dguralcc-by-4.0
droid-frames
5,172 RGB frames (1 fps, 640 px) decoded from 75 DROID robot-arm episodes for the Nebius x Voxel51 Physical AI workshop. Frames include robot state, CLIP embeddings, UMAP visualizations, Grounding DINO auto-labels, and YOLO11n predictions.
Downloads35
Episodes75
Why This Matters for Physical AI
This dataset provides annotated robotic manipulation frames with object detection labels and embeddings that enable training and evaluation of vision models for physical AI tasks.
Technical Profile
- Modalities
- rgbproprioception
- Robot Embodiments
- robot-arm
- Environment
- lab
- Task Types
- object-detectionmanipulation
- Episodes
- 75
- Data Format
- FiftyOne
- Annotation Types
- action_labelsbounding_boxes
- License
- cc-by-4.0
Access
Need custom rgb data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack