Semantic Segmentation on GTA to Cityscapes (val)
99.1Road AccuracyHoyer et al.
Evaluation results
| Method | Links | ||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Hoyer et al.Architecture=Transformer2025.12 | 99.1 | 74.8 | 95.2 | 61 | 53.1 | 59.8 | 70.7 | 68.6 | 96 | 63.4 | 98.7 | 84.2 | 69.9 | 95.5 | 88.9 | 84.1 | 74 | 70 | 69.7 | — | — | — | 77.8 | 14.2 | |
| Cui et al.Architecture=Transformer2025.12 | 98.8 | 77.7 | 94.1 | 71.4 | 56.8 | 63.7 | 77.2 | 71.2 | 95.3 | 63.5 | 99 | 84.1 | 72 | 94.6 | 90.1 | 89.2 | 68.5 | 76.4 | 80.2 | — | — | — | 80.2 | 12.6 | |
| DAFormer (baseline)Architecture=Transformer2025.12 | 98.3 | 75.4 | 94.9 | 53.8 | 57.8 | 53.6 | 70.2 | 59.7 | 96.2 | 57.6 | 99.1 | 83.5 | 67.6 | 95.5 | 87.1 | 84.3 | 73.8 | 74.5 | 77.5 | — | — | — | 76.9 | 15.3 | |
| MIC w/ MRMmodule=MRM2025.09 | 98.3 | 80.4 | 92.6 | 62.7 | 57 | 62.3 | 69.1 | 74.3 | 91.8 | 53.5 | 94.7 | 81.1 | 56.6 | 94.1 | 87.2 | 91.6 | 85.4 | 66.3 | 71.3 | 77.5 | — | — | — | — | |
| HALOBudget=5%, Backbone=SegFormer B42023.06 | 98.2 | 85.4 | 92.5 | 62.5 | 61.6 | 58.3 | 67.7 | 74.9 | 92.2 | 65.1 | 94.7 | 79.9 | 60.8 | 94.6 | 84.1 | 85.4 | 83.6 | 61.2 | 75.5 | 77.8 | — | — | — | — | |
| Menon et al.Architecture=Transformer2025.12 | 97.9 | 79.9 | 95 | 64.4 | 64.7 | 66.6 | 74.3 | 72.6 | 96.5 | 66.1 | 99.7 | 87.9 | 72.5 | 96.3 | 90.5 | 91.3 | 85.7 | 74.3 | 72.4 | — | — | — | 81.5 | 12.2 | |
| MIC + ECOCSegArch.=Transformer2025.12 | 97.9 | 81.4 | 91.9 | 62.2 | 54.3 | 64.2 | 67.4 | 76.1 | 92.9 | 54.4 | 94.2 | 82.1 | 53 | 95.2 | 89.6 | 90.8 | 82.3 | 61.9 | 69.4 | 76.9 | — | — | — | — | |
| HALOBudget=5%, Backbone=DeepLab-v3+2023.06 | 97.6 | 81 | 91.4 | 53.7 | 54.9 | 56.7 | 62.9 | 72.1 | 91.4 | 60.5 | 94.1 | 78 | 57.3 | 94 | 81.4 | 84.7 | 70.1 | 60 | 73.3 | 74.5 | — | — | — | — | |
| Hyper. Supervised DABudget=100%, Backbone=DeepLab-v3+2023.06 | 97.6 | 81.2 | 90.7 | 49.9 | 53.2 | 53.5 | 58 | 67.2 | 91 | 59.1 | 93.9 | 74.2 | 52.6 | 93.1 | 76.4 | 81 | 67 | 55 | 70.8 | 71.9 | — | — | — | — | |
| HALOBudget=2.2%, Backbone=DeepLab-v22023.06 | 97.5 | 79.9 | 90.2 | 55.6 | 51.5 | 45.3 | 56.2 | 66.2 | 90.2 | 58.6 | 92.8 | 73.3 | 53.5 | 92.6 | 76.9 | 76.2 | 64.2 | 55.2 | 70.1 | 70.8 | — | — | — | — | |
| BLDAArchitecture=Transformer2025.12 | 97.5 | 79.4 | 93.4 | 65.5 | 68 | 63.9 | 77.2 | 75.5 | 94.9 | 68.1 | 99 | 85.2 | 71.4 | 95.2 | 89.2 | 87.6 | 80.8 | 77 | 83.6 | — | — | — | 81.7 | 10.9 | |
| QuadMix2025.09 | 97.5 | 80.9 | 91.6 | 62.3 | 57.6 | 58.2 | 64.5 | 71.2 | 91.7 | 52.3 | 94.3 | 80 | 55.9 | 94.6 | 86.3 | 90.5 | 82.3 | 65.1 | 68.1 | 76.1 | — | — | — | — | |
| Eucl. Supervised DABudget=100%, Backbone=DeepLab-v3+2023.06 | 97.4 | 77.9 | 91.1 | 54.9 | 53.7 | 51.9 | 57.9 | 64.7 | 91.1 | 57.8 | 93.2 | 74.7 | 54.8 | 93.6 | 76.4 | 79.3 | 67.8 | 55.6 | 71.3 | 71.9 | — | — | — | — | |
| MIC*Model trained without terrain, truck, and train on the source domain=true2024.08 | 97.4 | 80.4 | 91.8 | 61.2 | 55.2 | 60 | 65.5 | 73 | 89.3 | — | 93.1 | 79.5 | 52.8 | 90.6 | — | 68.2 | — | 64.7 | 67 | — | 74.4 | 62.6 | — | — | |
| MICModel trained without terrain, truck, and train on the source domain=false2024.08 | 97.4 | 80.1 | 91.7 | 61.2 | 56.9 | 59.7 | 66 | 71.3 | 91.7 | 51.4 | 94.3 | 79.8 | 56.1 | 94.6 | 85.4 | 90.3 | 80.4 | 64.5 | 68.5 | — | 76.5 | 75.9 | — | — | |
| MICArch.=Transformer2025.12 | 97.4 | 80.1 | 91.7 | 61.2 | 56.9 | 59.7 | 66 | 71.3 | 91.7 | 51.4 | 94.3 | 79.8 | 56.1 | 94.6 | 85.4 | 90.3 | 80.4 | 64.5 | 68.5 | 75.9 | — | — | — | — | |
| MIC2025.09 | 97.4 | 80.1 | 91.7 | 61.2 | 56.9 | 59.7 | 66 | 71.3 | 91.7 | 51.4 | 94.3 | 79.8 | 56.1 | 94.6 | 85.4 | 90.3 | 80.4 | 64.5 | 68.5 | 75.9 | — | — | — | — | |
| MIC*+CSIModel trained without terrain, truck, and train on the source domain=true2024.08 | 97.3 | 80 | 91.8 | 62.1 | 54.4 | 58.9 | 65.2 | 72.4 | 90 | 39.2 | 92.8 | 79.5 | 54 | 93 | 68 | 84.1 | 74.8 | 64.5 | 67.4 | — | 75.5 | 73.1 | — | — | |
| CDACArch.=Transformer2025.12 | 97.1 | 78.7 | 91.8 | 59.6 | 57.1 | 59.1 | 66.1 | 72.2 | 91.8 | 53.1 | 94.5 | 79.4 | 51.6 | 94.6 | 84.9 | 87.8 | 78.7 | 64.9 | 67.6 | 75.3 | — | — | — | — | |
| D2ADABudget=5%, Backbone=DeepLab-v3+2023.06 | 97 | 77.8 | 90 | 46 | 55 | 52.7 | 58.7 | 65.8 | 90.4 | 58.9 | 92.1 | 75.7 | 54.4 | 92.3 | 69 | 78 | 68.5 | 59.1 | 72.3 | 71.3 | — | — | — | — | |
| RIPUBudget=5%, Backbone=DeepLab-v3+2023.06 | 97 | 77.3 | 90.4 | 54.6 | 53.2 | 47.7 | 55.9 | 64.1 | 90.2 | 59.2 | 93.2 | 75 | 54.8 | 92.7 | 73 | 79.7 | 68.9 | 55.5 | 70.3 | 71.2 | — | — | — | — | |
