Semantic Segmentation on ACDC (Night)
66.4mIoUCoDA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| CoDABackbone=SegFormer, Scene Specificity=Scene-Agnostic2024.03 | 66.4 | — | — | — | |
| ILE-YOLOinference time (s)=0.3062024.08 | 65.2 | — | — | — | |
| ExpoMambainference time (s)=0.2972024.08 | 64.3 | — | — | — | |
| YOLOv3inference time (s)=0.2492024.08 | 63.3 | — | — | — | |
| MBLLENinference time (s)=0.3322024.08 | 63 | — | — | — | |
| IATinference time (s)=0.282024.08 | 62.1 | — | — | — | |
| DeepLPFinference time (s)=0.8072024.08 | 61.9 | — | — | — | |
| Zero-DCEinference time (s)=0.32024.08 | 61.9 | — | — | — | |
| MICBackbone=SegFormer, Scene Specificity=Scene-Agnostic2024.03 | 57.2 | — | — | — | |
| HRDABackbone=SegFormer, Scene Specificity=Scene-Agnostic2024.03 | 53.1 | — | — | — | |
| DINOv2 + CDTrain. data=ACDC Night2026.02 | 53.1 | — | — | — | |
| DINOv2Train. data=ACDC Night2026.02 | 52.8 | — | — | — | |
| SePiCoBackbone=SegFormer, Scene Specificity=Scene-Agnostic2024.03 | 50.5 | — | — | — | |
| DINOv2 + CDTrain. data=CityScapes2026.02 | 49 | — | — | — | |
| STABackbone=SegFormer, Scene Specificity=Scene-Agnostic2024.03 | 48.4 | — | — | — | |
| DINOv2Train. data=CityScapes2026.02 | 46.4 | — | — | — | |
| CLOUDSEncoder=ConvNext-L, zero-shot Domain Adaptation=true2023.12 | 45.1 | — | — | — | |
| DAFormerBackbone=SegFormer, Scene Specificity=Scene-Agnostic2024.03 | 44.7 | — | — | — | |
| CoCoOp+CAKIBase Model=SEEM-Large2026.05 | 34.9 | — | — | — | |
| CoCoOpBase Model=SEEM-Large2026.05 | 34.3 | — | — | — | |
| DiffTPT + HisTPTModel Scale=Large2024.10 | 33.6 | — | — | — | |
| TPT + HisTPTModel Scale=Large2024.10 | 33.4 | — | — | — | |
| CoOp+CAKIBase Model=SEEM-Large2026.05 | 33.3 | — | — | — | |
| HisTPTModel Scale=Large2024.10 | 33.1 | — | — | — | |
| CLOUDSEncoder=ResNet-101, zero-shot Domain Adaptation=true2023.12 | 33 | — | — | — | |
| TPTModel Scale=Large2024.10 | 32.4 | — | — | — | |
| CoOpBase Model=SEEM-Large2026.05 | 32.3 | — | — | — | |
| DiffTPTModel Scale=Large2024.10 | 32.2 | — | — | — | |
| SEEMModel Scale=Large2024.10 | 32 | — | — | — | |
| SEEM-Large2026.05 | 32 | — | — | — | |
| CLOUDSEncoder=ResNet-50, zero-shot Domain Adaptation=true2023.12 | 29.4 | — | — | — | |
| ULDASource Domain=Cityscapes, Prompt=driving at night, Backbone=CLIP-ResNet-50, Base Model=DeepLabv3+2024.04 | 25.4 | — | — | — | |
| CoCoOp+CAKIBase Model=SEEM-Tiny2026.05 | 25.1 | — | — | — | |
| PØDASource Domain=Cityscapes, Prompt=driving at night, Backbone=CLIP-ResNet-50, Base Model=DeepLabv3+2024.04 | 25.03 | — | — | — | |
| PØDAAdaptation setting=0-shot, Prompt=driving at night2024.10 | 25.03 | — | — | — | |
| PØDASource domain=Cityscapes (CS), Target prompt (P)=driving at night, Zero-shot domain adaptation=true2024.10 | 25.03 | — | — | — | |
| PODAEncoder=ResNet-50, zero-shot Domain Adaptation=true2023.12 | 25 | — | — | — | |
| PIDAAdaptation setting=1-shot, Target image usage=single target image2024.10 | 24.29 | — | — | — | |
| CoOp+CAKIBase Model=SEEM-Tiny2026.05 | 23.9 | — | — | — | |
| CoCoOpBase Model=SEEM-Tiny2026.05 | 23.7 | — | — | — | |
| TPT + HisTPTModel Scale=Tiny2024.10 | 22.9 | — | — | — | |
| DiffTPT + HisTPTModel Scale=Tiny2024.10 | 22.7 | — | — | — | |
| CoOpBase Model=SEEM-Tiny2026.05 | 22.5 | — | — | — | |
| HisTPTModel Scale=Tiny2024.10 | 22.1 | — | — | — | |
| DiffTPTModel Scale=Tiny2024.10 | 21.6 | — | — | — | |
| TPTModel Scale=Tiny2024.10 | 21.4 | — | — | — | |
| CLIPstylerSource Domain=Cityscapes, Prompt=driving at night, Backbone=CLIP-ResNet-50, Base Model=DeepLabv3+2024.04 | 21.38 | — | — | — | |
| CLIPstylerSource domain=Cityscapes (CS), Target prompt (P)=driving at night, Zero-shot domain adaptation=true2024.10 | 21.38 | — | — | — | |
| CLIPStylerzero-shot Domain Adaptation=true2023.12 | 21.3 | — | — | — | |
| SEEMModel Scale=Tiny2024.10 | 20.7 | — | — | — | |
| SEEM-Tiny2026.05 | 20.7 | — | — | — | |
| source-onlySource Domain=Cityscapes, Prompt=driving at night, Backbone=CLIP-ResNet-50, Base Model=DeepLabv3+2024.04 | 18.31 | — | — | — | |
| source-only2024.10 | 18.31 | — | — | — | |
| source-onlySource domain=Cityscapes (CS), Target prompt (P)=driving at night, Zero-shot domain adaptation=true2024.10 | 18.31 | — | — | — | |
| AdaINAdaptation setting=1-shot, Target image usage=single target image2024.10 | 18.15 | — | — | — | |
| PØDASource Dataset=CS, Shots=0-shot, Backbone=ResNet-50, Pre-training=CLIP, Segmentation Head=DeepLabv3+2024.10 | — | 18.31 | 25.03 | 6.72 | |
| SM-PPMSource Dataset=CS, Shots=1-shot, Backbone=ResNet-101, Pre-training=ImageNet, Segmentation Head=DeepLabv22024.10 | — | 13.07 | 14.6 | 1.53 |