Package for action recognition with multimodal masked autoencoders. This package enables behavioral analysis from video, audio, depth cameras, etc.
The work has been road tested and developed for three recent projects from the lab:
- MammalAlps (CVPR highlight 2025)
- EPFL smart kitchen (NeurIPS 2025)
- DLC2action (biorxiv 2025)
It will appear in July 2026 -- we apologize for the delay!!