Semantic Segmentation on BDD-100K (val)
66.28mIoUCausal-Tune
Evaluation results
| Method | Links | |
|---|---|---|
| Causal-TuneBackbone Category=VFM-based, Training Source Domain=Cityscapes2025.12 | 66.28 | |
| FADABackbone Category=VFM-based, Training Source Domain=Cityscapes2025.12 | 65.12 | |
| SETBackbone Category=VFM-based, Training Source Domain=Cityscapes2025.12 | 65.07 | |
| ReinBackbone Category=VFM-based, