xuejf
EgoSuite-Open10K
Real-world egocentric human activity dataset with 10,000 hours of video across 7 environment families, captured from head and wrist viewpoints with structured pose and semantic supervision for embodied AI and robot learning.
Downloads0
Hours10000
Why This Matters for Physical AI
EgoSuite-Open10K provides large-scale real-world egocentric human demonstration data with multi-view synchronization and rich pose/semantic supervision, enabling training of embodied AI systems and world models for complex, long-horizon manipulation and activity understanding.
Technical Profile
- Modalities
- rgbhand_posebody_posesemantic_annotation
- Environment
- homehospitalityretailsportslogisticsofficeindustry
- Task Types
- cookingcleaningobject_organizationfood_preparationpickingrestockingpackingtool_useassemblymanipulationfine_manipulation
- Total Hours
- 10000
- Data Format
- WebDataset
- Annotation Types
- hand_posebody_posesemantic_labelslanguage_instructions
Access
Need custom rgb data?
Claru builds purpose-built datasets for home applications with dense human annotations and quality assurance.
Request a Sample Pack