excited-to-announce-th[…]m=member_desktop&rcm=ACoAAA4DIN4BIW_3ubfcfe3GqvBT-NI--ZBxMzQ

https://bsky.app/profile/did:plc:myaesxyk5su6tom3iil24ygf/post/3mlbdmleats2v

https://x.com/sarameghanbeery/status/2052383337570816062

Call for Applications | Sara Beery

We are VERY excited to announce the call for applications has just been released for the 5th CV4Ecology Short Course!! This two-and-a-half-week intensive program trains ecologists and conservation practitioners to develop their own AI tools for their own data. Tell your friends! Apply now! <https://lnkd.in/e3Pcy2tD> When: January 13-15, 2027(remote); January 18-29, 2027 (in person) Where: Smithsonian Conservation Biology Institute Deadline to apply: June 13, 2026

Original URL: https://www.linkedin.com/posts/sara-beery-1ba89166_we-are-very-excited-to-announce-the-call-share-7458147605382967297-iAy7?utm_source=share&utm_medium=member_desktop&rcm=ACoAAA4DIN4BIW_3ubfcfe3GqvBT-NI--ZBxMzQ

bluesky

Sara Beery (https://bsky.app/profile/sarameghanbeery.bsky.social)

We are VERY excited to announce the call for applications has just been released for the 5th CV4Ecology Short Course!! This two-and-a-half-week intensive program trains ecologists and conservation practitioners to develop their own AI tools for their own data. Tell your friends! Apply now! <https://cv4ecology.csail.mit.edu/call_for_applications.html>

Original URL: https://bsky.app/profile/did:plc:myaesxyk5su6tom3iil24ygf/post/3mlbdmleats2v

X (formerly Twitter)

CV4Ecology 2027 Application InfoSession

We are VERY excited to announce the call for applications has just been released for the 5th CV4Ecology Short Course!! This intensive program trains ecologists and conservation practitioners to develop their own AI tools for their own data. Tell your friends! (link below)

Original URL: https://x.com/sarameghanbeery/status/2052383337570816062

Shir Bar (shirbar@tauex.tau.ac.il)

2026-05-26 09:42:23

If you're interested in applying for CV4Ecology this year but have questions regarding the program or application process join us tomorrow for an Infosession at 11am ET (5pm CEST)
https://mit.zoom.us/j/92530183250

🎉 Timm Haucke, Alba Márquez-Rodríguez, murilo, Sara Beery, Andy Huynh, Aleksis Pirinen

❤️ Timm Haucke, murilo, Christoph Praschl, Sara Beery, Eric Snyder, Yseult Hb, Andy Huynh, Hiroto

😎 Timm Haucke, murilo, Sara Beery, Andy Huynh, Brian Geuther

👀 Ana Tarano

Shir Bar (shirbar@tauex.tau.ac.il)

2026-05-29 14:21:30

Bummed you missed this? Recording available:
https://youtu.be/7WKAcj820X0

YouTube

CV4Ecology (https://www.youtube.com/@cv4ecology)

Original URL: https://youtu.be/7WKAcj820X0

❤️ Sara Beery

J. Miguel Valverde (jmvma@dtu.dk)

2026-05-12 00:59:51

Hi all! I'm co-organizing an ECCV workshop with a Call for Papers that it is probably relevant for a many of you 🙂

🌿 The Computer Vision for Natural Heritage (CVNH) is an ECCV workshop that brings together computer vision researchers, natural heritage digitization experts, and domain scientists to advance methods for analyzing 2D, 3D, and multi-modal imaging of natural history collections and to identify open challenges.

📄 Abstract: Globally, large-scale digitization initiatives are generating massive image datasets from natural history collections. Computer vision is essential for unlocking the scientific value of these data, enabling automated extraction of specimen information and supporting research in biodiversity, ecology, and evolution, including studies of migration and ongoing mass extinction. Despite controlled imaging conditions and rich metadata, automated analysis remains challenging due to complex specimen structures, varying appearances, and handwritten labels. Natural history datasets span 2D images (e.g., photographs, multi-spectral scans), 3D volumetric data (e.g., micro-CT), and multi‑modal inputs (e.g., image-text pairs).

📢 Call for Papers: The covered topics include but are not limited to:

Digitization and mobilisation of specimen data from labels and archive cards

2D and 3D specimen imaging

Robotic vision for automated imaging

Species recognition and AI‑assisted species description

Phenology estimation from collection specimens

Benchmarks, datasets, and evaluation protocols

Vision-based quality control and error detection

Morphological trait extraction using CV

Computational challenges for species recognition (CT-scans of micro‑fossils and handwritten insect labels)

🏆 Kaggle challenges: We are also launching two Kaggle Challenges. The authors of the top-performing submissions will be invited to participate in a paper.

Foram2026 Challenge: Detection and classification of microCT 3D scans of Forameniferas.

SCAT2026: Text recognition and text type identification (e.g., “date”, “locality”) in museum label photographs.

🎤 Speakers: Elizabeth G. Campolongo (Senior Data Scientist for the Imageomics Institute, The Ohio State University (United States)); Emily Baird (Professor at the Stockholm University (Sweden)); Moritz Lürig (Assistant Professor at Bonn University (Germany)); Joakim Bruslund Haurum (Assistant Professor at the University of Southern Denmark (Denmark)).

👥 Organizers: Kim Steenstrup Pedersen (Professor, Natural History Museum Denmark, University of Copenhagen); Anders Bjorholm Dahl (Professor, Technical University of Denmark & QIM); Hans Martin Kjer (Associate Professor, Technical University of Denmark & QIM); Roberta Eleanor Hunt (Postdoctoral Researcher, University of Copenhagen); J. Miguel Valverde (Postdoctoral Researcher, Technical University of Denmark & QIM).

📅 Date: September 8th or 9th, 2026 (TBA)
📍 Location: Malmö (Sweden)
🌐 Website: https://computer-vision-for-natural-heritage.github.io/

🐛 Malte Pedersen, Roberta Hunt, Mitchell Rogers, Helena Russello, Thijs van der Plas, Robin Zbinden

😎 Jon Van Oast

🎉 Roberta Hunt, Hugo, Aleksis Pirinen

❤️ Sara Beery, Julia Chae, Amritha Pallavoor, Elizabeth Campolongo

Ellen Aikens (ellen.aikens@gmail.com)

2026-05-13 12:22:17

Anyone who is interested in pursuing a PhD at the intersection of Ecology, Evolution, and CS, I just wanted to make you aware of a fellowship opportunity through the Simons Foundation: https://www.simonsfoundation.org/grant/simons-graduate-fellowships-in-ecology-and-evolution/

Simons Foundation

Simons Graduate Fellowships in Ecology and Evolution

The purpose of these awards is to provide support for students entering U.S.-based Ph.D. programs with a plan to perform research in ecology and evolution. While we will consider all projects in ecology and evolution, we are particularly interested in interdisciplinary research that jointly considers processes on both ecological and evolutionary timescales, or that links studies of ecology and evolution to other STEM fields, including physics, computer science, mathematics, chemistry and other areas of biology, especially genetics.

Original URL: https://www.simonsfoundation.org/grant/simons-graduate-fellowships-in-ecology-and-evolution/

❤️ Ciara Sypherd

Ellen Aikens (ellen.aikens@gmail.com)

2026-05-13 12:24:38

And if anyone specifically has an interest in working with large GPS collar datasets on mule deer or multi-sensor data streams (high res GPS, ACC, IMU) in golden eagles, I'd be happy to brainstorm about potential proposal ideas...

Riley Knoedler (mknoedler@west-inc.com)

2026-05-15 18:19:23

Info for the 2026 CVPR bird walk, please share!

🌟 Atul Ingle, Levi Cai, Xiangteng He, murilo, Shir Bar, Justin Kay, Jason Holmberg (Wild Me), Antoine Salaün, Arpita C., Kevin Barnard, Mohamed Elhoseiny

🪿 Levi Cai, Xiangteng He, murilo, Shir Bar, Justin Kay, Jason Holmberg (Wild Me), Ben, Arpita C.

🎉 Sara Beery, Justin Kay, Jason Holmberg (Wild Me), Aleksis Pirinen

🦆 Mitchell Rogers, Jason Holmberg (Wild Me), Subhransu Maji, Lukas Picek, Arpita C.

❤️ Ted Schmitt, Sholto West

🙌 Laura Chrobak, Ben, Anthony Hoogs

Sara Beery (sara.beery@gmail.com)

2026-05-15 18:24:43

So excited for this!!!

🐦 Mitchell Rogers, Jianyang Gu, Mohamed Elhoseiny

Sara Beery (sara.beery@gmail.com)

2026-06-04 12:49:09

Happening tomorrow!!! See you there 😎

🐦 Oisin Mac Aodha, murilo

Roberta Hunt (r.hunt@di.ku.dk)

2026-05-19 06:58:04

Dear Community! We are very excited to announce that we just launched the Kaggle Competition "MuseumSCAT2026: Specimen Collection Annotation Task"!

It’s a computer vision challenge for structured recognition of 3500 hand/machine-written labels from the Natural History Museum of Denmark's Dung Beetle Collection.

The competition will be part of the CVNH workshop at ECCV 2026 in Malmö, Sweden in September. It’s a great opportunity to contribute methods that can help make museum collections more accessible for ecological and biodiversity research. We welcome contributions from academics and enthusiasts of all levels and backgrounds.

Competition link: https://www.kaggle.com/competitions/museumscat-specimen-collection-annotation-task
Workshop link: https://computer-vision-for-natural-heritage.github.io/

Please feel free to reach out if you have any questions (here, or on Kaggle, or via e-mail: r.hunt@di.ku.dk). :)

kaggle.com

MuseumSCAT: Specimen Collection Annotation Task@CVNH ECCV26

Help the Natural History Museum of Denmark automate label transcription on this Danish Dung Beetle collection.

Original URL: https://www.kaggle.com/competitions/museumscat-specimen-collection-annotation-task

🙌 Robin Zbinden, Malte Pedersen, Laura Chrobak, J. Miguel Valverde, staru, Ankit Kumar Upadhyay, Aleksis Pirinen, Elizabeth Campolongo

👍 Lukas Picek

😎 Jon Van Oast

🐞 Helena Russello

🙌:skin_tone_3: Alan Stenhouse

💩 Alan Stenhouse

Camille Wardlaw (cwardlaw@mit.edu)

2026-05-19 15:24:00

Might be of interest https://ai4ocean.org/2026workshop.htm

ai4ocean.org

AI4Ocean Workshop 2026

AI4Ocean WORKSHOP July 20-31, 2026, University of Washington. We invite scientists, students & early-career researchers to apply for a two-week workshop. AI4Ocean fosters synergies among experts in ocean dynamics, remote sensing & machine learning. Machine learning is rapidly transforming the landscape of oceanography. Meanwhile, in situ and satellite bservations from missions such as SWOT and PACE – together with high-resolution simulations – are revealing ocean variability at unprecedented spatial and temporal scales. This creates a unique opportunity to rethink how we observe, model, and understand the ocean as a coupled, data-rich system.

Original URL: https://ai4ocean.org/2026workshop.htm

🌊 Levi Cai, Eric Snyder, Shir Bar, Kalindi Fonda, Zee, Aleksis Pirinen

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2026-05-22 06:48:03

Hey guys,
I completely forgot to tell you (maybe some of you already saw it on LinkedIn 😄), but we’ve published our BAMBI UAV wildlife monitoring dataset 🚀🌲🦌

To the best of our knowledge, it’s the first large-scale multimodal, geo-referenced UAV dataset specifically focused on wildlife monitoring in forest environments (with a bias on Austria 😅).
A few quick stats:
• 386 paired RGB + thermal video sequences
• ~50h of flight footage (350+ GB)
• 5,100 annotated tracks
• 1.2M+ interpolated bounding boxes
• 12 species/classes
• RTK-level geo-referenced metadata per frame

The data was collected in forests and forest-adjacent habitats across Austria, so it includes lots of real-world challenges like canopy occlusion, seasonal variation, difficult lighting, etc.

The dataset is already available on Zenodo and can be accessed through our small Python toolkit: https://github.com/bambi-eco/Dataset

Big thanks to all people involved including @Hugo, @Sara Beery, @Michael Ørsted, and @David C. Schedl and many others🙂

And we are going to present it as part of the CV4Animal workshop. So if you are interested, Hugo and I are happy to talk with you!

🎉 Hugo, gyeongsu, Tedi Yankov, Mitchell Rogers, Sara Beery, Dan Morris, Shir Bar, Timm Haucke, Jason Holmberg (Wild Me), Alexander Merdian-Tarko, Robin Zbinden, Aleksis Pirinen, Alan Stenhouse

❤️ Hugo, Sara Beery, Timm Haucke, Josh Seltzer

🦌 Mitchell Rogers, Yseult Hb, Sara Beery, Shir Bar, Timm Haucke, Klára Janoušková, Malte Pedersen, Elizabeth Campolongo

🙌 Robin Sandfort

Zee (zaigham.randhawa@utah.edu)

2026-05-22 15:32:50

How long are the videos? Do you a histogram of the video lengths?

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2026-05-22 15:48:28

I don’t have it at hand (can tell you more when I am back in the office), but I would say typically something between 15-25min. but there are also some short outliers. So on average around 8min

❤️ Zee

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2026-05-23 07:15:06

@Zee ok sorry I had a wrong image in mind 😅 I thought since we mostly tried to max out our flight the videos should be longer. But seems so I was wrong:

👍 Zee

Karina Torres (karina.torres@siu.edu)

2026-05-25 09:41:13

Join us for a virtual, global vibe coding party on June 11, 2026! Register using the sign-up link below.

@Kalindi Fonda, @Filippo Varini, and I have teamed up to host two vibe coding parties, making it easy for anyone around the world to vibe code together!

Come with ideas for a tool, web app, game, or something you'd like to make. Get excited about discussions and networking with others who are interested in AI and vibe coding, particularly in the context of biology and ecology!