| HRDA w/ MRMmodule=MRM2025.09 | 97 | 78.3 | 90.9 | 59.5 | 52.8 | 60.4 | 66.6 | 72.4 | 91.9 | 51.8 | 94.3 | 79.3 | 55.5 | 94.4 | 84.4 | 87.9 | 82.7 | 65.4 | 68.2 | 75.4 | — | — | — | — | |
| DAFormer w/ MRMmodule=MRM2025.09 | 96.8 | 76.3 | 89.5 | 55.7 | 50.3 | 50.9 | 58.2 | 62.2 | 90 | 50.1 | 91.1 | 73.8 | 48.2 | 92.3 | 77.3 | 80.4 | 71.6 | 56.3 | 64 | 70.3 | — | — | — | — | |
| ADFormerArch.=Transformer2025.12 | 96.7 | 75.1 | 88.8 | 57.5 | 45.9 | 45.6 | 55.4 | 59.8 | 90.2 | 45.6 | 92.1 | 70.8 | 43 | 91 | 78.9 | 79.3 | 68.7 | 52.7 | 65 | 69.2 | — | — | — | — | |
| DAFormer + ECOCSegArch.=Transformer2025.12 | 96.7 | 75.6 | 89.4 | 54 | 51.4 | 55.1 | 59.4 | 61.9 | 90.1 | 46.6 | 90 | 71.5 | 42.4 | 92.8 | 79.7 | 85.4 | 79.1 | 60 | 58.2 | 70.5 | — | — | — | — | |
| LabORBudget=2.2%, Backbone=DeepLab-v22023.06 | 96.6 | 77 | 89.6 | 47.8 | 50.7 | 48 | 56.6 | 63.5 | 89.5 | 57.8 | 91.6 | 72 | 47.3 | 91.7 | 62.1 | 61.9 | 48.9 | 47.9 | 65.3 | 66.6 | — | — | — | — | |
| RIPUBudget=2.2%, Backbone=DeepLab-v22023.06 | 96.5 | 74.1 | 89.7 | 53.1 | 51 | 43.8 | 53.4 | 62.2 | 90 | 57.6 | 92.6 | 73 | 53 | 92.8 | 73.8 | 78.5 | 62 | 55.6 | 70 | 69.6 | — | — | — | — | |
| GANDA2025.09 | 96.5 | 74.8 | 91.4 | 61.7 | 57.3 | 59.2 | 65.4 | 68.8 | 91.5 | 49.9 | 94.7 | 79.6 | 54.8 | 94.1 | 81.3 | 86.8 | 74.6 | 64.8 | 68.2 | 74.5 | — | — | — | — | |
| HRDABackbone=MiT-B52023.04 | 96.4 | 74.4 | 91 | 61.6 | 51.5 | 57.1 | 63.9 | 69.3 | 91.3 | 48.4 | 94.2 | 79 | 52.9 | 93.9 | 84.1 | 85.7 | 75.9 | 63.9 | 67.5 | 73.8 | — | — | — | — | |
| HRDA2025.09 | 96.4 | 74.4 | 91 | 61.6 | 51.5 | 57.1 | 63.9 | 69.3 | 91.3 | 48.4 | 94.2 | 79 | 52.9 | 93.9 | 84.1 | 85.7 | 75.9 | 63.9 | 67.5 | 73.8 | — | — | — | — | |
| PiPa2025.09 | 96.1 | 72 | 90.3 | 56.6 | 52 | 55.1 | 61.8 | 63.7 | 90.8 | 52.6 | 93.6 | 74.3 | 43.6 | 93.5 | 78.4 | 84.2 | 77.3 | 59.9 | 66.7 | 71.7 | — | — | — | — | |
| DAFormerBackbone=MiT-B52023.04 | 95.7 | 70.2 | 89.4 | 53.5 | 48.1 | 49.6 | 55.8 | 59.4 | 89.9 | 47.9 | 92.5 | 72.2 | 44.7 | 92.3 | 74.5 | 78.2 | 65.1 | 55.9 | 61.8 | 68.3 | — | — | — | — | |
| DAFormerModel trained without terrain, truck, and train on the source domain=false2024.08 | 95.7 | 70.2 | 89.4 | 53.5 | 48.1 | 49.6 | 55.8 | 59.4 | 89.9 | 47.9 | 92.5 | 72.2 | 44.7 | 92.3 | 74.5 | 78.2 | 65.1 | 55.9 | 61.8 | — | 69.3 | 68.3 | — | — | |
