Codesota · Benchmark · CityscapesHome/Leaderboards/Vision & Documents/Semantic Segmentation/Cityscapes
Unknown

Cityscapes.

5,000 images with fine annotations and 20,000 with coarse annotations of urban street scenes.

Paper ↗Leaderboard ↓
§ 01 · SOTA history

Year over year.

§ 02 · Leaderboard

Results by metric.

Found a wrong score or missing run?
Use row edits to send a sourced correction into moderation.
Add / edit result ↗Report issue ↗

Miou

Miou is the reported evaluation metric for Cityscapes. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.

Higher is better

Trust tiers for Miouverifiedpapervendorcommunityunverified

Muted rows were not state of the art when published — an earlier or same-year result already scored better.

RankModelTrustScoreYearLinksFix
01EoMT (ViT-L)unverified84.22025Paper ↗Code ↗Looks wrong?
02SegFormer-B5
SegFormer-B5. mIoU on Cityscapes val. NVIDIA, NeurIPS 2021. Table 2.
verified842021Source ↗Looks wrong?
03Mask2Former (Swin-L)
Mask2Former with Swin-L backbone. mIoU on Cityscapes val. CVPR 2022. Table 3.
verified83.32021Source ↗Looks wrong?
04OneFormer (DiNAT-L)
OneFormer with DiNAT-L backbone. Single model for all segmentation tasks. mIoU on Cityscapes val. CVPR 2023. Table 3.
verified832022Source ↗Looks wrong?
05DINOv3 (7B)unverified81.12025Paper ↗Code ↗Looks wrong?
06DINOv2 (ViT-g/14)unverified812023Paper ↗Code ↗Looks wrong?
§ 04 · Submit a result

Add to the leaderboard.

← Back to Semantic Segmentation