Each session is 3 hours. Pick the one that works best for your schedule:

Session 1: Best for Asia-Pacific & the Americas

Asia-Pacific: Thursday, June 11 @ 8:00 AM SGT (Perth/Malaysia) / 10:00 AM AEST (Sydney) / 12:00 PM NZST (Auckland)

Americas: Wednesday, June 10 @ 5:00 PM PDT / 8:00 PM EDT

Session 2: Best for Europe, the Americas & Middle East

Thursday, June 11 @ 8:00 AM PDT / 11:00 AM EDT / 4:00 PM BST (London) / 5:00 PM CEST (Brussels) / 8:30 PM IST (Mumbai)

?? What is VIBE CODING ??
Vibe coding is the use of AI-assisted tools to build apps, tools, games, or code with natural language prompts, letting you focus on the "vibes" while the AI handles the code.

Register and learn more about vibe coding parties here

🥲 Adam Stewart

💃 Kalindi Fonda, Josh Seltzer, Sam Lapp, Elizabeth Campolongo

🎉 Dan Morris, Aleksis Pirinen, Malte Pedersen, Carly Batist

💃:skin_tone_4: Oc

Rupa Kurinchi-Vendhan (rupak272@mit.edu)

2026-05-25 13:24:41

🌱 The 3rd CV4Ecology Workshop is taking place at ECCV 2026 this year! 🌱

CV4Ecology aims to bring together researchers working at the intersection of computer vision, machine learning, ecology, and biodiversity monitoring. We welcome submissions and participation from researchers interested in developing AI methods for real-world ecological and environmental challenges.

Deadlines:
• July 10, 2026 — Archival Track
• August 14, 2026 — Non-archival Track

Check out our website for more details! https://cv4e-workshop.github.io/

👀 Frank Fundel, Hiroto, Andy Huynh, Robin Zbinden, Sara Beery, Elena Sierra

🎉 Shir Bar, Risa Shinoda, Oisin Mac Aodha, Roberta Hunt, Andy Huynh, Thijs van der Plas, Robin Zbinden, Aleksis Pirinen, Sara Beery, Malte Pedersen, Juan Sebastián Cañas, Elizabeth Campolongo, Elena Sierra

❤️ Christoph Praschl, Frank Fundel, Risa Shinoda, Jonathan Sauder, Ankit Kumar Upadhyay, Andy Huynh, Robin Zbinden, Aleksis Pirinen, Malte Pedersen, Devis Tuia, Patrik Reiske, Elena Sierra

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2026-05-25 14:13:00

There are different deadlines on linkedin and on the homepage. 😄 so what is true? 🙂

Rupa Kurinchi-Vendhan (rupak272@mit.edu)

2026-05-25 14:29:51

We're working on updating everything! The final deadlines are July 10th and August 14th.

👍 Christoph Praschl, Hiroto

❤️ Justin Kay

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2026-05-25 14:51:38

Sounds great thanks :)

Yuqian Fu (yuqian.fu@kaust.edu.sa)

2026-05-25 14:57:50

The post in LinkedIn has been updated

👍 Christoph Praschl

Hugo (hma@reallyarobot.com)

2026-05-25 16:55:05

Are there any information about page limits etc. for the different tracks? I tried to find it on the GitHub page without luck

Rupa Kurinchi-Vendhan (rupak272@mit.edu)

2026-05-25 18:39:56

In the current website, we refer to 7 page submissions as short form and 14 page submissions as full length. We’ll update the website to be more precise, but in the meantime:

Proceedings (archival papers): Use ECCV standard formatting. Proceedings papers may be either short-form (7 pages) or full-length (14 pages). References are not counted toward the page limit.

Non-archival papers and abstracts: Submissions may span a range up to full proceedings (archival) length (14 pages) and should follow ECCV standard formatting. References are not counted toward the page limit. Relevant work accepted at other conferences within the previous 6 months may be submitted as non-archival contributions regardless of length and formatting.

✅ Hugo, Sara Beery

Seven Song (qsong2001@gmail.com)

2026-05-26 01:01:34

ECCV standard formatting is not 8 pages : )

Rupa Kurinchi-Vendhan (rupak272@mit.edu)

2026-05-26 01:39:22

Updated!

Moritz Lürig (moritz.luerig@gmail.com)

2026-05-29 08:39:48

call for papers at ECCV - might be of interest for some of you: https://computer-vision-for-natural-heritage.github.io/cfp

CVNH - ECCV Workshop 2026

Call for Papers

Computer Vision Natural Heritage CVNH ECCV Workshop 2026

Original URL: https://computer-vision-for-natural-heritage.github.io/cfp

❤️ Roberta Hunt, Prince Ravi Leow, Elizabeth Campolongo, Zee, Navodita Mathur, Aleksis Pirinen

Moritz Lürig (moritz.luerig@gmail.com)

2026-05-29 12:21:09

(I just realized this has been posted before - my apologies)

✌️ Vincent Lostanlen, Robin Zbinden, Zee, Roberta Hunt

Sal Keith (sally.a.keith@gmail.com)

2026-06-02 06:13:37

Opportunity based at Lancaster University - MARS has an Environment (which includes ECOLOGY🎉🐠🌴🐞) theme!

MARS: Call for Early Career Visiting Fellows 2026–2027. MARS: Mathematics for AI in Real-world Systems is offering three funded visiting fellowships, lasting four weeks each, to PhD students and postdoctoral researchers. Successful candidates will collaborate with MARS staff on either mathematical machine learning and AI methods, or applied mathematics aimed at real-world problems; further information about MARS’ research is available online. Fellowships may be taken up between 1st October 2026 and 31st July 2027, with travel and accommodation expenses reimbursed up to £4,000. The deadline is 17:00 BST on Friday 31st July 2026. Eligibility criteria and application details are available on the MARS visiting fellows page.

Lancaster University

Mathematics for AI in Real-world Systems - Lancaster University

Original URL: https://www.lancaster.ac.uk/mathematics-for-ai-in-real-world-systems/

Lancaster University

Research - Lancaster University

Original URL: https://www.lancaster.ac.uk/mathematics-for-ai-in-real-world-systems/research/

Lancaster University

Visiting Fellows Scheme - Lancaster University

Original URL: https://www.lancaster.ac.uk/mathematics-for-ai-in-real-world-systems/research/visiting-fellows-scheme/

Barney Stannah (barney@aerialconservation.co.uk)

2026-06-02 12:51:24

Hi everyone 👋
Great to be part of the community.
I’m Barney, founder of Aerial Conservation Monitoring. We’re building AI-powered aerial monitoring for conservation, using drone footage to detect, count and map wildlife, livestock, humans and vehicles across large landscapes 🐘🦏📡
The aim is to help protected areas move from expensive, infrequent aerial surveys to faster, lower-carbon and more repeatable monitoring, with outputs that can feed into platforms like EarthRanger and GIS systems.
We completed our first proof of concept in Kenya and are now working on further model development, field validation and deployment with conservation partners.
Really happy to be here and to connect with anyone interested in aerial surveys, AI detection, drones, wildlife monitoring or conservation tech more broadly.
Website: www.aerialconservation.co.uk
LinkedIn: www.linkedin.com/in/brs13
Very happy to chat with anyone who wants to discuss ACM in more detail.

🐘 Jennifer, Dan Morris, Hiroto, Mitchell Rogers, Owen Martin, staru, Daniele Silvestro, Shir Bar, Christoph Praschl, Sara Beery

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2026-06-03 12:42:11

I am always open to chat about drone based monitoring 😄 Maybe we also have workflows that are re-usable for your use case 😉

Holly (holly@earthspecies.org)

2026-06-04 04:47:37

Hi Everyone,

Earth Species Project is hiring a Director of Development to help shape and lead our fundraising efforts as we continue to push the boundaries of interspecies understanding.

You’d be joining a small, ambitious team, partnering closely with leadership and researchers, and helping translate complex science into compelling stories to inspire people to support our work.

If you’ve raised at scale, enjoy building relationships from the ground up, and care about animals and science, we’d love to hear from you! Apply here. 🙏🐋

Mitchell Rogers (mitchell.rogers@auckland.ac.nz)

2026-06-04 16:47:26

Also, if you are at CVPR and would like to have a beer the night before the bird walk, @Leah Brickson has booked some tables at Wynkoop brewing company. We will meet there around 7pm!

All are welcome!

👀 Prince Ravi Leow, murilo, gyeongsu, Zee

❤️ Sara Beery, Zee, Matthias Zuerl

Hugo (hma@reallyarobot.com)

2026-06-04 22:06:07

@Mitchell Rogers are you still there? We arrived a bit late and can't seem to find anyone

Mitchell Rogers (mitchell.rogers@auckland.ac.nz)

2026-06-04 22:12:07

Yes! Upstairs

Hugo (hma@reallyarobot.com)

2026-06-04 22:13:41

We must be blind 😄

staru (arusharmazxx000@gmail.com)

2026-06-04 23:24:32

Hi, I'm currently working on a project related to dog emotion analysis, and I'm a bit confused about the best way to annotate the video data. I'd appreciate any guidance or suggestions.

Zee (zaigham.randhawa@utah.edu)

2026-06-05 00:54:42

https://proceedings.neurips.cc/paper_files/paper/2023/file/8f0d446441a938d9de420a8ab8d7fd36-Paper-Conference.pdf

Unsupervised temporal event segmentation?

staru (arusharmazxx000@gmail.com)

2026-06-05 00:57:54

We will be using some unsupervised/self supervised methods later on for sure but we need some golden truths to get started with

Blair Costelloe (blaircostelloe@gmail.com)

2026-06-05 04:25:26

We use BORIS for manually annotating behavior in videos: https://www.boris.unito.it/

staru (arusharmazxx000@gmail.com)

2026-06-05 04:49:52

Thanks will check this one

Jenna Kline (jennamkline@gmail.com)

2026-06-05 08:56:08

Our group used CVAT to annotate video data for behavior analysis, we have some tip and tricks to get started here: https://imageomics.github.io/kabr-tools/cvat/cvat-guide/

imageomics.github.io

Setup and Usage - KABR Tools Documentation

Tools for working with data for annotating animal behavior

Original URL: https://imageomics.github.io/kabr-tools/cvat/cvat-guide/

staru (arusharmazxx000@gmail.com)

2026-06-05 11:30:48

Thanks this is really helpful

Sara Beery (sara.beery@gmail.com)

2026-06-05 09:00:21

For the bird walk, is anyone in the group walking over from the conference? Are you guys doing ok?

Atul Ingle (ingle@uwalumni.com)

2026-06-05 09:02:31

I'm running just a tad behind; ETA 7:05 at Commons Park.

Nina Grace Baranduin (ngb34@cam.ac.uk)

2026-06-05 09:43:30

Hello, I am looking at trying to use high resolution remote sensing data to find livestock pens (fenced areas livestock is kept at night) in Romania. Using the google map basemap in ArcGIS I can clearly see the sheepfolds. However google earth engine is a different resolution and the livestock pens cant be seen. I’m looking at the whole of Romania, so buying it isn't really an option I don’t think…Does anyone have any advice on where I might be able to get some high resolution images for Romania that I can use to analyse? Looking for anything within the years 2008-2023 if that helps at all.

Sara Beery (sara.beery@gmail.com)

2026-06-05 10:17:33

It was so fun!!

🪶 Shir Bar, Jaycee Choi, Owen Martin, Julia Wilkins, Elizabeth Bondi-Kelly, Risa Shinoda, Don Cosseboom, Jianyang Gu, murilo, Mohamed Elhoseiny, Subhransu Maji, Levi Cai, gyeongsu, Kevin Barnard, Patrik Reiske, Brian Geuther, Alexander Merdian-Tarko, Malte Pedersen

🎉 Riley Knoedler, Risa Shinoda, Don Cosseboom, murilo, Levi Cai, Jose Ruiz-Munoz, Hayley Smith

Sara Beery (sara.beery@gmail.com)

2026-06-05 10:18:07

:yay_frog: Atul Ingle, Shir Bar, Risa Shinoda

Sara Beery (sara.beery@gmail.com)

2026-06-05 10:20:59

Huge thanks to @Riley Knoedler for organizing!!!;

Riley Knoedler (mknoedler@west-inc.com)

2026-06-05 11:09:49

Is it alright to share these photos on other platforms?

Sara Beery (sara.beery@gmail.com)

2026-06-05 15:05:43

Yep, I think so

Riley Knoedler (mknoedler@west-inc.com)

2026-06-05 16:23:33

https://ebird.org/checklist/S352754456 Here's the ebird checklist of what we saw!

ebird.org

eBird Checklist - 5 Jun 2026 - Confluence & Commons Parks - 15 species

Submitted by Nicholas Faraco-Hadlock.

Original URL: https://ebird.org/checklist/S352754456

😎 Sara Beery

Subhransu Maji (smaji@cs.umass.edu)

2026-06-05 12:34:27

Thanks for organizing the bird walk @Riley Knoedler and great seeing you all!

Come by our posters today at 10:45–12:45 in Poster Session 1 + Exhibit Hall (ExHall A–F) to see more birds! 🐦

183: Not All Birds Look The Same — identity-preserving bird generation, led by Aaron Sun
226: RealBirdID — benchmarking bird species ID with MLLMs, led by @Logan Lawrence

😎 Sara Beery, gyeongsu, Mitchell Rogers, Owen Martin

Alex Chan (hoi-hang.chan@uni-konstanz.de)

2026-06-06 09:35:44

Checkout the soundtrack we made for our CVPR paper this year, something we think (and hope!) this community would appreciate 👻

If you are at CVPR today, come by our poster #389 in the lunch poster session!

https://youtu.be/uZdewa2WuOc

YouTube

Alex Chan (https://www.youtube.com/@AlexHHChan212)

"Frame by Frame" official track for CHIRP dataset CVPR 2026

Original URL: https://youtu.be/uZdewa2WuOc

🪶 Patrik Reiske, murilo, Owen Martin, Sara Beery, Robin Zbinden, Elizabeth Campolongo

👍 Xiangteng He, murilo, Owen Martin, Sara Beery, gyeongsu, Ellen Aikens

😂 Mitchell Rogers, Elizabeth Campolongo, Prince Ravi Leow, Laura Berman

Orit Peleg (Orit.Peleg@Colorado.EDU)

2026-06-07 13:30:50

Hi everyone, it was great meeting many of you during the CV4Animals workshop this week! For those I didn’t get a chance to meet: I’m Orit, and my lab at CU Boulder studies collective behavior and communication in animals. One of our main model species is fireflies, so I’m mostly here to lobby for adding a firefly emoji to the channel fireflyfirefly

Excited to be here, and happy to connect with anyone working on computer vision, animal behavior, or collective intelligence!

