Datasets:
Formats:
parquet
Languages:
English
Size:
100K - 1M
Tags:
autonomous-driving
vision-language-model
visual-grounding
chain-of-thought
long-tail
motion-planning
License:
Commit History
-
Add AnchorReasoning annotations: 416,119 frames of VG-CoT + instance masks (Parquet)
Verified signatureTwo configs keyed to WOD-E2E frame.context.name: annotations (scene context, traffic events, decision-critical elements with boxes/attributes/impact rank/implications, rationale, action plan) and masks (COCO compressed RLE). Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>ee12d56