| DAFormerArch.=Transformer2025.12 | 95.7 | 70.2 | 89.4 | 53.5 | 48.1 | 49.6 | 55.8 | 59.4 | 89.9 | 47.9 | 92.5 | 72.2 | 44.7 | 92.3 | 74.5 | 78.2 | 65.1 | 55.9 | 61.8 | 68.3 | — | — | — | — | |
| DAFormer2025.09 | 95.7 | 70.2 | 89.4 | 53.5 | 48.1 | 49.6 | 55.8 | 59.4 | 89.9 | 47.9 | 92.5 | 72.2 | 44.7 | 92.3 | 74.5 | 78.2 | 65.1 | 55.9 | 61.8 | 68.3 | — | — | — | — | |
| DACS + ECOCSegArch.=CNN2025.12 | 95.6 | 71.8 | 90.2 | 37.8 | 31.4 | 44.8 | 50.8 | 58.8 | 90.4 | 50.3 | 91.3 | 68.6 | 23.5 | 91.2 | 49.8 | 55.4 | 8.8 | 15.2 | 9.8 | 54.5 | — | — | — | — | |
| DAFormer*Model trained without terrain, truck, and train on the source domain=true2024.08 | 95.5 | 68.5 | 89.5 | 53.2 | 47.3 | 50.2 | 55.4 | 60.1 | 87.1 | — | 91 | 72.1 | 44.9 | 89 | — | 63.8 | — | 53.7 | 62.1 | — | 67.7 | 57 | — | — | |
| MADABudget=5%, Backbone=DeepLab-v3+2023.06 | 95.1 | 69.8 | 88.5 | 43.3 | 48.7 | 45.7 | 53.3 | 59.2 | 89.1 | 46.7 | 91.5 | 73.9 | 50.1 | 91.2 | 60.6 | 56.9 | 48.4 | 51.6 | 68.7 | 64.9 | — | — | — | — | |
| CorDABackbone=ResNet-1012023.04 | 94.7 | 63.1 | 87.6 | 30.7 | 40.6 | 40.2 | 47.8 | 51.6 | 87.6 | 47 | 89.7 | 66.7 | 35.9 | 90.2 | 48.9 | 57.5 | 0 | 39.8 | 56 | 56.6 | — | — | — | — | |
| TransDAArch.=Transformer2025.12 | 94.7 | 64.2 | 89.2 | 48.1 | 45.8 | 50.1 | 60.2 | 40.8 | 90.4 | 50.2 | 93.7 | 76.7 | 47.6 | 92.5 | 56.8 | 60.1 | 47.6 | 49.6 | 55.4 | 63.9 | — | — | — | — | |
| DACS w/ MRMmodule=MRM2025.09 | 94.7 | 68.3 | 87.6 | 38.1 | 27.4 | 42 | 51.9 | 59.2 | 86.9 | 45 | 87.7 | 66.3 | 32.3 | 89.6 | 53.9 | 56.6 | 0 | 33.2 | 42.1 | 55.9 | — | — | — | — | |
| DAFormer*+CSIModel trained without terrain, truck, and train on the source domain=true2024.08 | 93.2 | 60.7 | 89.4 | 53.4 | 48 | 50.6 | 55.4 | 60.6 | 88.5 | 41.5 | 90.6 | 71.7 | 44 | 90.6 | 42.1 | 73.5 | 63 | 55.3 | 61.2 | — | 67.9 | 64.9 | — | — | |
| CPSLArch.=CNN2025.12 | 92.3 | 59.5 | 84.9 | 45.7 | 29.7 | 52.8 | 61.5 | 59.5 | 87.9 | 41.6 | 85 | 73 | 35.5 | 90.4 | 48.7 | 73.9 | 26.3 | 53.8 | 53.9 | 60.8 | — | — | — | — | |
| AADABudget=5%, Backbone=DeepLab-v3+2023.06 | 92.2 | 59.9 | 87.3 | 36.4 | 45.7 | 46.1 | 50.6 | 59.5 | 88.3 | 44 | 90.2 | 69.7 | 38.2 | 90 | 55.3 | 45.1 | 32 | 32.6 | 62.9 | 59.3 | — | — | — | — | |
