Pose + Behavior: Is a Unified AI Model More Helpful?

Описание: Pose-estimation tools have made it very easy to track body parts, but these tools often require post-processing of tracked body parts and separate classifiers to get to behaviors. Even then, manual verification of clustered behaviors is often required to match clusters to ethologically relevant behaviors. So, I was wondering if it would be possible to train a single model to do both pose-estimations and behavior identification (at least for certain visually distinct behaviors). In this video, I demonstrated the feasibility of this approach and request community feedback regarding this approach's usefulness. Feel free to leave a comment and let me know what you think.