Recognition and Spatio-temporal Modeling of Animal Species in the Mercantour National Park Thanks to AI
This project is based on a collaboration between the Mercantour National Park, on the one hand, and the 3IA Côte d’Azur Institute and the MSI of Université Côte d’Azur, on the other, with the aim of using AI tools to conduct an analysis of audio-visual data concerning the animal species living in the park.
The database was obtained thanks to intensive photographic and video trapping, carried out in the Mercantour Park. It bears witness to the richness of the park - one of the 11 French national parks - in terms of biodiversity and in particular variety of species and ecosystems.
To make the best use of the data provided by the videos and images, Université Côte d'Azur and the 3IA Côte d'Azur Institute will provide the project with a team of researchers and engineers specializing in artificial intelligence. The project has the following objectives:
- Detection of species present in a video and counting of individuals.
- Study of the connection between the observation of certain species and human presence.
- Spatial and temporal modeling of the presence of certain species and linking with field surveys (GPS tracks) and genetic data.
To do so, the research team will use the appropriate AI tools available, such as deep learning, active learning and learning transfer algorithms. Dedicated learning methodologies will certainly have to be developed for the counting part (objective 1) and for linking with external data of various types (objective 3).
