Hi everyone.
I am currently working on a project looking at the density and distribution of lions and hyenas in Nairobi National Park using spatially explicit capture-recapture models. As part of the model requirements, I am required to individually identify each lion and hyena. For lions, this is done via their whisker spots and for hyenas their spot patterns. My camera traps have been set up to capture both flanks of the animal.
I have so far conducted a short pilot study and collected 28 days worth of camera trap video footage. I would like to ask if anybody knows of a good machine learning model that can be used to:
a) Sort through camera trap footage, remove blanks and categorise species.
b) Individually identify lions and hyenas
Please let me know if anybody knows where to start. I'm new to the machine learning and AI world, but I am always up for learning something new!
28 January 2026 3:14pm
Yo could possibility use something like addax AI offline.
AddaxAI - Simplifying camera trap image analysis with AI
Previously known as EcoAssist. An AI platform to analyse images with machine learning models for automatic detection, so that ecologists can save time and focus on conservation efforts.
28 January 2026 4:30pm
28 January 2026 10:07pm
Carly covers it well, adding a couple of notes:
- I would encourage you not to require one system that does everything. There's definitely not a single AI model that does all those things, and while I'm aware of one platform that does all of those things (with different AI models) (more on this in a second), you will likely ignore a lot of other factors that might end up being important to you if you decide that, for example, you have to do your species classification and individual ID in the same system.
- I would encourage you to evaluate tools not on whether they can automate these tasks, but rather on whether they make you faster and "click" with you in terms of ergonomics and workflow. None of these steps (except maybe blank elimination) can be fully automated yet. I posted a longer version of this spiel in an answer on this thread.
- All of that said, TrapTagger is the only platform I'm aware of that directly integrates all of the steps you talked about: (a) blank elimination, (b) species ID (for South African wildlife, but I would encourage you to think of that as close enough to East African wildlife in terms of the way it speeds up a workflow), and (c) individual ID, especially for video (many tools/platforms don't support video yet). I'm not telling you to use TrapTagger (and in full disclosure I've never tried it with video, though the documentation indicates that it supports video); it's one of many great tools/platforms out there, and as per above you might not want to do everything in one tool. But FWIW TrapTagger is the closest I'm aware of to a "one-stop shop" for your requirements.
Hope that helps! If I can help you with any other questions, feel free to drop me an email ( [email protected] ).
1 February 2026 9:47am
Thanks everybody for the useful feedback. I'm going to try 3 or 4 different models like the ones Carly suggested and find the one that works for me. Dan I like your points, its likely I'll end up using different models for species categorisation and individual ID.
Arky