COT5415_Computer_Vision / docs /data /ADDITIONAL_DATASET_SCOPE.md
iiCEMAN's picture
Add PACO-LVIS (v2 unified release, annotations only; images from COCO 2017)
b0d869f verified
|
Raw History Blame Contribute Delete
1.85 kB

Additional dataset scope

Current local release: PartImageNet OOD and the supplied Compositor-style Pascal-Part subset. These are the two converted, validated sources. This scope describes the release contents; it does not remove the proposal's additional-dataset commitment.

Next source: PACO-LVIS

Use PACO-LVIS as the next planned extension, rather than counting a Pascal-derived person subset as an independent source. Its official annotations contain object and part masks plus obj_ann_id links, providing instance-level part/whole relations beyond the current semantic groupings. Retain image IDs, original splits, category namespaces, annotation IDs, masks and exhaustive/negative-category metadata. Do not discard these fields or derive parent identity merely by overlap.

Official resources:

Integration acceptance criteria: resolve the documented annotation version and source images; preserve explicit object links and mask geometry; identify void/negative/non-exhaustive semantics; validate split disjointness and every parent reference; extend the shared loader without fabricating missing masks; publish source-specific counts and license records. An annotation-only preview must be labelled as such and cannot be counted as a complete image dataset.

PACO-EGO4D is outside the immediate extension because video access and temporal instances add requirements not needed to demonstrate a third image source. Status 2026-11-18: PACO-LVIS v1 has been downloaded, converted and validated as the third source (see DATASET_CARD.md and UNIFIED_SCHEMA.md); it has not been benchmarked with CAST. The official repository's software license is not automatically a blanket license for underlying COCO/LVIS/Ego4D images.