| Lin et al.Architecture=Transformer2025.12 | 91.7 | 63.3 | 94.6 | 66.4 | 50.2 | 54.4 | 63.9 | 55.1 | 95.6 | 66.9 | 98.6 | 84.9 | 58.5 | 92.6 | 84.4 | 81.9 | 70 | 68.8 | 68.2 | — | — | — | 74.2 | 15.2 | |
| DACSBackbone=ResNet-1012023.04 | 89.9 | 39.7 | 87.9 | 30.7 | 39.5 | 38.5 | 46.4 | 52.8 | 88 | 44 | 88.8 | 67.2 | 35.8 | 84.5 | 45.7 | 50.2 | 0 | 27.3 | 34 | 52.1 | — | — | — | — | |
| DACSArch.=CNN2025.12 | 89.9 | 39.7 | 87.9 | 39.7 | 39.5 | 38.5 | 46.4 | 52.8 | 88 | 44 | 88.8 | 67.2 | 35.8 | 84.5 | 45.7 | 50.2 | 0.2 | 27.3 | 34 | 52.1 | — | — | — | — | |
| DACS2025.09 | 89.9 | 39.7 | 87.9 | 30.7 | 39.5 | 38.5 | 46.4 | 52.8 | 88 | 44 | 88.8 | 67.2 | 35.8 | 84.5 | 45.7 | 50.2 | 0 | 27.3 | 34 | 52.1 | — | — | — | — | |
| ADVENTBackbone=ResNet-1012023.04 | 89.4 | 33.1 | 81 | 26.6 | 26.8 | 27.2 | 33.5 | 24.7 | 83.9 | 36.7 | 78.8 | 58.7 | 30.5 | 84.8 | 38.5 | 44.5 | 1.7 | 31.6 | 32.4 | 45.5 | — | — | — | — | |
| DACSBackbone=MiT-B5, additional daytime/clear-weather reference images=true2023.04 | 88.9 | 50 | 88.4 | 46.4 | 43.9 | 43.1 | 53.4 | 54.8 | 89.9 | 51.2 | 92.8 | 64.2 | 9.4 | 91.4 | 77.3 | 63.3 | 0 | 47.4 | 49.8 | 58.2 | — | — | — | — | |
| ProDABackbone=ResNet-1012023.04 | 87.8 | 56 | 79.7 | 46.3 | 44.8 | 45.6 | 53.5 | 53.5 | 88.6 | 45.2 | 82.1 | 70.7 | 39.2 | 88.8 | 45.5 | 59.4 | 1 | 48.9 | 56.4 | 57.5 | — | — | — | — | |
| ProDAArch.=CNN2025.12 | 87.8 | 56 | 79.7 | 46.3 | 44.8 | 45.6 | 53.5 | 53.5 | 88.6 | 45.2 | 82.1 | 70.7 | 39.2 | 88.8 | 45.5 | 50.4 | 1 | 48.9 | 56.4 | 57.5 | — | — | — | — | |
| DecoupleNetBackbone=ResNet-1012023.04 | 87.6 | 49.3 | 87.2 | 42.5 | 41.6 | 46.6 | 57.4 | 44 | 89 | 43.9 | 90.6 | 73 | 43.8 | 88.1 | 32.9 | 53.7 | 44.3 | 49.8 | 57.2 | 59.1 | — | — | — | — | |
| AdaptSegBackbone=ResNet-1012023.04 | 86.5 | 25.9 | 79.8 | 22.1 | 20 | 23.6 | 33.1 | 21.8 | 81.8 | 25.9 | 75.9 | 57.3 | 26.2 | 76.3 | 29.8 | 32.1 | 7.2 | 29.5 | 32.5 | 41.4 | — | — | — | — | |
| AdaptSegBackbone=MiT-B5, additional daytime/clear-weather reference images=true2023.04 | 85.2 | 20.4 | 85.5 | 38.2 | 30.9 | 34.5 | 43 | 26.2 | 87.4 | 40.3 | 86.4 | 63.6 | 23.7 | 88.6 | 48.5 | 50.6 | 5.8 | 33.1 | 16.2 | 47.8 | — | — | — | — |