:firefly: Shir Bar, Owen Martin, Shayan Azhar, Sara Beery, Yseult Hb, Alexander Merdian-Tarko, Sarthak Arora, Elizabeth Campolongo, Viktor Domazetoski, Arthur Caillau, Ellen Aikens, Jaycee Choi, Ankit Kumar Upadhyay, Ben, Prince Ravi Leow

Felix Benjamin Mueller (felixbmuller@gmail.com)

2026-06-07 13:45:27

Thanks for the great CV4Animals workshop at CVPR! In my lab at University of Goettingen, we think that the biggest leverage we have for advancing animal behavior recognition is to combine data from as many diverse application settings as possible to push into the generalization regime and get better models with less labeled data.

At this year’s CVPR, we present a first step in that direction with a large-scale diverse pretraining dataset for primate behavior. If you are at CVPR, stop by our poster PriVi in the afternoon session today 3pm-5:30pm #345!

We are also collecting more data to scale this project further (including labeled data and potentially other orders). If you are interested in collaborating on that, please shoot me a message! Our project page is https://privi.eckerlab.org/

privi.eckerlab.org

PriVi — Towards a General-Purpose Video Model for Primate Behavior in the Wild

PriVi: A large-scale primate-centric video pretraining dataset with 424 hours of curated video for behavioral research. CVPR 2026.

Original URL: https://privi.eckerlab.org/

🐒 Alexander Merdian-Tarko, Elizabeth Campolongo, Dante Wasmuht

👏 Sal Keith, Prince Ravi Leow

👍 Aimee du Luart

Sal Keith (sally.a.keith@gmail.com)

2026-06-08 05:42:36

This is awesome! I'm trying to get funding to do something similar with fish. If we get it, I'll be in touch and perhaps we can work on a cross-taxon version 😉 Thanks for sharing - it is the same approach we are proposing with the self-supervision and then fine-tuning, so it's really great to see it works

👍 Felix Benjamin Mueller

Zee (zaigham.randhawa@utah.edu)

2026-06-08 05:49:31

Can you perhaps post here a breakdown of the video lengths?

Felix Benjamin Mueller (felixbmuller@gmail.com)

2026-06-21 10:38:26

Hi, sorry for the late response! I was on vacation after CVPR. I attached the length breakdown. For training models, we always used 3s video snippets extracted from the raw footage and cropped to primates. Below is the length breakdown of the raw footage.

For datasets marked with **, we only released 3s snippets (both cropped to primates and uncropped) because we have a ton of video material for those and only used a small fraction of it for model training. But we can make it available if there is interest (either to individual researchers or completely open depending on interest). For the datasets without , raw footage is already publicly available, look for the uncropped_.zip files in the dataset release.

For youtube videos, the duration statistics are calculated after cutting videos into individual scenes, which is why they are mostly so short. For YT, we can only publicly release video ids, but if downloading does not work for you, PM me and I can also share them.

Screenshot 2026-06-19 at 12.07.27.png

❤️ Zee

Sal Keith (sally.a.keith@gmail.com)

2026-07-17 06:58:21

Brilliant, thanks!

Simon Elwen (simon.elwen@gmail.com)

2026-06-08 06:32:26

We welcome abstract submission to the 5th African Bioacoustics Community CONFERENCE - 21 – 27 of November 2026

BIOACOUSTICS is a rapidly growing field of research globally, and there is huge potential for bioacoustics research in Africa. Africa hosts an immense diversity of behaviourally unique acoustic species, environments, and increasing development, resulting in noise pollution.

It is with these realities in mind that the African Bioacoustics Community was initiated in March 2018. The ABC was founded to support growth in the field of bioacoustic science on the African continent, striving to provide opportunities for African scientists, students and wildlife workers to learn from and network with experts in the field through international conferences, workshops, and courses.

The 5th African Bioacoustics Community CONFERENCE - 21 – 27 of November 2026, Kenya

The abstract deadline for the 5th African Bioacoustics Community conference, taking place 21 – 27 of November 2026 at Sawela Lodges, Lake Naivasha, Kenya, has been extended until the 15th of June!

Can't present in person? Don't worry, we are hosting another hybrid event, so you can join us online and present your work virtually.

See the African Bioacoustics Community website to submit your abstract, register, and find out all about this year’s event: https://www.africanbioacoustics.org/

Join the African Bioacoustics Community & Co for the Little Bioacoustics Jamii @ Mpala!
Following directly after the conference, the African Bioacoustics Community is also hosting the Little Bioacoustics Jamii Workshop at the Mpala Research Centre, Nanyuki. This is an engaging, participant driven, residential workshop to encourage organic collaboration in bioacoustics science. Serenaded by the savanna soundscape - enjoy a flexible agenda that prioritizes open discussion, networking, scientific and creative exploration of natural sound. From sound recording and equipment, to big data challenges and machine learning methods, we have created a welcoming space for engagement between attendees at all points in their bioacoustics journey.
Workshop dates: 29th November - 4th December 2026
All information & registration:
African Bioacoustics
African Bioacoustics Community

African Bioacoustics

African Bioacoustics Community

Bioacoustics is a rapidly growing research field globally, but research in Africa and by African scientists is lagging. The African Bioacoustics Community, therefore, strives to put Africa first and shine a light on the amazing bioacoustics research happening in Africa.

Original URL: https://www.africanbioacoustics.org/

😆 Adam Stewart, gyeongsu, Vincent Lostanlen

Erin Roche (roch042@gmail.com)

2026-06-08 09:48:16

Hi all, Erin Roche here, wildlife researcher at Idaho Department of Fish and Game. Using cameras to estimate species abundance. Thanks for the invite to this channel!

👋 Timm Haucke, Peter van Lunteren, Tedi Yankov, Devi Ayyagari, Paul, Shir Bar, Dan Morris, Ben Weinstein, Jason Holmberg (Wild Me), Jon Van Oast, Sara Beery, Viktor Domazetoski

🙌 Robin Sandfort

Dan Morris (agentmorris@gmail.com)

2026-06-08 13:49:45

In addition to being a MegaDetector/SpeciesNet power user, I would like to point out that Erin designed the "MegaDetector logo" (as defined by the logo I use on the MegaDetector repo).

🎉 Ben Weinstein

🔥 Kalindi Fonda, Viktor Domazetoski

🤩 Atul Ingle

👀 Robin Sandfort

Tom August (tomaug@ceh.ac.uk)

2026-06-08 12:16:03

Interested in biodiversity monitoring and turning data into action for nature?

We're recruiting a Biodiversity Monitoring Scientist to join the team at UKCEH in Wallingford. You'll work across field ecology, data analysis in R, stakeholder engagement, and national biodiversity monitoring projects, helping generate evidence that informs conservation and land management decisions.

The role also offers the opportunity to work with emerging technologies, including AI methods for acoustic monitoring and computer vision, helping develop new approaches to understanding biodiversity. No prior AI experience is required.

📍 Wallingford, Oxfordshire
📋 Permanent position
💷 £31,942–£33,898 + excellent benefits

This role would suit someone with a background in ecology or environmental science who enjoys combining fieldwork, data analysis, and applied research to support real-world environmental outcomes.

More info: https://ceh.wd3.myworkdayjobs.com/en-US/CEH_Careers/job/Biodiversity-Monitoring-Scientist_JR1566

📣 Oisin Mac Aodha, Sara Beery, Nathan Fox, Juan Sebastián Cañas, Abhi M, Aleksis Pirinen

Sara Beery (sara.beery@gmail.com)

2026-06-08 18:16:35

There is a little less than a week left to apply for the 5th CV4Ecology Short Course!!

I might be biased, but I really believe this program is a one-of-a-kind opportunity for ecologists to DIRECTLY build robust, applied AI expertise that is immediately usable and useful in their own research, for their own projects, on their own data.

Come join us at the Smithsonian's National Zoo and Conservation Biology Institute's Smithsonian-Mason School of Conservation next January!!

The application is lightweight, no excuses not to apply if you're interested :)

Please share to folks you think would benefit from hands-on skill building in AI for ecology!

Learn more at https://cv4ecology.csail.mit.edu/

❤️ Shir Bar, gyeongsu, Digital Naturalism Laboratories (Andrew Quitmeyer), Alba Márquez-Rodríguez, Yseult Hb, Matthias Zuerl, Brett Ford, Subhransu Maji, Irina Tolkova, Robin Zbinden, Hiroto

🐠 Shir Bar

🎉 Jon Van Oast, Dan Morris, Alexander Merdian-Tarko, Malte Pedersen, Kalindi Fonda, Aleksis Pirinen

Pranjal (pranjal.pok4@gmail.com)

2026-06-09 11:00:36

Hello Everyone, I am Pranjal. I am new here. I have recently completed my undergraduate studies in CS from Kathmandu University, Nepal and joined UKALI(ukali.org) as a Research Assistant where my work focuses on Early Warning Systems in the Himalayas for reducing Human-Wildlife Conflict. It's great to meet you all.

👋 Chris Lange, Shir Bar, Jason Holmberg (Wild Me), Dan Morris, Abhi M, Irina Tolkova, Sara Beery, Alexander Merdian-Tarko

Digital Naturalism Laboratories (Andrew Quitmeyer) (andrew.quitmeyer@gmail.com)

2026-06-10 18:09:37

Hi Pranjal! Do you know about the Gathering for Open Science Hardware Gathering that will happen in October in Nepal! https://openhardware.science/gosh-2026-kathmandu-nepal/ Def help share with any local contacts you have there too!

Gathering for Open Science Hardware - Gathering for Open Science Hardware

- Gathering for Open Science Hardware

We are delighted to share that GOSH 2026 will be hosted by our partners at Karkhana Samuha, a long-time GOSH collaborator celebrating its 10th year of reimagining learning through making and experimentation. Join October 4 - 6, 2026 in Kathmandu for the next GOSH gathering Apply for scholarships! Apply Here Scholarship Deadline: April 30, 2026This

Original URL: https://openhardware.science/gosh-2026-kathmandu-nepal/

Nina Grace Baranduin (ngb34@cam.ac.uk)

2026-07-20 10:03:29

Hey Pranjal, my work is on human-wildlife conflict too! I’d be really interested to hear more about your work, although I dont really touch on early warning systems, I focus on predicting hotspots and understanding why the hotspots are where they are.

Rupa Kurinchi-Vendhan (rupak272@mit.edu)

2026-06-09 15:50:49

@Sara Beery and I just released our new preprint, "Finding Needles in the Haystack: Transductive Active Labeling in Ecology"!

We argue that ecological annotation should often be viewed as transductive active labeling: efficiently labeling a fixed pool of collected data, rather than simply collecting enough labels to train a classifier.

This matters most for rare species, behaviors, and other long-tail classes, where the bottleneck is often sampling these data for expert annotation, not classification. We formalize this as a “needles in haystacks” problem and introduce a rarefaction-inspired stopping criterion for deciding when continued labeling is still likely to surface rare-class examples.

Check out the paper on arXiv: https://arxiv.org/abs/2606.03821!

😎 Timm Haucke, Shir Bar, Juan Pierella Karlusich

❤️ Timm Haucke, Riley Knoedler, Cheng Zhang, Malte Pedersen, Shir Bar, Owen Martin, Joe Nash, Julia Wilkins, Irina Tolkova, Burak Ekim, Yunxuan Chai (wsyxbcl), Justin Kay, Viktor Domazetoski, Juan Sebastián Cañas, Cameron Trotter, Tedi Yankov, Elizabeth Campolongo, Hugo, Alan Stenhouse

👀 Jianyang Gu, Oisin Mac Aodha

🙌 Levi Cai, Yuyan Chen, Eric Snyder, Hiroto, Lukas Picek, Malte Pedersen, Shir Bar, Vincent Lostanlen, Tiziana Gelmi Candusso, Owen Martin, Juan Pierella Karlusich, Burak Ekim, Navodita Mathur, Alexander Merdian-Tarko, Justin Kay, Ben McEwen, Ragib Nihal, Robin Zbinden, Aleksis Pirinen, Burooj Ghani, Hugo

🙌:skin_tone_2: Cara Appel

👍 Cheng Zhang, Srikumar Sastry, Justin Kay

🎉 Jenna Kline, Dan Morris, Justin Kay, Amritha Pallavoor

🚀 Leonardo Viotti

Sara Beery (sara.beery@gmail.com)

2026-06-09 15:58:38

TLDR: we might not be doing as badly for rare species as it seems when we use active learning to label new datasets

💯 Vincent Lostanlen, Ben Weinstein, Cameron Trotter, Tedi Yankov

Filippo Varini (fppvrn@gmail.com)

2026-06-10 04:40:11

Congrats Rupa! Very interesting

❤️ Rupa Kurinchi-Vendhan

Burooj Ghani (buroojghani@gmail.com)

2026-06-11 12:02:11

Thanks for formalizing this. Definitely the way to go!

❤️ Rupa Kurinchi-Vendhan, Sara Beery

Austen (austensorochak@gmail.com)

2026-06-09 21:57:09

Hi everyone 👋 I'm Austen

I'm based in Victoria, BC. I work at Ocean Networks Canada — a leading ocean observatory that operates a vast network of sensors and instruments in the Pacific, Atlantic, Arctic and Southern oceans — where I use data analytics, visualization, and code to track and communicate the impact of ocean research infrastructure.

Before that I worked as a software dev building digital tools for oceanography research.

