gagandeepreehal2026cc-by-nc-4.0
INDUS-AD: Indian Dataset of Unstructured Urban Scenes for Autonomous Driving
A 4.006 TB research dataset built from 1,469 driving recordings across Bengaluru, India, containing 9,927 face/plate-blurred multi-camera MP4s, 588.57 aggregate camera-view hours, normalized sensor tables, and derived semantic and depth layers for autonomous driving research.
Downloads3K
Episodes1469
Hours588.57
Likes1
Why This Matters for Physical AI
INDUS-AD provides large-scale multi-camera autonomous driving data from unstructured urban Indian traffic, supporting research in vision-based perception, end-to-end driving models, and real-world robotics applications in diverse driving conditions.
Technical Profile
- Modalities
- rgbdepthgnsssemantic_segmentationcamera_calibrationpose
- Environment
- urbanoutdoor
- Task Types
- autonomous_drivingperceptionlocalizationdepth_estimationsemantic_segmentation
- Episodes
- 1469
- Total Hours
- 588.57
- Data Format
- parquet, MP4, JSONL
- Annotation Types
- semantic_segmentationdepth_estimationcamera_annotationsweather_labels
- License
- cc-by-nc-4.0
Community Signals
Top 25% by downloads
Access
Need custom rgb data?
Claru builds purpose-built datasets for urban applications with dense human annotations and quality assurance.
Request a Sample Pack