List your company
ResearchFormerly WizWand

Semantic Segmentation on CityScapes (mIoU)

86.9mIoUState-of-the-Art (Liu et al., 2021)
-0.14822.45145.0567.649Apr 14, 2023Oct 23, 2023May 3, 2024Nov 12, 2024May 24, 2025Dec 3, 2025Jun 14, 2026
Updated 4mo ago

Evaluation results

MethodLinks
State-of-the-Art (Liu et al., 2021)Mode=Absolute state-of-the-art2023.0486.9
DINOv2Arch.=ViT-g/14, Evaluation=Multiscale (+ms)2023.0481
DINOv2Arch.=ViT-L/14, Evaluation=Multiscale (+ms)2023.0480.9
DINOv2Arch.=ViT-B/14, Evaluation=Multiscale (+ms)2023.0480
DINOv2Arch.=ViT-S/14, Evaluation=Multiscale (+ms)2023.0477.1
Ours-XL-tCategory=Ours, Backbone=SDXL2024.1075.89
iBOTArch.=ViT-L/16, Evaluation=Multiscale (+ms)2023.0474.5
Ours-XLCategory=Ours, Backbone=SDXL2024.1074.47
Meta PromptsCategory=SOTA2024.1071.94
Legacy-XLCategory=Baseline, Backbone=SDXL2024.1071.67
DINOv2Arch.=ViT-g/14, Evaluation=Linear2023.0471.3
FP BASELINEArchitecture=DeepLabv32024.0570.7
OpenCLIPArch.=ViT-G/14, Evaluation=Multiscale (+ms)2023.0470.3
DINOv2Arch.=ViT-L/14, Evaluation=Linear2023.0470.3
DINOv2Arch.=ViT-B/14, Evaluation=Linear2023.0469.4
BOLDArchitecture=DeepLabv3, Backbone=Boolean ResNet-182024.0567.4
DINOv2Arch.=ViT-S/14, Evaluation=Linear2023.0466.6
DINOArch.=ViT-B/8, Evaluation=Multiscale (+ms)2023.0466.2
iBOTArch.=ViT-L/16, Evaluation=Linear2023.0464.8
Ours-v1.5Category=Ours, Backbone=Stable Diffusion v1.52024.1064.1
Legacy-v1.5Category=Baseline, Backbone=Stable Diffusion v1.52024.1064.01
MAEArch.=ViT-H/14, Evaluation=Multiscale (+ms)2023.0461
OpenCLIPArch.=ViT-G/14, Evaluation=Linear2023.0460.3
MAEArch.=ViT-H/14, Evaluation=Linear2023.0458.4
BINARY DAD-NET2024.0558.1
DINOArch.=ViT-B/8, Evaluation=Linear2023.0456.9
VPDCategory=SOTA2024.1055.06
OursBackbone=SDv1.4, Resolution=512 x 5122026.0627.8
DiffSegBackbone=SDv1.4, Resolution=512 x 5122026.0622.2
EagleAX (Auxiliary image requirement)=true2026.0622.1
STEGOLD (Language Dependency)=true, AX (Auxiliary image requirement)=true2026.0621
ReCOLD (Language Dependency)=true, AX (Auxiliary image requirement)=true2026.0619.3
MaskCut2026.0618.7
DiffCutBackbone=SDv1.4, Resolution=512 x 5122026.0613.4
PiCIEUA (Unsupervised Adaptation)=true2026.0612.3
ProMerge2026.0611.8
MaskCLIPLD (Language Dependency)=true2026.0610
DINOv3 + Recursive-NCut2026.069
MDCUA (Unsupervised Adaptation)=true2026.067.1
IICUA (Unsupervised Adaptation)=true2026.066.4
CuVLERUA (Unsupervised Adaptation)=true2026.063.2