DexCap

Stanford research team

Hand motion capture and scene observations for learning dexterous manipulation from human demonstrations.

What you can explore

Study the path from recorded human motion to hand–arm policies.

Before you use it

The reported robot setup uses LEAP Hand. Other embodiments require retargeting and validation.

Original contributors

Chen Wang, Haochen Shi, Weizhuo Wang, Ruohan Zhang, Li Fei-Fei and C. Karen Liu

License / access: Check data and code terms at source

Project, data & code

Editorial reference · checked 2026-09-22