I'm also working on a bioacoustics platform in my spare time.

Looking forward to connecting with folks here, and I've got an exciting community-driven event to share that hopefully some of you will be interested in!

ca.linkedin.com

Austen Sorochak - Ocean Networks Canada | LinkedIn

I bridge environmental science and technology, using data analytics, visualization, and… · Experience: Ocean Networks Canada · Education: Royal Roads University · Location: Victoria · 500+ connections on LinkedIn. View Austen Sorochak’s profile on LinkedIn, a professional community of 1 billion members.

Original URL: https://www.linkedin.com/in/austensorochak/

👋 Shir Bar, Pranjal, Alexander Merdian-Tarko, Sebastien Ouellet, Malte Pedersen, Eric Snyder, Nicole Link, Ankit Kumar Upadhyay, Noah Barrett

👋:skin_tone_2: Cara Appel

👋:skin_tone_3: Alan Stenhouse

Austen (austensorochak@gmail.com)

2026-06-10 10:37:10

Thank you!

Austen (austensorochak@gmail.com)

2026-06-10 10:50:42

🌊 Applications are open for OceanHackWeek Canada 2026
Bamfield Marine Sciences Centre, Vancouver Island, BC
August 24-28, 2026

The first OceanHackWeek ever hosted in Canada. Join us for five days of hands-on tutorials, data exploration, software development, presentations, collaborative hack projects and community networking at a world-class marine research station. Bamfield sits on the wild Pacific coast of Vancouver Island, surrounded by old-growth rainforest, rocky shorelines, and some of the richest marine habitat in the world.

What to expect:

Tutorials on data science tools and workflows

Hands-on project work with open datasets from Ocean Networks Canada, the Canadian Integrated Ocean Observing System (CIOOS), and the BMSC PEARL Lab

Build lasting connections with a diverse international community

Past participants have launched careers, built lasting collaborations, and found new approaches to their research at the intersection of ocean and data science.

This event is co-organized with individuals from institutions including the University of Washington Applied Physics Laboratory, UW eScience Institute, NOAA Fisheries, and Gulf of Maine Research Institute, alongside Canadian partners ASL Environmental Sciences, Ocean Networks Canada, CIOOS Pacific, and the Bamfield Marine Sciences Centre (BMSC).

💰 $550 CAD — covers 6 nights accommodation and all meals at BMSC. Bursaries available — don’t let cost stop you from applying.

⏰ Application deadline: June 15th

🔗 Apply here: oceanhackweek.org

We're also looking for sponsors to help make this event even better and keep costs accessible for participants. If your organization might be interested or you have a lead, feel free to reach out.

🌊 Alexander Merdian-Tarko, Malte Pedersen, Kalindi Fonda, Aleksis Pirinen, Sam Lokanc

😎 Jon Van Oast

Daniela Ruiz (da.ruizl1@uniandes.edu.co)

2026-06-11 15:49:41

Hi everyone! Our biodiversity team at Microsoft AI for Good just published a new paper in Marine Mammal Science: "Adaptive Acoustic Monitoring for Endangered Cook Inlet Beluga Whales in Complex Soundscapes."
We developed an adaptive AI-based passive acoustic monitoring framework that combines deep learning, contrastive audio-language models, and active learning to improve detection of endangered Cook Inlet belugas while simultaneously distinguishing killer whale and humpback whale vocalizations.
The workflow was designed to remain robust across changing recording systems, new acoustic environments, and highly imbalanced datasets—common challenges in long-term conservation monitoring programs.
📄 Check out the paper here: http://dx.doi.org/10.1111/mms.70213 !

🎉 Jason Holmberg (Wild Me), Zee, Keenan Johnson, Ben Weinstein, Gustavo Perez, Owen Martin, Noah Barrett, Ariel Raizman, Malika Nisal Ratnayake, Jose Ruiz-Munoz, Ankit Kumar Upadhyay, Aleksis Pirinen

🐋 Shir Bar, Keenan Johnson, Rupa Kurinchi-Vendhan, Mitchell Rogers, Sara Beery, Owen Martin, Juan Sebastián Cañas, Alexander Merdian-Tarko, Jacqueline Joya, Noah Barrett, Ariel Raizman, Hugo

❤️ Owen Martin, Juan Sebastián Cañas, Hugo, Ankit Kumar Upadhyay

Yuqian Fu (yuqian.fu@kaust.edu.sa)

2026-06-12 01:34:04

Hi everyone,
We are excited to announce that we have just released the FishNet Challenge under the 3rd CV4Ecology Workshop @ ECCV 2026.
The challenge focuses on text-guided open-set fine-grained fish classification. More details can be found in the poster below and on our website:
https://cv4e-workshop.github.io/competition.html
You are very welcome to check it out and participate!

cv4e-workshop.github.io

CV4Ecology 2026 - Competition

Competition page for the Computer Vision for Ecology workshop at ECCV 2026.

Original URL: https://cv4e-workshop.github.io/competition.html

FloraPalooza-2026-flyer.png

🐠 Shir Bar, Elizabeth Campolongo, Alexander Merdian-Tarko, Brian Geuther, Hugo, Aleksis Pirinen

🙌 Rupa Kurinchi-Vendhan, Daniela Ruiz

Ana Maria Quintero (quinteroossa37@gmail.com)

2026-06-12 12:59:55

🌍 The Climate Change AI Virtual Summer School is back (Jul 19-Aug 22, 2026), and registration is now open!

The program consists of five weeks of online lectures and hands-on tutorials providing essential grounding in AI, climate, and their intersection, as well as exposure to a wide range of AI-for-climate topics.

The program is designed for a global audience from across sectors and career stages, including researchers, practitioners, decision-makers, and students across academia, industry, entrepreneurship, civil society, and the public sector. If you are enthusiastic about ways to responsibly and impactfully leverage AI to tackle major climate problems and are looking to learn more about this space, we encourage you to check out the program!

📅 Dates: July 19 – August 22, 2026
📍 Location: Virtual
🔗 Registration and program info: www.climatechange.ai/events/summer_school2026

The full session schedule will be posted closer to the start date. For questions, please contact summerschool+virtual@climatechange.ai

🙌 Daniela Ruiz, Yuyan Chen, Alexander Merdian-Tarko, Elizabeth Campolongo, Aleksis Pirinen, Carly Batist, David Rolnick

👍 Ajay Goverdhan, David Rolnick

Elizabeth Campolongo (e.campolongo479@gmail.com)

2026-06-15 15:59:44

🌱 Growing the future of biodiversity science!
Applications are now open for FloraPalooza 2026—a collaborative workshop bringing together researchers across plant science, biodiversity, ecology, and AI to explore what's possible when we combine plant images with machine learning.
📅 August 17–20, 2026 | Columbus, OH
⏳ Apply by June 25, 2026

📄 Call for Participation + Application details: https://github.com/Imageomics/FloraPalooza-2026/blob/main/CfP.md

🌸 Jenna Kline, Yuyan Chen, Jaycee Choi, Dan Morris, Mitchell Rogers, Ankit Kumar Upadhyay, Yseult Hb, Jianyang Gu, Daniela Ruiz

🥀 Digital Naturalism Laboratories (Andrew Quitmeyer), Ankit Kumar Upadhyay

stefano puliti (stefano.puliti@nibio.no)

2026-06-16 01:27:19

Happy to contribute to the 3D forest computer vision field with SegmentAnyTreev2 our latest tree instance 🌲 🌳 🌴 and semantic (wood/leaf 🪾🌿) segmentation model https://www.linkedin.com/posts/segmentanytreev2-segmentanytree-segmentanytreev2-share-7472189319508656129-Bgnj/?utm_source=share&utm_medium=member_ios&rcm=ACoAAAUn2sABvkN71_GkuJHYSyqwV11nnLUTrBU

#segmentanytreev2 #segmentanytree #segmentanytreev2 #segmentanytreev2 #pointtransformerv3 #opendata #openmodels #forestai #remotesensing #lidar #deeplearning #pointclouds #forestry #digitalforestry… | SmartForest4.0

🌲 Introducing #SegmentAnyTreeV2 (SaTv2) 🤖 After the success of #SegmentAnyTree, we are excited to share #SegmentAnyTreeV2 🔥 — the latest contribution by SmartForest4.0 to build robust, scalable, and operational AI tools for single tree forest inventory and management 🚀 read the preprint 📰 : <https://lnkd.in/ep9Q5nmD> Built to work across sensors, platforms, and forest types, #SegmentAnyTreeV2 expands from well-managed conifer forests to a much broader range of forest conditions, including structurally complex broadleaved, mixed, and tropical forests. 🌍🌳 The model combines a #PointTransformerV3 backbone with a tree-focused instance decoder, enabling stronger separation of individual trees even in dense and complex stands. Key highlights: 📊 Introducing FOR-instance V3 a new and more CHALLENGING benchmark dataset that standardizes several open benchmarks ✅ SegmentAnyTreeV2 achieves SOTA performance --> more trees are correctly segmented ✅ Improved coverage --> better tree masks ✅ Improved semantic --> better leaf/wood separation ✅ Strong zero-shot generalization across unseen forest sites Well done to Maciej Wielgosz, Stefano Puliti, Rasmus Astrup and a huge thanks to everyone involved in pushing 3D forest computer vision with #openData and #OpenModels! #ForestAI #RemoteSensing #LiDAR #DeepLearning #PointClouds #Forestry #DigitalForestry #AI #SegmentAnyTreeV2 #TreeSegmentation

Original URL: https://www.linkedin.com/posts/segmentanytreev2-segmentanytree-segmentanytreev2-share-7472189319508656129-Bgnj/?utm_source=share&utm_medium=member_ios&rcm=ACoAAAUn2sABvkN71_GkuJHYSyqwV11nnLUTrBU

🎉 Cameron Trotter, Dan Morris, Tedi Yankov, Zee, Elizabeth Campolongo, Hiroto, Arpita C., Malte Pedersen, Jose Ruiz-Munoz, Digital Naturalism Laboratories (Andrew Quitmeyer), Aleksis Pirinen, Viktor Domazetoski, Daniela Ruiz, Ben Weinstein

🙌 Isaac Corley, Zee

🌲 Mitchell Rogers, Wenxin Yang

Isaac Corley (isaac.a.corley@gmail.com)

2026-06-16 09:02:21

FOR-instance is clever, great work!

Digital Naturalism Laboratories (Andrew Quitmeyer) (andrew.quitmeyer@gmail.com)

2026-06-17 03:26:13

-Travel Funding apparently available-
“Integrative Social-Ecological Knowledge Systems: Conceptual and Computational Foundations for Ecological Synthesis” is now open for abstract submissions.
The workshop will take place September 28–October 1, 2026 in Jena, Germany, bringing together researchers working across ecology, AI, knowledge graphs, philosophy of science, and social-ecological systems.
We are especially interested in contributions on:
– AI and LLM-based approaches to ecological synthesis
– semantic knowledge graphs and ontologies
– social-ecological knowledge integration
– epistemic standards, validation, and interpretability
– conceptual foundations for computational synthesis
We’re very pleased to now have support from the DFG and Ernst Abbe Stiftung. Travel subsidies are available for active participants, including speakers and poster presenters.
Abstract deadline: June 30, 2026
Details and submission: https://indico.rz.uni-jena.de/event/329/
Please share widely with colleagues and students who may be interested.

Indico

Integrative Social-Ecological Knowledge Systems: Conceptual and Computational Foundations for Ecological Synthesis

Ecological research increasingly relies on the integration of heterogeneous knowledge sources, including field observations, experimental results, theoretical models, and socio-economic data. However, relevant knowledge remains highly fragmented across disciplines, conceptual frameworks, and data infrastructures, which limits the capacity for ecological synthesis. Recent advances in large language models, semantic knowledge graphs, and agent-based AI systems make it technically feasible to...

Original URL: https://indico.rz.uni-jena.de/event/329/?utm_source=chatgpt.com

🌱 Kalindi Fonda

🙌 Keenan Johnson

👀 Keenan Johnson

Larry Clarfeld (Laurence.Clarfeld@uvm.edu)

2026-06-18 16:03:58

Hi everyone,

I'm happy to share a new composite dataset on LILA called "AMMonitor Camera Traps". The dataset includes around 1.2 million photos compiled from 10 datasets collected in the northeastern U.S. states of Vermont, New Hampshire, Maine, and Massachusetts.

I used the original datasets (published on USGS ScienceBase) to train DeepFaune New England, a classification model for northeastern North America. Thanks to @Dan Morris for finding these datasets and making them available through LILA, where I hope they will be useful to others.

And thanks to all the individuals and organizations who collected these data. I've been very fortunate to work with them and they are acknowledged in the individual citations for each dataset.

https://lila.science/datasets/ammonitor-camera-traps/

Cheers,
Larry

GMS-GMS1_071_12-27-2016_22-24.JPG

🦊 Dan Morris, Ankit Kumar Upadhyay, Sara Beery, Alexander Merdian-Tarko, Timm Haucke, Cara Appel, Viktor Domazetoski, Yunxuan Chai (wsyxbcl)

🎉 Jon Van Oast, Ankit Kumar Upadhyay, Daniela Ruiz, Sara Beery, Timm Haucke

Dan Morris (agentmorris@gmail.com)

2026-06-19 12:27:00

Also my obligatory reminder that individual camera trap datasets on LILA are an illusion; most folks training models probably want to use the Very Giant CSV File where all the datasets are mapped to a common taxonomy:

https://lila.science/taxonomy-mapping-for-camera-trap-data-sets/

👍 Larry Clarfeld

Arthur Caillau (arthur@caillau.me)

2026-06-19 04:10:46

Hi Larry,

This is fantastic! Thank you for compiling and sharing it, and for training DeepFaune New England on top of it. Composite datasets like this are exactly what the community needs.

Nice timing too: we just added your dataset to Biowatch in our latest release, so you can now explore the dataset interactively - browse the 1.2M images, filter by species and deployment, and look at activity patterns and species distributions on a map, all offline on your own machine. It imports the LILA datasets directly, so AMMonitor works out of the box.

