Self-supervised Learning on Camera Trap Footage Yields a Strong Universal Face Embedder
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866911055002730496 |
|---|---|
| author | Iashin, Vladimir Lee, Horace Schofield, Dan Zisserman, Andrew |
| author_facet | Iashin, Vladimir Lee, Horace Schofield, Dan Zisserman, Andrew |
| contents | Camera traps are revolutionising wildlife monitoring by capturing vast amounts of visual data; however, the manual identification of individual animals remains a significant bottleneck. This study introduces a fully self-supervised approach to learning robust chimpanzee face embeddings from unlabeled camera-trap footage. Leveraging the DINOv2 framework, we train Vision Transformers on automatically mined face crops, eliminating the need for identity labels. Our method demonstrates strong open-set re-identification performance, surpassing supervised baselines on challenging benchmarks such as Bossou, despite utilising no labelled data during training. This work underscores the potential of self-supervised learning in biodiversity monitoring and paves the way for scalable, non-invasive population studies. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_10552 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Self-supervised Learning on Camera Trap Footage Yields a Strong Universal Face Embedder Iashin, Vladimir Lee, Horace Schofield, Dan Zisserman, Andrew Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning Camera traps are revolutionising wildlife monitoring by capturing vast amounts of visual data; however, the manual identification of individual animals remains a significant bottleneck. This study introduces a fully self-supervised approach to learning robust chimpanzee face embeddings from unlabeled camera-trap footage. Leveraging the DINOv2 framework, we train Vision Transformers on automatically mined face crops, eliminating the need for identity labels. Our method demonstrates strong open-set re-identification performance, surpassing supervised baselines on challenging benchmarks such as Bossou, despite utilising no labelled data during training. This work underscores the potential of self-supervised learning in biodiversity monitoring and paves the way for scalable, non-invasive population studies. |
| title | Self-supervised Learning on Camera Trap Footage Yields a Strong Universal Face Embedder |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2507.10552 |