siltframe2026cc-by-sa-4.0
Off-Road Segmentation Stress Test
165 labelled images of off-road scenes under various adverse conditions (dust, night, fog, rain on lens, mud on lens) at multiple severities, plus real rain and snow frames, designed to evaluate robustness of perception models.
Downloads73
Episodes165
Why This Matters for Physical AI
This dataset is critical for evaluating robustness of autonomous vehicle perception systems under adverse weather and lighting conditions, revealing that dataset coverage significantly outperforms augmentation for handling real-world environmental degradation.
Technical Profile
- Modalities
- rgbsemantic_segmentation_labels
- Robot Embodiments
- autonomous_vehicle
- Environment
- outdooroff-road
- Task Types
- semantic_segmentationperception
- Episodes
- 165
- Data Format
- JPEG images with PNG labels
- Annotation Types
- semantic_segmentationpixel-level_labels
- License
- cc-by-sa-4.0
Access
Need custom rgb data?
Claru builds purpose-built datasets for outdoor applications with dense human annotations and quality assurance.
Request a Sample Pack