Cityscapes

Introduced by Cordts et al. in The Cityscapes Dataset for Semantic Urban Scene Understanding

Cityscapes is a large-scale database which focuses on semantic understanding of urban street scenes. It provides semantic, instance-wise, and dense pixel annotations for 30 classes grouped into 8 categories (flat surfaces, humans, vehicles, constructions, objects, nature, sky, and void). The dataset consists of around 5000 fine annotated images and 20000 coarse annotated ones. Data was captured in 50 cities during several months, daytimes, and good weather conditions. It was originally recorded as video so the frames were manually selected to have the following features: large number of dynamic objects, varying scene layout, and varying background.

Source: A Review on Deep Learning Techniques Applied to Semantic Segmentation

Homepage

Benchmarks

Add a new result Link an existing benchmark

Task	Dataset Variant	Best Model
Semantic Segmentation	Cityscapes test	VLTSeg
Semantic Segmentation	Cityscapes val	SERNet-Former
Panoptic Segmentation	Cityscapes val	OneFormer
Real-Time Semantic Segmentation	Cityscapes test	PIDNet-L
Semi-Supervised Semantic Segmentation	Cityscapes 12.5% labeled	SemiVL
Semi-Supervised Semantic Segmentation	Cityscapes 25% labeled	SemiVL
Image-to-Image Translation	Cityscapes Labels-to-Photo	DP-SIMS
Semi-Supervised Semantic Segmentation	Cityscapes 50% labeled	SemiVL
Instance Segmentation	Cityscapes val	OneFormer
Instance Segmentation	Cityscapes test	Deep Watershed Transform
Semi-Supervised Semantic Segmentation	Cityscapes 6.25% labeled	SemiVL
Semi-Supervised Semantic Segmentation	Cityscapes 100 samples labeled	SemiVL
Panoptic Segmentation	Cityscapes test	OneFormer
Unsupervised Monocular Depth Estimation	Cityscapes	RM-Depth
Unsupervised Semantic Segmentation	Cityscapes test	CAUSE
Video Semantic Segmentation	Cityscapes val	TMANet-50
Federated Learning	Cityscapes heterogeneous	SiloBN + ASAM
Real-Time Semantic Segmentation	Cityscapes val	PIDNet-L
Robust Object Detection	Cityscapes	OA-DG
Unsupervised Semantic Segmentation with Language-image Pre-training	Cityscapes val	TTD
Image Generation	Cityscapes	Projected GAN
Image-to-Image Translation	Cityscapes Photo-to-Labels	pix2pix
Domain 1-1	Cityscapes	RCNet-101
Domain 11-5	Cityscapes	RCNet-101
Domain 11-1	Cityscapes	RCNet-101
Open Vocabulary Semantic Segmentation	Cityscapes	FC-CLIP
Video Prediction	Cityscapes 128x128	GHVAEs
Semi-Supervised Semantic Segmentation	Cityscapes 2% labeled	GIST and RIST
Semi-Supervised Semantic Segmentation	Cityscapes 5% labeled	GIST and RIST
Video Prediction	Cityscapes	DMVFN
Semi-Supervised Semantic Segmentation	Cityscapes 93 labeled	AEL
Domain 1-1	Cityscapes val	PLOP
Domain 11-5	Cityscapes val	PLOP
Real-Time Semantic Segmentation	Cityscapes	S^2-FPN34
Semantic Segmentation	Cityscapes	SPFNet34M
Robust Object Detection	Cityscapes test	Faster R-CNN with Stylized Training Data
Multi-Task Learning	Cityscapes test	Nash-MTL
Semi-Supervised Semantic Segmentation	Cityscapes with extra (no coarse labels)	Dense FixMatch
Monocular Depth Estimation	Cityscapes	SwinMTL
Domain 11-1	Cityscapes val	PLOP
Image Generation	Cityscapes-5K 256x512	SB-GAN
Edge Detection	Cityscapes test	CASENet
Semi-Supervised Semantic Segmentation	Cityscapes 10% labeled	IM++
Weakly-Supervised Semantic Segmentation	Cityscapes val	CARB
Overlapped 10-1	Cityscapes	MiB+AWT
Overlapped 14-1	Cityscapes	MiB+AWT
Weakly-Supervised Semantic Segmentation	Cityscapes test	CARB
Interactive Segmentation	Cityscapes val	IOG
Unsupervised Semantic Segmentation	Cityscapes val	Segmenter ViT-S/16
Depth Estimation	Cityscapes test	SDC-Depth
Scene Parsing	Cityscapes test	VCD No Coarse
2D Semantic Segmentation	Cityscapes val	SERNet-Former
Real-time Instance Segmentation	Cityscapes test	CenterPoly
Image Generation	Cityscapes-25K 256x512	SB-GAN