Biowatch is a free, open-source desktop app for camera-trap data analysis that we build at earthtoolsmaker. Might be a fun way for folks to get a feel for the data before diving in. Happy to help anyone who wants to try it.

- Biowatch: https://www.earthtoolsmaker.org/tools/biowatch/
- Biowatch Manual: https://biowatch.earthtoolsmaker.org
- earthtoolsmaker: https://www.earthtoolsmaker.org

Cheers!

biowatch-ammonitor-camera-traps.png

biowatch-ammonitor-camera-traps-media.png

🎉 Alexander Merdian-Tarko, Larry Clarfeld, Dan Morris, Nuno Sa, Elizabeth Campolongo, Gaspard Dussert

Keenan Johnson (keenan.johnson@gmail.com)

2026-06-20 09:46:41

Anyone here based in London? I've just relocated to london and looking to connect with anyone / find my people here.

❤️ Digital Naturalism Laboratories (Andrew Quitmeyer), Shayan Azhar

Shayan Azhar (shay@wildmon.ai)

2026-06-21 05:28:37

Not London based but am in town until the 25th! Then I usually am up north, closer to Manchester

Keenan Johnson (keenan.johnson@gmail.com)

2026-06-21 08:33:26

Oh great I'll message you !

Avishek Dangol (dangolavishek@gmail.com)

2026-06-21 15:41:04

Hi @Keenan Johnson, I am based in London. Whereabouts in London are you based?

Keenan Johnson (keenan.johnson@gmail.com)

2026-06-21 15:41:15

Also London!

ash (a.h.blum@gmail.com)

2026-06-22 13:01:42

I'll be in town late July/early August, would love to catch up! What part of London are you in?

Avishek Dangol (dangolavishek@gmail.com)

2026-06-22 15:13:21

I'm in Fulham

Satyam Mohla (satyammohla@gmail.com)

2026-06-24 00:55:31

Hi from Oxford!

Keenan Johnson (keenan.johnson@gmail.com)

2026-06-25 09:40:11

Hello @ash, @Avishek Dangol, and @Satyam Mohla! Good to connect with all of you! I had a lovely chat with @Shayan Azhar yesterday and would love to catch up with each of you next time you are around. I'm sorta in central london, near tower bridge area etc, but can easily travel to other parts

👍 ash

Yuqian Fu (yuqian.fu@kaust.edu.sa)

2026-06-22 22:12:38

Dear everyone, our fishnet challenge @ ECCV26 CV4Ecology workshop is still ongoing, you are very welcome to have a look and participate at the competition: https://cv4e-workshop.github.io/competition.html 🙌🙌

cv4e-workshop.github.io

CV4Ecology 2026 - Competition

Competition page for the Computer Vision for Ecology workshop at ECCV 2026.

Original URL: https://cv4e-workshop.github.io/competition.html

Elizabeth Campolongo (e.campolongo479@gmail.com)

2026-06-23 09:20:20

There’s still time to apply!
Submit your application by June 25 to join us for FloraPalooza 2026.

Elizabeth Campolongo (https://aiforconservation.slack.com/team/U0515V75N56)

🌱 Growing the future of biodiversity science! Applications are now open for FloraPalooza 2026—a collaborative workshop bringing together researchers across plant science, biodiversity, ecology, and AI to explore what's possible when we combine plant images with machine learning. 📅 August 17–20, 2026 | Columbus, OH ⏳ Apply by June 25, 2026 📄 Call for Participation + Application details: <a href="https://github.com/Imageomics/FloraPalooza-2026/blob/main/CfP.md">https://github.com/Imageomics/FloraPalooza-2026/blob/main/CfP.md</a>

Digital Naturalism Laboratories (Andrew Quitmeyer) (andrew.quitmeyer@gmail.com)

2026-06-23 13:24:18

I don't know if they posted it here too, but the Imageomics team also has this cool course! It's a whole year of learning about DIY technology for conservation (and you visit like 5 different field sites!). And it's a course you can join from whatever institution i think!
docs.google.com/forms/…/viewform

Elizabeth Campolongo (e.campolongo479@gmail.com)

2026-06-23 14:23:43

They only go to one field site, but yes, it is a virtual course (other than the field component in January). We've had it recognized at different institutions and run as an independent study at others.

IMG_1375

✅ Digital Naturalism Laboratories (Andrew Quitmeyer)

👀 Irina Tolkova

Jennifer (jzhuge@alumni.cmu.edu)

2026-06-23 21:17:24

wow this is so cool thank you for sharing! Do I have to be a student to apply for this?

👀 Cathy Richards

Dan Morris (agentmorris@gmail.com)

2026-06-25 19:30:40

We've been meaning to write up a tutorial about fine-tuning SpeciesNet, I finally got my act together to do this (with help from AI... and a vibe coding party!):

https://agentmorris.github.io/speciesnet-fine-tuning

Thanks to @Peter Bull who ported the SpeciesNet weights to a fine-tuning-friendly container; that's what I used for this tutorial.

This was also a good opportunity to reflect on what a "tutorial" means in 2026. Although I think the tutorial text is readable enough, this was the first time I produced a big chunk of text where I don't really want humans to read it, I want humans to ask agents to read it so the humans can do a thing (see the "consider using AI to take it from here" section).

With that mindset, instead of explaining, e.g., how to install git and Python, and becoming the 77,000,000th tutorial on the Internet that starts with explaining how to install git and Python, which is also impossible because everyone's machine is different, tutorial authors can focus on the ideas you want to communicate that AI wouldn't know by default.

E.g. for a tutorial about fine-tuning SpeciesNet, I'd rather spend more text talking about things that are unique to camera traps - splitting based on cameras, tips for mapping categories, how to decide whether to fine-tune a model in the first place - that AI isn't necessarily aware of out of the box, and that the human user will want to weigh in on, rather than boilerplate setup stuff, or detailed torch/CUDA stuff that AI knows way more about than I do.

🎉 Timm Haucke, Shir Bar, Jon Van Oast, Paul, Yunxuan Chai (wsyxbcl), Christoph Praschl, Jenna Kline, Karina Torres, Cara Appel, Angela Zhu, Aleena Munshi, Viktor Domazetoski, Alexander Merdian-Tarko, Devi Ayyagari, Elizabeth Campolongo, Alan Stenhouse, Carly Batist

👀 Shir Bar, Bernie Boscoe, Malte Pedersen, Sarah

🤯 Paul

❤️ Sara Beery, Malte Pedersen, Otto Brookes, Burak Ekim, Riley Knoedler

Paul (morescopm@gmail.com)

2026-06-25 20:02:26

This is the best news all week, thanks Dan!! And also Peter!

Peter Bull (peter@drivendata.org)

2026-06-25 20:10:42

splitting based on cameras, tips for mapping categories, how to decide whether to fine-tune a model in the first place

Yes! So many folks have asked us for similar takes and just gotten long emails in response. This is so valuable.

Also, I'm digging in to land speciesnet as a base model in zamba and zamba cloud as well for some other options for where/how to do the mechanics of the finetuning, but the decision making criteria here is great.

Casey Clifton (caseyclifton@proton.me)

2026-06-29 04:51:38

Very cool! 🤩

J. Miguel Valverde (jmvma@dtu.dk)

2026-06-30 01:20:33

Hi,

We, the organizers of Computer Vision for Natural Heritage (ECCV 2026 Workshop), have just launched the Kaggle Competition "Forams2026: Detection and Classification of shells of planktonic foraminifera (forams) in volumetric micro-CT images."

The data consists of 2425 labelled 3D volumes of individual forams (training set) and 95 unlabelled 3D volumes contained mixed specimens (test set).

Competition link: kaggle.com/competitions/forams-2026
Workshop link: computer-vision-for-natural-heritage.github.io

Furthermore, we have extended the submission deadline for our Call for Papers: July 14, 2026 (AoE).

kaggle.com

Forams2026

Computer Vision for Natural Heritage competition on detecting Forams in micro-CT volumes

Original URL: https://www.kaggle.com/competitions/forams-2026

CVNH - ECCV Workshop 2026

CVNH

Computer Vision Natural Heritage CVNH ECCV Workshop 2026

Original URL: https://computer-vision-for-natural-heritage.github.io

forams_train.jpeg

🎉 Roberta Hunt

Samuel Repka (Samuel.Repka@lut.fi)

2026-06-30 03:11:04

Hi, I think you should check the Rules section on your Kaggle page. I feel like you left in a lot of LLM instructions in there.

Digital Naturalism Laboratories (Andrew Quitmeyer) (andrew.quitmeyer@gmail.com)

2026-06-30 09:40:22

DEADLINE TODAY: June 30 - Call for talks and posters and abstracts for with some travel funding

"Toward Integrative Social-Ecological Knowledge Systems: Conceptual and Computational Foundations for Ecological Synthesis"
Friedrich Schiller University Jena

indico.rz.uni-jena.de/event/329/abstracts

Indico

Integrative Social-Ecological Knowledge Systems: Conceptual and Computational Foundations for Ecological Synthesis

Original URL: https://indico.rz.uni-jena.de/event/329/abstracts/

Catherine Breen (cbreen@uw.edu)

2026-07-01 11:57:53

Hi All! This year there will be a session on "AI for Snow" ❄️ at the American Geophysical Union meeting in SF. The inspiration for the need for this session is following a bunch of conversations with people in this group and how important snow is for ecosystems! The session is very open for submissions for ML for snow-related challenges that include issues related to conservation (climate impacts, biodiversity impacts related to snow change, water security, and snow drought). If you're doing anything ml+ snow-related, it would be awesome to have you submit! Details below and hope to connect with some of you in person later this year! Feel free to reach out with any questions too.

Session: H099: Machine Learning for Snow: Model Development, Applications, and Lessons Learned
Link: agu.confex.com/agu/agu26/…/279746
due: 5 Aug 2026

AGU - AGU26

Machine Learning for Snow: Model Development, Applications, and Lessons Learned

Machine learning (ML) use cases for snow have rapidly increased over recent years, with applications to snow modeling, snow remote sensing, and data processing. Snow, including seasonal snow, presents a unique use case for ML frameworks as it is variable over multiple spatial and temporal scales, and current sensors and observation networks present limitations due to clouds, forests, and the need for observations in hard-to-reach places. In this session, we welcome published and in-progress work on applications of ML for snow, whether for dataset development, hydrologic modeling, or snow remote sensing. We also welcome experiments that identify model structures and workflows specific to snow datasets, as well as lessons learned for capacity building for ML for snow. Unlike broader cryosphere or hydrology ML sessions, this session focuses specifically on the unique data characteristics and observational constraints of snow, and prioritizes community knowledge-sharing and capacity building alongside scientific results.

Original URL: https://agu.confex.com/agu/agu26/prelim.cgi/Session/279746

Picture1.png

🙌 Justin Kay, murilo, Mark Goldwater, Bernie Boscoe, Viktor Domazetoski

Elizabeth Madin (emadin@hawaii.edu)

2026-07-01 21:20:21

Aloha colleagues,

My lab at the Hawai‘i Institute of Marine Biology is currently advertising a postdoc position for someone with solid quantitative skills plus remote sensing and/or AI/machine learning expertise - the link below has details. If you or anyone you know may be interested and are looking for a postdoc, please feel free to apply or pass on to them.

https://www.oceansphere.org/news/2026/7/1/2-year-postdoc-available-in-ai-and-geospatial-approaches-to-observing-reef-health

We’ll also be advertising very soon for a Project and Lab Manager - this is a Master’s-or-comparable-experience level position for someone who ideally has some geospatial / remote sensing skills. Please email me for details.

Thanks!

Liz Madin

Elizabeth Madin Lab

2-year postdoc available in AI and geospatial approaches to observing reef healtH — Elizabeth Madin Lab

Post-Doctoral Research Fellowship in AI, Remote Sensing & Quantitative Conservation Ecology &nbsp; The Marine Conservation Innovation Group ( <https://www.himb.hawaii.edu/lab/marine-conservation-innovation-group> ) and the Elizabeth Madin Laboratory ( <a href="http://www.oceansphere.org">www.oceansphere.org</a> ) at the U

Original URL: https://www.oceansphere.org/news/2026/7/1/2-year-postdoc-available-in-ai-and-geospatial-approaches-to-observing-reef-health

:fishcowboy: Jaycee Choi, Zee

❤️ Hannah Murray, Zee

Julia Chae (chaenayo@mit.edu)

2026-07-03 18:42:59

Hi everyone, just a reminder that 🌱The 3rd CV4Ecology Workshop 🌱 Archival Track deadline is coming up soon and has been extended to July 15th 2026! Looking forward to another great round of submissions from the AI4ecology community -- please share widely with anyone who might be interested and feel free to reach out for any questions 🙂

Screenshot 2026-07-03 at 6.38.22 PM.png

🎉 Shir Bar, Rupa Kurinchi-Vendhan, Tedi Yankov, gyeongsu, Mitchell Rogers, Burak Ekim, Otto Brookes, Jonathan Sauder, Nina van Tiel, Robin Zbinden, parto, Risa Shinoda, Aleksis Pirinen, Owen Martin, murilo

👍 Cheng Zhang, Rupa Kurinchi-Vendhan, Tedi Yankov, Robin Zbinden, Aleksis Pirinen, murilo

Digital Naturalism Laboratories (Andrew Quitmeyer) (andrew.quitmeyer@gmail.com)

2026-07-07 13:20:45

The latest Widlabs awards have been announced! Congrats to all the groups on another load of neat projects! wildlabs.net/en/article/announcing-wildlabs-awards-2026-grantees

✅ Keenan Johnson, Ben Weinstein

🎉 Jon Van Oast, Yseult Hb, Kalindi Fonda, Alexander Merdian-Tarko, Elizabeth Campolongo

Riley Knoedler (mknoedler@west-inc.com)

2026-07-07 14:23:53

Western EcoSystems Technology, Inc. (WEST) is hiring a US-based Machine Learning Engineer / Data Scientist! Our work focuses on computer vision and developing custom models for NGOs, state and federal agencies, university labs, and private institutions to monitor wildlife. west-inc.hua.hrsmart.com/hr/ats/…/1969

west-inc.hua.hrsmart.com

Deltek Talent Management - JS Job Details

Original URL: https://west-inc.hua.hrsmart.com/hr/ats/Posting/view/1969

❤️ Owen Martin, Angela Zhu, Luke Meyers

Piotr Tynecki (piotr@tynecki.pl)

2026-07-08 04:28:44

Hello everyone.

