CREA Mont-Blanc : ScienceSandwich

AI and camera traps: how can you recognize wildlife and its behaviour automatically?

Camera traps have become an indispensable tool for the census, monitoring and management of many animal species. They provide low-cost data over large areas, with continuous temporal monitoring. However, the images collected - often in the hundreds of thousands - must be inspected one by one, a long and tedious task, in order to identify the species of the individuals present in the images obtained.

In this presentation, we'll look at how artificial intelligence (AI) can automate the analysis of these images. As part of my thesis and the DeepFaune initiative, we are developing new methods and a tool designed to improve the automatic classification of species in photo-trap images. In collaboration with CREA Mont-Blanc, we have also developed a new approach to automatically classify the behavior of observed individuals.

Gaspard Dussert is a third-year PhD student at the Laboratoire de Biométrie et Biologie Évolutive in Lyon and a member of the DeepFaune team (https://www.deepfaune.cnrs.fr/).