jagennath-hari2012mit
NYUv2
A preprocessed RGB-D dataset for indoor scene understanding tasks including semantic segmentation, depth estimation, and instance segmentation. The dataset contains aligned RGB images, depth maps, semantic masks, and instance masks converted from the original NYU Depth Dataset V2 into a modern ML-friendly format.
Downloads265
Episodes1449
Why This Matters for Physical AI
NYUv2 provides benchmark RGB-D scene understanding capabilities essential for robotic perception in indoor environments, enabling vision-based navigation and manipulation through depth estimation and semantic scene segmentation.
Technical Profile
- Modalities
- rgbdepth
- Environment
- indoor
- Task Types
- depth_estimationsemantic_segmentationinstance_segmentation
- Episodes
- 1449
- Data Format
- HDF5
- Annotation Types
- semantic_segmentationinstance_segmentationdepth_labels
- License
- mit
Community Signals
Top 50% by downloads
HuggingFace Discussions1
Access
Need custom rgb data?
Claru builds purpose-built datasets for indoor applications with dense human annotations and quality assurance.
Request a Sample Pack