I am currently working on developing an open-source computer vision model dedicated to detecting seabirds 🐦 and marine mammals from aerial survey imagery ✈️ (offshore context), as well as classifying their behavior (between birds in flight and those sitting on the water).

So far, the project has gained some great momentum. I have successfully collected over 90k high-quality samples covering diverse geographical areas, species and various weather conditions. I also have a strong GPU infrastructure ready for fine-tuning and large-scale experiments. Since high-quality, bbox-like annotated aerial data for marine wildlife is notoriously hard to come by, I am reaching out to our community to connect, share guidance and look for potential co-op. Feel free to talk.

🙂 Tyler Marino

🎉 Jon Van Oast, Alexander Merdian-Tarko

Arthur Caillau (arthur@caillau.me)

2026-07-08 04:35:11

Hey Piotr,
We are building a platform for doing exactly this end to end from drone survey image uploads down to collecting biological insights (number of birds per species, number of nests, dead birds, behavior classification too).
We also have a large data corpus and some models for detection and classification.
Could be interesting to have a chat at some point!

👍 Piotr Tynecki

Piotr Tynecki (piotr@tynecki.pl)

2026-07-08 04:41:56

@Arthur Caillau Thanks for fast response.

As I know that drone surveys have different flight protocols and settings compared to airplane-based surveys. I have a plan to expand into the drone context in the next stage of the project. My current focus right now is strictly on aircraft-based surveys of seabirds and marine mammals.

❤️ Arthur Caillau, Owen Martin

👍 Arthur Caillau

Piotr Tynecki (piotr@tynecki.pl)

2026-07-08 05:11:24

@Kyle Landolt I watched your talk from the State of the Science 2024. Are you still in the business 🙂?

Arthur Caillau (arthur@caillau.me)

2026-07-08 05:52:01

Regarding aerial surveys, we do some of this for seals. You can learn more here: earthtoolsmaker.org/projects/wadden_sea_seal_monitoring

earthtoolsmaker

Wadden Sea Seal Monitoring | earthtoolsmaker

Automated seal population monitoring system using AI to count, classify, and identify individual seals from aerial imagery in the Wadden Sea.

Original URL: https://www.earthtoolsmaker.org/projects/wadden_sea_seal_monitoring/

🙌 Tyler Marino

Dan Morris (agentmorris@gmail.com)

2026-07-08 11:40:38

I track all of the public datasets I'm aware of in this area (drone/aerial detection datasets) here:

https://github.com/agentmorris/drone-wildlife-datasets

Those are not exclusive to marine environments, so many will not be relevant, but hopefully you find some that help you out. Let me know if you know of any that I'm missing.

👍 Piotr Tynecki

Ben Weinstein (benweinstein2010@gmail.com)

2026-07-08 12:25:47

We have a marine animal detector from more than 2 million annotations, including all that @Kyle Landolt and others have worked on. The data isn't ready to share, but I think the model weights would be okay. huggingface.co/weecology/deepforest-marine-biodiversity/discussions/1 I highly suggest you start there.

from deepforest import main
m = main.deepforest()
m.load_model("Weecology/marine-biodiversity-detection")
m.predict_tile(,).

the patch size can be ground resolution specific, so try a range. Then you can fine tune with your specific dataset.

huggingface.co

weecology/deepforest-marine-biodiversity · Add BOEM detection model checkpoint

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Original URL: https://huggingface.co/weecology/deepforest-marine-biodiversity/discussions/1

staru (arusharmazxx000@gmail.com)

2026-07-09 02:58:25

Hii I am working on animal behaviour analysis using SOTA vision algorithms so did a lot of work on segmentation and detection. Would love to collaborate if you planning to publish research in the open

Dan Stowell (dan.stowell@naturalis.nl)

2026-07-08 11:26:46

Delighted to officially publish the results of BioDCASE 2026! Six different tasks in animal sound, each with methods proposed by multiple teams. Dive in to the latest state of the art here: biodcase.github.io/challenge2026

biodcase.github.io

BioDCASE Challenge - BioDCASE

Tasks Results are now available for the 2026 BioDCASE...

Original URL: https://biodcase.github.io/challenge2026/

🦟 Inês Nolasco, Oisin Mac Aodha

🐋 Inês Nolasco, Robin Zbinden

🐦 Inês Nolasco

👍 Holger Klinck, Benjamin Hoffman, Aleksis Pirinen

🎉 Ben McEwen, Jose Ruiz-Munoz, Aleksis Pirinen, Sara Beery, Robin Zbinden

TrevorCaughlin (trevorcaughlin@boisestate.edu)

2026-07-09 12:41:36

We’re hiring a Postdoctoral Research Associate at Boise State University to develop next-generation deep learning methods for identifying plant species from drone imagery. The postdoc will help build open-source AI tools for mapping rangeland plant communities, working with an interdisciplinary team across Boise State, USDA ARS, and the Sevilleta LTER.

The position offers a salary of $78,000 plus benefits, a two-year initial appointment, and the option to be based in Boise, Idaho, or Burns, Oregon. Review of applications begins immediately; interested applicants should email: trevorcaughlin@boisestate.edu with the subject line “Deep learning postdoctoral position.” for more info see:
https://caughlinlab.com/join-the-lab/

❤️ Owen Martin, Sara Beery, Jason Holmberg (Wild Me)

Douglas Mbura (douglas.mbura@gmail.com)

2026-07-11 08:03:22

Join the Climate Change AI Virtual Summer School 2026.

This program provides participants with the opportunity to learn about the responsible, impactful use of AI for climate action. It is designed for a global audience from across sectors and career stages, including researchers, practitioners, decision-makers, and students across academia, industry, entrepreneurship, civil society, and the public sector.

🕑 July 19 to August 22, 2026

🌿 Virtual with synchronous (recorded) lectures and asynchronous interactive coding tutorials

Learn more and register at climatechange.ai/events/summer_school2026

Climate Change AI

Climate Change AI Virtual Summer School 2026

Climate Change AI - Virtual Summer School 2026

Original URL: https://www.climatechange.ai/events/summer_school2026

🎉 Alexander Merdian-Tarko, Kalindi Fonda, Tedi Yankov, Robin Zbinden, Otto Brookes, Carly Batist

👍 Aleksis Pirinen

Brett Buzzanga (brett@conservationta.org)

2026-07-13 12:39:31

If anyone is at the ESRI conference and looking to connect send me a message!

Yuqian Fu (yuqian.fu@kaust.edu.sa)

2026-07-14 15:27:22

Dear everyone, a friendly reminder about the CV4Ecology @ ECCV 2026 Workshop:
The archival track closes on July 15 (AoE). If you are interested in the workshop and have not submitted your paper yet, please remember to do so before the deadline.

The non-archival track will remain open until August 14 (AoE)
🌿 Workshop date: September 8, 2026
🔗 cv4e-workshop.github.io

Thanks for your attention ^-^

cv4e-workshop.github.io

CV4Ecology 2026

Computer Vision for Ecology.

Original URL: https://cv4e-workshop.github.io/

Introducing alp-data: A Shared Data Layer for Animal Language Processing

❤️ Risa Shinoda, Yseult Hb, Oisin Mac Aodha, Rupa Kurinchi-Vendhan, Aleksis Pirinen

Justin Kay (justinkay92@gmail.com)

2026-07-14 15:28:41

Website says July 15 for archival track - is it extended again?

Yuqian Fu (yuqian.fu@kaust.edu.sa)

2026-07-14 15:30:04

it is still 15th July, my bad == have edited the text above

👍 Justin Kay

Yuqian Fu (yuqian.fu@kaust.edu.sa)

2026-07-14 15:30:39

Thanks for noticing it @Justin Kay

👍 Justin Kay

Frank Fundel (frank.fundel@outlook.com)

2026-07-14 17:10:04

and there is no plan to extend the archival deadline? 😅

Yuqian Fu (yuqian.fu@kaust.edu.sa)

2026-07-15 04:16:46

it has been extended once.

✅ Frank Fundel

Christoph Praschl (christoph.praschl@fh-hagenberg.at)

2026-08-05 13:33:12

@Yuqian Fu Is there any info/requirements about poster formats and sizes? 🙂

Yuqian Fu (yuqian.fu@kaust.edu.sa)

2026-08-05 14:31:00

@Rupa Kurinchi-Vendhan do you know any information regarding posters ?

Jan Dirk Wegner (jandirk.wegner@uzh.ch)

2026-07-17 06:24:49

🌲🚀 Join us at the University of Zurich: 3x PhD Positions in AI, Computer Vision & Earth Observation

🌍 The EcoVision Lab at the Department of Mathematical Modeling and Machine Learning (DM3L), led by Prof. Jan Dirk Wegner, and the Remote Sensing of Environmental Change (RSE) group, led by Prof. Livia Piermattei at the Department of Geography, UZH, are recruiting three PhD candidates for the SNSF-funded project "From Alps to Arctic: Satellite-based Assessment of Forest Canopy Height across Decades".

Together, we aim to reconstruct forest change across the Alps and Arctic using modern and historical satellite imagery and state-of-the-art AI, creating the first harmonized, multi-decadal canopy height dataset at scale.

🔬 Research topics include:
• Deep learning for satellite image time-series
• Self-supervised and multimodal learning
• Domain adaptation
• Super-resolution of historical imagery
• Photogrammetry and laser altimetry
• Uncertainty estimation
• Forest changes

🌱 This is a unique opportunity to combine cutting-edge machine learning research with meaningful environmental impact. We welcome applications from candidates with a strong background in Machine Learning, Computer Vision, Data Science, or related fields.

📍 Zurich, Switzerland
📅 Start: January 2027 (or by agreement)
⏳ Application deadline: 16 August 2026

👉 Apply here:

• 2x PhD in Data Science:
jobs.uzh.ch/job-vacancies/…/a704a064-07e0-4920-a0bb-0afe5a2c2b5d

• 1x PhD in Geography:
jobs.uzh.ch/job-vacancies/…/f49ee8f9-3652-4f0e-84ab-94c99e374da8

Please share this opportunity with your network

📣 Oisin Mac Aodha, Nico Lang, Justin Kay, Robin Zbinden, Digital Naturalism Laboratories (Andrew Quitmeyer), Aleksis Pirinen

❤️ Nico Lang, Benjamin Kellenberger, Victor Boussange, Robin Zbinden

Adam Stewart (ajstewart426@gmail.com)

2026-07-17 08:30:29

Feel free to share in the TorchGeo slack as well

Matthias Zuerl (matthias.zuerl@fau.de)

2026-07-20 05:39:29

Hi Jan, any chance one can apply for one of the positions as a PostDoc? 🙂
Cheers, Matthias

Ghazi Randhawa (muhammadghazirandhawa@gmail.com)

2026-07-27 02:53:37

@WasilaRehman

❤️ WasilaRehman

Jan Dirk Wegner (jandirk.wegner@uzh.ch)

2026-08-13 04:54:29

"Hi Jan, any chance one can apply for one of the positions as a PostDoc? 🙂
Cheers, Matthias" ==> no, unfortunately not

Holly (holly@earthspecies.org)

2026-07-20 10:01:59

Earth Species Project is hiring for a Senior Accountant! 🚀

We're looking for someone who can build and strengthen the financial systems that support our mission to decode animal communication with frontier AI.

You'll join a small, ambitious team, partnering closely with Operations and leadership to ensure our accounting, reporting, and financial processes are accurate, scalable, and ready to support the next phase of our growth.

If you're excited by this, we'd love to hear from you. Apply today!

Karina Torres (karina.torres@siu.edu)

2026-07-20 10:26:13

Join us for another virtual, global vibe coding party this week on July 22/23, 2026!

Register here: forms.gle/…

@Kalindi Fonda, @Filippo Varini, and I have teamed up again to host two vibe coding parties, making it easy for anyone around the world to vibe code together! Come with ideas for a tool, website, game, or something you'd like to make. Get excited about discussions and networking with others who are interested in AI and vibe coding, particularly in the context of biology and ecology!

Each session is 3 hours. This month we will only be holding Session 2:

Session 2: Best for Europe, Africa, some of Asia & the Americas
➡️ Thursday, July 23 @ 8:00 AM PDT / 11:00 AM EDT / 4:00 PM BST (London) / 5:00 PM CEST (Brussels) / 8:30 PM IST (Mumbai)

Learn more about vibe coding parties here!

🌱 Kalindi Fonda

Sal Keith (sally.a.keith@gmail.com)

2026-07-21 06:25:30

I will shortly be advertising THREE postdoctoral positions (2 x 3yr, 1 x 2yr) at the intersection of AI + behavioural ecology + conservation based at Lancaster University, UK. We will be using coral reefs as a model system, with later transfer to UK waters.

Keep an eye out here for full adverts - I will post them once they are live. Please do feel free to email / DM me in the mean time for further info 🙂

❤️ Cameron Trotter, Aly Mtsumi, Shir Bar, Matthias Zuerl, Yseult Hb, Dasha Simons, Aimee du Luart

🐟 Kalindi Fonda

🎉 Elizabeth Madin

Elizabeth Madin (emadin@hawaii.edu)

2026-07-21 22:30:33

@Sal Keith: definitely email me the advert link when it's ready! I might have some good applicants to send your way!!

🙌 Sal Keith

Laura Chrobak (lchrobak@mbari.org)

2026-07-21 11:31:42

Has anyone used the test set eval resource, EvALL before? Curious to hear how it might have gone if so!

