Start of funding 01.01.2024

Video-based Artificial Intelligence Methods for Re-Identification and Behaviour Prediction for Animals

Prof. Dr. Björn Eskofier
Friedrich-Alexander-University of Erlangen-Nuremberg
Machine Learning and Data Analytics Lab

Dr. Mathias Tobler
San Diego Zoo Wildlife Alliance
Conservation Science and Wildlife Health



The Video-based Re-Identification and Behaviour Analysis for Animals (VERA) project is an interdisciplinary collaboration led by the Machine Learning and Data Analytics Lab (MaD Lab). The VERA project forms a bridge between biological and ecological research and the machine learning field in order to develop a new generation of video-based algorithms for the tasks of re- identification of individual animals as well as detection of shown behaviour. Our focus here is on implementing the models in such a way that they utilise the movement of the animals as a temporal feature. This helps both in the differentiation of individual animals as well as in the detection of displayed behaviour.