The DeepFaune initiative: a collaborative effort towards the automatic identification of European fauna in camera trap images
Noa Rigoudy, Gaspard Dussert, Abdelbaki Benyoub, Aurélien Besnard, Carole Birck, Jérome Boyer, Yoann Bollet, Yoann Bunz, Gérard Caussimont, Elias Chetouane, Jules Chiffard, Pierre Cornette, Anne Delestrade, Nina De Backer, Lucie Dispan, Maden Le Barh, Jeanne Duhayer, Jean-François Elder, Jean-Baptiste Fanjul, Jocelyn Fonderflick, Nicolas Froustey, Mathieu Garel, William Gaudry, Agathe Gérard, Olivier Giménez, Arzhela Hemery, Audrey Hemon, Jean‐Michel Jullien, Daniel Knitter, Isabelle Malafosse, Mircea Mărginean, Louise Ménard, Alice Ouvrier, Gwennaelle Pariset, Vincent Prunet, Julien Rabault, Malory Randon, Yann Raulet, Antoine Régnier, Romain Ribière, Jean-Claude Ricci, Sandrine Ruette, Yann Schneylin, Jérôme Sentilles, Nathalie Siefert, Bethany R. Smith, Guillaume Terpereau, Pierrick Touchet, Wilfried Thuiller, Antonio Uzal, Valentin Vautrain, Ruppert Vimal, Julian Weber, B. Spataro, Vincent Mièle, Simon Chamaillé‐Jammes
Abstract
Camera traps have revolutionized how ecologists monitor wildlife, but their full potential is realized only when the hundreds of thousands of collected images can be readily classified with minimal human intervention. Deep learning classification models have allowed extraordinary progress towards this end, but trained models remain rare and are only now emerging for European fauna. We report on the first milestone of the DeepFaune initiative ( https://www.deepfaune.cnrs.fr ), a large-scale collaboration between more than 50 partners involved in wildlife research, conservation and management in France. We developed a classification model trained to recognize 26 species or higher-level taxa that are common in Europe, with an emphasis on mammals. The classification model achieved 0.97 validation accuracy and often > 0.95 precision and recall for many classes. These performances were generally higher than 0.90 when tested on independent out-of-sample datasets for which we used image redundancy contained in sequences of images. We implemented our model in a software to classify images stored locally on a personal computer, so as to provide a free, user-friendly, and high-performance tool for wildlife practitioners to automatically classify camera trap images. The DeepFaune initiative is an ongoing project, with new partners joining regularly, which allows us to continuously add new species to the classification model.