Data for "PriVi: Towards a General-Purpose Video Model for Primate Behavior in the Wild"

Résumé fourni par la source

Non-human primates are our closest living relatives, and analyzing their behavior is central to research in cognition, evolution, and conservation. Computer vision could greatly aid this research, but existing methods often rely on human-centric pretrained models and focus on single datasets, which limits generalization. We address this limitation by shifting from a model-centric to a data-centric approach and introduce PriVi, a large-scale primate-centric video pretraining dataset. PriVi contains 424 hours of curated video, combining 174 hours from behavioral research across 11 settings with 250 hours of diverse web-sourced footage, assembled through a scalable data curation pipeline. We find that primate-centric pretraining substantially improves performance, data efficiency and generalization, making it a promising approach for low-label applications. Dataset, code, and models are available at https://privi.eckerlab.org. **Note:** This release only contains preprocessed video snippets so far. Raw videos and clip embeddings will be released shortly. For early access, contact [felix.mueller@cs.uni-goettingen.de](mailto:felix.mueller@cs.uni-goettingen.de).

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.