Research map / Video / Action recognition
MammalNet: A Large-Scale Video Benchmark for Mammal Recognition and Behavior Understanding
Jun Chen, Ming Hu, Darren J. Coker, Michael L. Berumen, Blair Costelloe, Sara Beery, Anna Rohrbach, Mohamed Elhoseiny
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| initialize MViT V2 with Kinetics 400 pretraining | MViT V2 · MammalNet standard classification | animal per-class top-1 accuracy: 19.7 to 32.5 | Improved |
| initialize MViT V2 with Kinetics 400 pretraining | MViT V2 · MammalNet standard classification | behavior per-class top-1 accuracy: 27.7 to 37.8 | Improved |
| add one labeled example per animal-behavior composition | MViT V2* · MammalNet compositional low-shot | behavior per-example accuracy: 26.2 to 28.9 | Improved |
| add five labeled examples per animal-behavior composition rather than zero | MViT V2* · MammalNet compositional low-shot | behavior per-example accuracy: 26.2 to 31.7 | Improved |
| add five labeled examples per animal-behavior composition rather than zero | MViT V2* · MammalNet compositional low-shot | animal per-example accuracy: 32.2 to 39.3 | Improved |
| train animal and behavior recognition jointly instead of separately | MammalNet classification model · MammalNet standard classification | behavior per-class accuracy: ~2.2 per-class accuracy | Improved |
| use joint animal-and-behavior training rather than animal-only prediction | MammalNet classification model · MammalNet standard classification | animal recognition | Hurt |
| transfer behavior knowledge from seen animals to unseen animal-behavior compositions | MViT V2* · MammalNet compositional low-shot (0-shot) | behavior per-example accuracy: 26.2 | Improved |
Closest papers by meaning
- MammAlps: A Multi-view Video Behavior Monitoring Dataset of Wild Mammals in the Swiss Alps (CVPR 2025 · similarity 0.93)
- Animal Kingdom: A Large and Diverse Dataset for Animal Behavior Understanding (CVPR 2022 · similarity 0.91)
- The SA-FARI Dataset: Segment Anything in Footage of Animals for Recognition and Identification (CVPR 2026 · similarity 0.90)
- The PanAf-FGBG Dataset: Understanding the Impact of Backgrounds in Wildlife Behaviour Recognition (CVPR 2025 · similarity 0.90)
- Animal-Bench: Benchmarking Multimodal Video Models for Animal-centric Video Understanding (NeurIPS 2024 · similarity 0.89)
- BuckTales: A multi-UAV dataset for multi-object tracking and re-identification of wild antelopes (NeurIPS 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
