Imagine opening a camera trap and finding thousands of wildlife photographs waiting to be sorted. An open-source artificial intelligence (AI) model can now take over much of that work. Trained on more than 1.1 million camera-trap images collected from over 10,000 locations across the dry and moist deciduous forests of Central India, the model identifies 40 classes of wild animals with 97.3 per cent accuracy, offering wildlife researchers and forest department officials a faster way to analyse data generated during wildlife surveys.

Manual challenge

Wildlife researchers and forest department officials using camera traps often collect thousands, and sometimes lakhs, of images. Sorting these photographs and identifying species manually is tedious and time-consuming.