Benchmark
COCO-Stuff
A scene understanding dataset extending MS-COCO with pixel-level segmentation annotations for 91 amorphous background region categories—such as grass, sky, and water—supplementing the original object annotations and supporting evaluation of complex scene semantic understanding.
Timeline (last 90 days)
Oct 12
The research team validated HnS on two datasets, CelebA and COCO-Stuff.
Unverified50%
Oct 12
COCO-Stuff is a scene understanding dataset extended from MS-COCO, supplementing the original object annotations with pixel-level segmentation annotations for 91 categories of stuff.
Unverified50%