COCO-Stuff (Common Objects in COntext-stuff)

Introduced by Caesar et al. in COCO-Stuff: Thing and Stuff Classes in Context

The Common Objects in COntext-stuff (COCO-stuff) dataset is a dataset for scene understanding tasks like semantic segmentation, object detection and image captioning. It is constructed by annotating the original COCO dataset, which originally annotated things while neglecting stuff annotations. There are 164k images in COCO-stuff dataset that span over 172 categories including 80 things, 91 stuff, and 1 unlabeled class.

Source: Image Colorization: A Survey and Dataset

Homepage

Benchmarks

Add a new result Link an existing benchmark

Task	Dataset Variant	Best Model
Semantic Segmentation	COCO-Stuff test	EVA
Unsupervised Semantic Segmentation	COCO-Stuff-27	CAUSE
Image-to-Image Translation	COCO-Stuff Labels-to-Photos	DP-SIMS
Unsupervised Semantic Segmentation with Language-image Pre-training	COCO-Stuff-171	TagAlign
Open Vocabulary Semantic Segmentation	COCO-Stuff-171	POMP
Unsupervised Semantic Segmentation	COCO-Stuff-3	IIC
Layout-to-Image Generation	COCO-Stuff 256x256	LayoutDiffusion
Layout-to-Image Generation	COCO-Stuff 64x64	OC-GAN
Layout-to-Image Generation	COCO-Stuff 128x128	CAL2IM
Unsupervised Semantic Segmentation with Language-image Pre-training	COCO-Stuff-27	ReCo+
Unsupervised Semantic Segmentation	COCO-Stuff-171	CAUSE-TR
Unsupervised Semantic Segmentation	COCO-Stuff-81	CAUSE-TR
Unsupervised Semantic Segmentation	COCO-Stuff-15	InfoSeg
Sketch-to-Image Translation	COCO-Stuff	PITI
Real-Time Semantic Segmentation	COCO-Stuff	BiSeNet V2-Large
Unsupervised Image Segmentation	COCO-Stuff-27	SAN
Semantic Segmentation	COCO-Stuff full	SegFormer-B5
Semantic Segmentation	COCO-Stuff	Deeplab v2
Zero-Shot Semantic Segmentation	COCO-Stuff	MAFT
Semantic Segmentation	COCO-Stuff-27	DiffSeg