Holly (holly@earthspecies.org)

2026-07-22 12:11:47

Hi everyone,

I’m excited to share that Earth Species Project has just open-sourced alp-data, the first piece of shared infrastructure for Animal Language Processing (ALP) - an approach that uses AI to study animal communication at scale.

This tool seeks to make it easier for anyone working with bioacoustics to access siloed datasets that live in different repositories and libraries. BirdSet uses one schema, Xeno-Canto another, Watkins yet another. Each comes with its own sampling rates, file structures, metadata conventions, and licensing terms. alp-data replaces all of that data engineering with one consistent interface: every dataset returns audio and sample_rate keys, supports streaming for large corpora, and can be filtered, combined, or reconfigured through YAML.

ESP is offering it to the field as an open-source Python package that gives unified access to 35+ bioacoustic datasets (birds, marine mammals, primates, insects, anurans, and multi-taxon benchmarks) through a single interface. The package is built on the data tools that ESP uses internally to train and evaluate our AI models. Even better, it's open, free, and ready to use today!

Learn more and get started:

- Blog post
- Docs and background
- Seven tutorial notebooks to help you get started
- PyPi project page
- Github Repo

In the meantime, you are invited to explore the tool - including the ALP Data Explorer, which allows you to geographically interact with and preview a curated subset of data. And if you have any prompt feedback, feel free to join the alp-data technical discussions at the ESP’s Discourse forum. Finally, please note that currently dataset contributions to alp-data will be reviewed case by case via this form. Thank you!

Earth Species Project

We are starting to see the study of animal communication shift from narrowly defined, species-specific, and hypothesis-limited analyses toward data-driven discovery at scale. We refer to this conv…

Original URL: http://earthspecies.org/2026/07/22/introducing-alp-data-a-shared-data-layer-for-animal-language-processing

👏 Julia Wilkins, Ariel Raizman, Jessica Kendall-Bar, Dan Morris, Boe Zhou, Sam Lapp, Karina Torres, Tedi Yankov, Naama Marom-Yochai, Sal Keith

🔥 Julia Wilkins, Ariel Raizman, Karina Torres

❤️ Owen Martin, Yossi Hoffman, Karina Torres, melissa.ecologist, Ken Pierce

🎵 Molly Blank

Jessica Kendall-Bar (jessica.kendallbar@gmail.com)

2026-07-22 15:15:01

amazing! i would love to use this in my biological oceanography courses at CSUMB. I'll look through the tutorials, but if anyone would be willing to give me a handheld workshop via Zoom I'd love that! Please send me an email if that would be possible to set up- jkendall-bar@csumb.edu Thank you!

Brittany Solano (brittany@earthspecies.org)

2026-07-23 15:29:19

Hi Jessica! We'll plan a webinar / walkthrough in the coming weeks and make sure you're on the list. In the meantime, if you have questions as you work through the tutorials, start a conversation on the Discourse! earthspeciesproject.discourse.group/c/alp-data/6

Earth Species Project

alp-data

alp-data is the first piece of shared infrastructure for Animal Language Processing (ALP), the approach that uses AI to study animal communication at scale. You can create discussion topics under this forum category to share feedback, contribution ideas, help/feature requests, etc. Please use tags (i.e. “feedback”, “troubleshooting”, “contribution”) to identify your needs.

Original URL: https://earthspeciesproject.discourse.group/c/alp-data/6

✅ Jessica Kendall-Bar

❤️ Owen Martin

Owen Martin (owma6084@colorado.edu)

2026-07-29 14:09:47

Would you be interested in presenting on this at the Wyoming computing symposium in October ? I'm a research scientist there and we have a ton of ongoing work in the animal communication AI realm but many/most researchers there are unaware of this work . I can provide more details privately if you like !

👀 Brittany Solano

Holly (holly@earthspecies.org)

2026-08-03 05:47:04

Hi @Owen Martin, thank you for your comment! We'd be keen to learn more details 🙂 please feel free to DM me!

❤️ Brittany Solano

Josh Seltzer (jyseltz@gmail.com)

2026-07-23 19:04:54

Hi everyone, you're warmly invited to the WILDLABS Group Leads Panel Discussion, an online meetup next Friday where we'll share with you updates from around the world of conservation tech.

We'd love to see you all there and to hopefully start some new conversations and build new connections! Registration link below 👇

luma.com/tte8wdfi?tk=6D5hky

luma.com

WILDLABS Group Leads Panel Discussion · Zoom · Luma

Leads from nine WILDLABS groups - AI for Conservation, Human x Wildlife Coexistence, Bioacoustics, Edge Computing, Open Source Solutions, eDNA & Genomics,…

Original URL: https://luma.com/tte8wdfi?tk=6D5hky

👀 murilo, Digital Naturalism Laboratories (Andrew Quitmeyer), Karina Torres, Ariel Raizman

🌱 Kalindi Fonda, Viktor Domazetoski, Karina Torres

Piero Visconti (visconti@iiasa.ac.at)

2026-07-27 08:10:44

📣 Job alert! Two post-doc positions open in the Biodiversity, Ecology and Conservation Group at the International Institute for Applied Systems Analysis, in Vienna, Austria.

🏛️ The BEC Research Group develops and applies state-of-the-art ecological knowledge, methods, and tools to understand ecosystem functioning, ecological responses, and feedback within coupled social-economic-ecological systems. The Group applies this knowledge to investigate management options that support sustainability goals with ecological implications.

We have recently opened two positions within the group funded by two Horizon Europe international research projects:

Position 1 - Spatial Prioritization for Land-Use Planning
This researcher will apply spatial prioritization methods for integrated land-use planning at national and continental scale. Candidates with strong literacy in spatial multi-criteria optimization methods, reinforcement learning, or other state-of-the-art spatial planning approaches are of particular interest.

Position 2 - Ecological Modeling of Population Dynamics
This researcher will develop statistical and process-based ecological models simulating spatial and temporal colonization and population dynamics, as a function of land management (including restoration and conservation actions), species interactions, and other abiotic effects such as climate.

📍 Laxenburg, Vienna (Austria)
📅 Start: Preferably by October 2026 - duration: 2 years, extendable
⏳ Application deadline: 6 August 2026

👉 Apply here: iiasa.ac.at/employment/job-openings?jh=zdltvx6vbbsljhr1r0d1…

Further information and contacts are available at the link above.

Please share this opportunity with your network.

Thank you!

📣 Robin Zbinden

👏 Matthias Zuerl

❤️ Andrea Zampetti

J. Miguel Valverde (jmvma@dtu.dk)

2026-07-29 04:03:38

ECCV Workshop - Computer Vision for Natural Heritage

We have included a non-archival track to the workshop so that you can submit unfinished work and papers published elsewhere. If you're attending ECCV, this is a great opportunity to present your previous and ongoing research to the community.

Format: 1 page PDF.
Deadline: 23rd of August
Read more: computer-vision-for-natural-heritage.github.io/cfp

CVNH - ECCV Workshop 2026

Call for Papers

Computer Vision Natural Heritage CVNH ECCV Workshop 2026

Original URL: https://computer-vision-for-natural-heritage.github.io/cfp

🙌 Nico Lang, murilo, Malte Pedersen

Jennifer (jzhuge@alumni.cmu.edu)

2026-09-03 16:40:57

hi! could you use any volunteers? I'm a conservation AI engineer, but I'm not registered for ECCV unfortunately, but I'd love to help out with anything

Diego Marcos (diego.marcos.gonzalez@gmail.com)

2026-07-31 04:07:02

We are pleased to invite submissions to the Machine Vision for Earth Observation, Environmental Monitoring, and Climate Change (MVEO) Workshop, to be held as a full-day workshop at BMVC 2026 in Lancaster, UK, on November 26th.
MVEO aims to bring together researchers, practitioners, policymakers, and industry experts working at the intersection of Computer Vision, Earth Observation, and Environmental Monitoring. The workshop will explore how machine vision and AI can support climate action, sustainability, environmental resilience, and evidence-based decision-making.
We welcome original research contributions, position papers, datasets, benchmarks, and application-focused studies spanning topics such as:

Foundation Models and Geospatial AI

Vision-Language and Multimodal Learning

Remote Sensing and Earth Observation Analytics

Environmental Monitoring and Climate Informatics

Biodiversity and Ecosystem Assessment

Disaster Response and Resilience

Precision Agriculture and Sustainable Cities

Climate Risk Prediction and Adaptation

Trustworthy, Explainable, and Green AI

Paper Submission Deadline: 31 August 2026

🔗 Workshop Website: mveo.github.io

mveo.github.io

Workshop on Machine Vision for Earth Observation, Environmental Monitoring, and Climate Change

Workshop on Machine Vision for Earth Observation and Environment Monitoring

Original URL: https://mveo.github.io/

📣 Oisin Mac Aodha, Nico Lang, murilo, Malte Pedersen

Kazi S. Mehrab (ksmehrab@vt.edu)

2026-07-31 17:15:04

Wanted to share this dataset in case it's helpful: https://lnkd.in/p/gNmM23-G

#computervision #machinelearning #ai #datacuration | Jimmy Guerrero

From open water to museum drawer: our new post explores Fish-Vista—60,000 museum fish specimens across ~1,900 species—inside FiftyOne - <https://lnkd.in/g68WgZBJ> • Pixel-level trait segmentation of 9 anatomical parts (eye, head, barbel, and six fins) • Multi-label trait tags read straight from the masks—filter "has a barbel" with zero extra annotation • A long tail of ~1,900 imbalanced species you can actually click into, singletons and all • Embeddings that cluster by body plan, so taxonomy emerges visually • The twist: a wild-trained detector run cold on the specimens found the fish in 100% of images at ~0.97 mAP Same tool, new modality—and one genuinely surprising result at the end. Shout out to Kazi Sajeed Mehrab on the Fish-Vista research #ComputerVision #MachineLearning #AI #DataCuration

Original URL: https://lnkd.in/p/gNmM23-G

🙌 murilo, Yseult Hb

👍 Ken Pierce

murilo (murilogustineli@gatech.edu)

2026-07-31 18:31:49

hey 👋 I work at Voxel51 and do research in biodiversity conservation as well. feel free to reach out if you have any questions about using fiftyone, we do have an open source version of our software :)

❤️ Kazi S. Mehrab, Mitchell Rogers

Yuqian Fu (yuqian.fu@kaust.edu.sa)

2026-08-01 10:14:48

📢 Reviewer Recruitment — CV4Ecology @ ECCV 2026
We are recruiting reviewers for the 3rd Workshop on Computer Vision for Ecology (CV4Ecology), taking place at ECCV 2026 on September 8 in Malmö, Sweden.
We welcome researchers working in computer vision, machine learning, ecology, biodiversity, conservation, and related areas to join our reviewer pool and help support high-quality interdisciplinary research.
📝 Reviewer sign-up form:
forms.gle/…
🌐 Workshop website:
cv4e-workshop.github.io
Please feel free to share this with colleagues and students who may be interested. Thank you for helping us spread the word!

Google Docs

Reviewer Nomination for CV4Ecology @ECCV 2026

3rd Workshop on Computer Vision for Ecology at ECCV The review period for the non-archival round will be August 14–20, 2026

Original URL: https://forms.gle/DPNboDfpjn45rS9b6

cv4e-workshop.github.io

CV4Ecology 2026

Computer Vision for Ecology.

Original URL: https://cv4e-workshop.github.io/

❤️ Mitchell Rogers, Andy Huynh, Sara Beery, Aleksis Pirinen

Jennifer (jzhuge@alumni.cmu.edu)

2026-08-02 19:12:36

Hi! Do you need any volunteer help?

Yuqian Fu (yuqian.fu@kaust.edu.sa)

2026-08-03 02:05:14

If you can sign up as a volunteer reviewer would be great 😊

Jennifer (jzhuge@alumni.cmu.edu)

2026-08-03 10:14:59

I'm not a researcher (yet), I'm a Conservation AI Engineer, so I'm not sure I can review, but I'm happy to be there in person and help out with logistics :)

Yuqian Fu (yuqian.fu@kaust.edu.sa)

2026-08-03 12:05:54

that is so sweet; looking forward to seeing you in person

Hugo (hma@reallyarobot.com)

2026-08-07 05:04:34

Hey @Yuqian Fu - do you know the poster format for the workshop? I can't find the information on the website

Arya Hanif (aryasenaria@gmail.com)

2026-08-04 05:02:53

Hello, I'm Arya and I've been part of a team called SeaLens. We are developing a tool to help scientists from coral restoration groups to do fish monitoring by utilizing computer vision to quickly detect, count and specify fish. We started from a restoration site in Bali and are testing the app right now, evaluating if it properly support scientists in their restoration process. The goal is to help scientists with faster decision making for their next restoration steps and to get biodiversity data and reports more easily.

Please feel free to try out the app here: sealens.app. Our current challenge is the lack of indonesian underwater fish images available, so we are actively processing videos in that region to gather more images and use it for further improvement of our dataset.

If you are part of or in contact with a coral restoration organization, we would be happy to connect

Welcome to SORCE - SORCE - Sustainable Oceanic Research, Conservation and Education

🐟 Cameron Trotter, Sebastien Ouellet, Robin Zbinden, Jennifer, Digital Naturalism Laboratories (Andrew Quitmeyer), Jose Ruiz-Munoz, Arya Hanif, Alexander Merdian-Tarko, Andy Huynh

😎 Jon Van Oast, Andy Huynh

Digital Naturalism Laboratories (Andrew Quitmeyer) (andrew.quitmeyer@gmail.com)

2026-08-04 17:05:18

Which restoration site in Bali did you work with? Sea communities?

Kalindi Fonda (kalindi.fonda@gmail.com)

2026-08-04 17:24:39

Hello @Arya Hanif really interesting, I am working on Spyfish (accelerating marine monitoring using ML, Citizen Science, Experts), which has a very similar goal to yours, and would love to connect/learn where your project is at.

