Dex-GEM2025odbl
DexCanvas
A dataset of 111,857 complete right-hand trajectories and 55,442,683 frames of rigid hand–object interaction with dual-camera captures, hand kinematics, object poses, and per-frame contact forces.
Downloads328
Episodes111857
Likes3
Why This Matters for Physical AI
DexCanvas provides large-scale hand-object interaction data with contact forces and kinematic annotations that enables training robust dexterous manipulation policies grounded in real human demonstrations.
Technical Profile
- Modalities
- rgbproprioceptionforce_torque
- Robot Embodiments
- hand
- Action Space
- joint_positions
- Environment
- lab
- Task Types
- dexterous_manipulationhand_object_interactiongrasping
- Episodes
- 111857
- Data Format
- Parquet
- Annotation Types
- action_labelscontact_forcessegmentation
- License
- odbl
Access
Need custom rgb data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack