Felidae Conservation Fund: volunteers and AI sorting camera-trap photos of pumas
Felidae Conservation Fund runs camera traps across California to study pumas and bobcats. Its platform, WildePod, runs open-source AI models that throw out empty photos and suggest species, so more than 300 trained volunteers can spend their time on the photos that matter.
What they learned
Letting AI throw out empty frames first let volunteers focus on the photos that matter
From public sources: Felidae writes that wind, shadows and moving plants set off its cameras, producing thousands of empty photos in a single batch. Running MegaDetector first removes those, and SpeciesNet then suggests species. Felidae reports that volunteers have annotated over two million images since 2021 this way. It does not publish accuracy or time-saved figures.
The problem
Monitoring wild cats over years produces millions of photos, many of them empty, and a small nonprofit can't pay people to look at them all.
What the AI does
MegaDetector flags photos with animals, people or vehicles and removes the empty ones. SpeciesNet then suggests which species is in each remaining photo, and its newest version can label photos on its own when it is confident enough. Volunteers check and annotate the rest.