Spaces:

---

kehong

/

MoCapAnythingV2

like 3

Running on Zero

Space: https://huggingface.co/spaces/kehong/MoCapAnythingV2

Official interactive demo for MoCapAnything V2 (featured in HF Daily Papers: https://huggingface.co/papers/2604.28130). End-to-end motion capture & retargeting for arbitrary skeletons from monocular video — pick a video and a target species/object, get the predicted 3D pose. Weights and demo data are already on the Hub (kehong/MoCapAnythingV2-weights, kehong/MoCapAnythingV2-data-sample) and the GitHub repo runs end-to-end.

Requesting a ZeroGPU community grant to keep the demo running for the community. @NielsRogge from the HF open-source team offered to help attach it (GitHub issue: https://github.com/phongdaot/MocapAnything/issues/9).