Arya Hanif (aryasenaria@gmail.com)

2026-08-04 23:47:01

@Digital Naturalism Laboratories (Andrew Quitmeyer) We started with livingseas, and now in contact with other restoration sites as well like sorce and reefcheck. Are you maybe familiar with reef restoration project that would be interested on trying it out?

livingseasfoundation.org

Livingseas Foundation

Restoring, Conserving & Rebuilding Coastal and Marine Ecosystems. Based in Indonesia

Original URL: https://www.livingseasfoundation.org/

SORCE - Sustainable Oceanic Research, Conservation and Education

SORCE - Sustainable Oceanic Research, Conservation and Education -

Original URL: https://sorce.org/

Reef Check Malaysia

Home - Reef Check Malaysia

Sustainably managed marine ecosystems in Malaysia starts here

Written by

admin

Time to read

16 minutes

Original URL: https://reefcheck.org.my/

Arya Hanif (aryasenaria@gmail.com)

2026-08-04 23:49:01

@Kalindi Fonda Heii, I would really love to. Where would you like to discuss it more? I've also send a linkedin connect request to you. Thankyou for responding.

🐟 Kalindi Fonda, Arya Hanif

Kalindi Fonda (kalindi.fonda@gmail.com)

2026-08-04 23:58:01

Awesome, I messaged you. 🪸

🙌 Arya Hanif

Digital Naturalism Laboratories (Andrew Quitmeyer) (andrew.quitmeyer@gmail.com)

2026-08-07 07:11:02

Here's a really wonderful group that we worked with last year in bali, in case you won another awesome group seacommunities.com/

Sea Communities

Conservation & Community-Based Tourism | Sea Communities

Explore Sea Communities' tourism programs on coral restoration, volunteer programs, community service, CCAS, and educational tours.

Original URL: http://seacommunities.com/

👀 Arya Hanif

Arya Hanif (aryasenaria@gmail.com)

2026-08-08 01:25:13

@Digital Naturalism Laboratories (Andrew Quitmeyer) Do you mind if I dm you a bit regarding the collaboration that you did with them?

Digital Naturalism Laboratories (Andrew Quitmeyer) (andrew.quitmeyer@gmail.com)

2026-08-08 09:38:32

Sounds good

Aleksis Pirinen (aleksis.pirinen@ri.se)

2026-08-07 00:36:00

🧭1 day left to submit to the 2nd Workshop on AI for Climate and Conservation (AICC-2) that we're organizing at ECCV! 🦏🌏

https://sites.google.com/g.harvard.edu/aicc2eccv26/

sites.google.com

AICC @ ECCV

News IMPORTANT UPDATE: ECCV registration deadline is Aug 10th, so please register for ECCV before that!(Registration: <https://eccv.ecva.net/Conferences/2026/Registration> ) Call for Participation is now open! Submission page: <https://cmt3.research.microsoft.com/AICCECCV2026>. Deadline: Aug 7th. The

Original URL: https://sites.google.com/g.harvard.edu/aicc2eccv26/

🔥 Hugo, Cameron Trotter, Robin Zbinden, Andy Huynh

🌎 murilo

Jennifer (jzhuge@alumni.cmu.edu)

2026-09-03 16:40:32

hi! could you use any volunteers? I'm a conservation AI engineer, but I'm not registered for ECCV unfortunately, but I'd love to help out with anything

Aleksis Pirinen (aleksis.pirinen@ri.se)

2026-09-04 02:40:40

Hi Jennifer! That's very kind of you to ask 🙂

I'd say we're quite set for the workshop now, and since you are not attending ECCV, I don't think there's much more to do currently. But again, appreciate you suggesting this!

Cheers!

Jennifer (jzhuge@alumni.cmu.edu)

2026-09-04 05:05:54

No worries thanks for the response :)

❤️ Aleksis Pirinen

Karina Torres (karina.torres@siu.edu)

2026-08-11 10:23:57

Hey friends! Join us for a virtual, global vibe coding party on August 25/26, 2026! Register here: Sign up for the August Vibe Coding Party

@Kalindi Fonda, @Filippo Varini and I have teamed up to host two vibe coding parties, making it easy for anyone around the world to vibe code together! Come with ideas for a tool, website, game, or something you'd like to make. Get excited about discussions and networking with others who are interested in AI and vibe coding, particularly in the context of biology and ecology! And if you don’t feel like vibe coding, come along for a coworking session.

Each session is 3 hours. Pick the one that works best for your schedule:

Session 1: Best for Asia-Pacific & the Americas
➡️ Asia-Pacific: Wednesday, August 26 @ 8:00 AM SGT (Perth/Malaysia) / 10:00 AM AEST (Sydney) / 12:00 PM NZST (Auckland)
➡️ Americas: Tuesday, August 25 @ 5:00 PM PDT / 8:00 PM EDT

Session 2: Best for Europe, Africa, some of Asia & the Americas
➡️ Wednesday, August 26 @ 8:00 AM PDT / 11:00 AM EDT / 4:00 PM BST (London) / 5:00 PM CEST (Brussels) / 8:30 PM IST (Mumbai)

🤔 What is VIBE CODING?
Vibe coding is the use of AI-assisted tools to build apps, tools, games, or code with natural language prompts, letting you focus on the "vibes" while the AI handles the code.

✏️ Learn more about vibe coding parties here: August Vibe Coding Party Info Doc

⭐Check out projects created during our monthly parties at our Vibe Coding Party Gallery website!

🎉 Malte Pedersen, Dan Morris, Kalindi Fonda, Sara Beery, Dan Voyce, John Quinn

Dan Morris (agentmorris@gmail.com)

2026-08-11 19:26:01

I've gotten a bee in my bonnet lately about small animal cameras (camera traps pointed at the ground, typically inside small enclosures, to find Very Small Things). For purposes of this Slack, I consider this a different modality than "vanilla" camera traps, because (a) these cameras target taxa you would basically never see in standard camera traps (e.g. small amphibians), and (b) none of the AI tools we use for typical camera traps are relevant for small animal cameras. MegaDetector, for example, is hit or miss at best on these cameras (but also unnecessary, because all the animals of interest are actually giant in terms of pixel size, so there's much less reason to take a two-stage, detect-then-classify approach).

So, I trained a classifier for this type of image, primarily to help colleagues at California F&W, who have a mountain of these images for which they weren't getting any help from AI:

https://github.com/agentmorris/small-animal-classifier

So far so good for CDFW, and with help from @Nathaniel Rindlaub, we've tested anecdotally on a few other datasets too, with good results.

This is just a first pass, I hope to keep improving this as available training data grows for this type of camera. If others are working on this type of image, let me know!

Also I'm pumped to set up one of these cameras in my already-way-overmonitored backyard (I think my yard has more camera traps than any other random suburban yard in North America)... CDFW has a neat guide to setting up these systems here:

https://storymaps.arcgis.com/stories/97b5182d6c7f4067937309cd758f9ee8

There is the small issue of needing to swap the lens out for a short-focus lens, which is not a major operation, but will be easier on some cameras than others, and everyone I've talked to so far who is doing this at scale is special-ordering pre-modified cameras from Reconyx. If anyone has set these up and has suggestions for how to swap a camera trap lens without special-ordering, I'm all ears.

snake~2004886_sa-fcwa4-2024__f53a6408-2130-4dc6-a98a-f9b7d9247e30 (1).jpg

❤️ Sara Beery, Nathaniel Rindlaub, Jason Holmberg (Wild Me), Yseult Hb, Eric Snyder, Toryn Schafer, Hannah Murray, Andrea Zampetti, Tiziana Gelmi Candusso, Kyra Swanson

😍 Jon Van Oast, Jason Holmberg (Wild Me), Sara Beery, Shir Bar, Yseult Hb, Alexander Merdian-Tarko

🙌 Mitchell Rogers, Gliday Yuka

👍 Ken Pierce

Levi Cai (levi@levicai.com)

2026-08-11 19:42:07

Are there some good off-the-shelf (or close) backyard camera trap setups that like..auto email you the cool photos of the day...?

Dan Morris (agentmorris@gmail.com)

2026-08-11 19:57:32

I'm not an expert on connected cameras, so YMMV, but I'll try...

There are lots of connected (cellular) cameras that send pictures to a dashboard that you can view in the cloud; search Amazon for "cellular trail camera".

Caveats:

I've never set one of those up because you have to supply a SIM card, and it's sort of a hassle (and not free) to have another SIM card in my life just for more fun pictures from my yard. GardePro makes a WiFi system, but it's expensive and requires a special WiFi base station, i.e. it doesn't just connect to your home WiFi.

AFAIK all of the cell cameras you would buy on Amazon send images to a manufacturer-specific dashboard or via SMS, so you can't easily view them in Google Photos or whatever, but you can, e.g., ask @Peter van Lunteren about AddaxAI Connect (https://addaxdatascience.com/addaxai-connect).

❤️ Levi Cai, Peter van Lunteren

Levi Cai (levi@levicai.com)

2026-08-11 20:03:05

Hmm.....maybe I'll just dig up an old computer + webcam and setup something much simpler for now 😛 (only recently have access to a backyard with real wildlife)

Digital Naturalism Laboratories (Andrew Quitmeyer) (andrew.quitmeyer@gmail.com)

2026-08-12 01:04:12

A cheap easy hack is just to double side tape some reading glasses over your cameras to make them focus closer

🤓 Kalindi Fonda, Dan Morris, Mitchell Rogers

Nuno Sa (Nuno.Cesar.Sa@wwf.de)

2026-08-17 05:22:43

I had some colleagues working on roadkill detection that ran to similar issues (they were targeting small lizards and other small mammals). So i wonder of much of what they do is relevant to you: mdpi.com/2220-9964/8/12/565

MDPI

An Improved Mobile Mapping System to Detect Road-Killed Amphibians and Small Birds

Roads represent a major source of mortality for many species. To mitigate road mortality, it is essential to know where collisions with vehicles are happening and which species and populations are most affected. For this, moving platforms such as mobile mapping systems (MMS) can be used to automatically detect road-killed animals on the road surface. We recently developed an MMS to detect road-killed amphibians, composed of a scanning system on a trailer. We present here a smaller and improved version of this system (MMS2) for detecting road-killed amphibians and small birds. It is composed of a stereo multi-spectral and high definition camera (ZED), a high-power processing laptop, a global positioning system (GPS) device, a support device, and a lighter charger. The MMS2 can be easily attached to any vehicle and the surveys can be performed by any person with or without sampling skills. To evaluate the system’s effectiveness, we performed several controlled and real surveys in the Évora district (Portugal). In real surveys, the system detected approximately 78% of the amphibians and birds present on surveyed roads (overlooking 22%) and generated approximately 17% of false positives. Our system can improve the implementation of conservation measures, saving time for researchers and transportation planning professionals.

Original URL: https://www.mdpi.com/2220-9964/8/12/565

👍 Dan Morris

Peter van Lunteren (peter@addaxdatascience.com)

2026-08-18 03:08:54

✅ Implemented into AddaxAI: docs.addaxai.com/docs/reference/model-zoo#…

docs.addaxai.com

Model zoo | AddaxAI

Every model AddaxAI ships, in one table. Select a model to read its full description, licence and citation.

Original URL: https://docs.addaxai.com/docs/reference/model-zoo#SAC-DMO-v1

🎉 Dan Morris, Nathaniel Rindlaub, Owen Martin

Dan Morris (agentmorris@gmail.com)

2026-08-18 10:31:43

To the earlier discussion of setting up a short-focus camera... a colleague pointed me to a GardePro camera that has a short focus (20cm) out of the box:

https://www.trailcampro.com/products/gardepro-t5cf-pro-non-cellular

As much as I like the idea of hot-gluing reading lenses to a regular camera, there's plenty of room for DIY spirit in setting up the enclosure, so using a camera that already supports a short focus seems like a good solution.

Nathaniel Rindlaub (nathaniel.rindlaub@tnc.org)

2026-08-20 18:38:21

This model is also now live in Animl! HMU or click through our sign up form here animl.camera if you'd like to try it out in the cloud or with cell/wireless cameras

🎉 Dan Morris, Peter van Lunteren

Nathaniel Rindlaub (nathaniel.rindlaub@tnc.org)

2026-08-11 19:37:17

Stoked for this Dan; thanks so much for your work on this model! For a low-tech, extremely hacky solution I've had ok luck hot gluing dollar-store reading glasses lenses to regular camera traps to change their focal length

😎 Sara Beery, Dan Morris

Nathaniel Rindlaub (nathaniel.rindlaub@tnc.org)

2026-08-11 19:38:01

but also if you have existing reconyx cams, you can send them back and they will adjust the focus to your specifications for you

😮 Jon Van Oast

Jon Van Oast (jon@wildme.org)

2026-08-11 19:39:40

we were just commenting on how we might catch some images of the little whiptail lizards (Aspidoscelis neomexicanus) that run around our backyard. now i might have a weekend project. 🤔

🦎 Dan Morris, Jason Holmberg (Wild Me), Yseult Hb, Carly Batist

Nathaniel Rindlaub (nathaniel.rindlaub@tnc.org)

2026-08-11 19:41:40

And if anyone wants an easy way to give this model a whirl, it will be available in Animl very soon.

🎉 Jon Van Oast, Dan Morris, Sara Beery, Yseult Hb

Jason Holmberg (Wild Me) (holmbergius@gmail.com)

2026-08-11 19:59:37

Annotated amphibian data!!!!! Yesssssssss..... 🐍🦎yay-frog

🎉 Jon Van Oast, Dan Morris, Carly Batist

Rupa Kurinchi-Vendhan (rupak272@mit.edu)

2026-08-13 09:31:19

🌍🍹AI for Planetary Health – ECCV 2026 Social 🍹 🌍

Attending ECCV 2026 in Malmö this year? Join us for an informal evening social bringing together folks intersection of computer vision, AI, climate, biodiversity, conservation, and oceans!

📍Location: Cantin, Malmö
🕖 Date & time: Sept 10th, 19:00–22:00
🎟️ Registration via Luma: luma.com/…

This community event is jointly organized by:
Climate AI Nordics
ECCV Marine Computer Vision Workshop & Community
AI for Conservation & CV4Ecology Community
asterisk labs