harpreetsahota2025apache-2.0
lingbot-depth-subset
A representative subset of the LingBot-Depth RGB-D training corpus (3.02M samples, 2.71–3.03 TB) reformatted into FiftyOne's grouped-dataset structure for depth completion and estimation tasks. Contains 13,149 samples across 3 sub-collections (RobbyReal, RobbyVla, RobbySim) spanning real-world and simulated RGB-D captures.
Downloads0
Episodes13,149 samples (10,207 groups)
Likes1
Why This Matters for Physical AI
This dataset enables research into masked depth modeling and depth completion for RGB-D perception systems, critical for enabling robots to reconstruct dense metric depth from partial sensor readings in real-world manipulation tasks.
Technical Profile
- Modalities
- rgbdepth
- Robot Embodiments
- Frankahumanoid
- Environment
- labsimulation
- Task Types
- depth-completiondepth-estimation
- Episodes
- 13,149 samples (10,207 groups)
- Data Format
- FiftyOne
- Annotation Types
- depth_ground_truth
- License
- apache-2.0
Access
Need custom rgb data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack