Upstream training splits; known explicitly identified test/eval rows excluded. Cross-dataset benchmark overlap is not guaranteed. Published as a raw, manually gated release; source annotation caveats remain.

Task views reuse original image archives. Coordinates are normalized 0–1000. Within each query targets are sorted left-to-right then top-to-bottom.

HF preview configs contain 10 examples per view, not the complete training split. Full training uses metadataset.yaml and examples/read_energon.py (Energon 7.4).

Source	View	Records	Targets	Examples
dair_v2x	boxes	3,042	41,754	10 annotated examples
waymo	boxes	648,338	7,817,150	10 annotated examples

What is not claimed

No guarantee of cross-dataset benchmark disjointness, global image uniqueness, semantic annotation correctness, or redistribution/license clearance. Shared box/point views are alternate tasks, not distinct images. Source-specific caveats remain applicable.

The package preserves upstream query text; metadata cleanup does not alter images or embedded image metadata.

Downloads last month: 21

Total file size:

241 GB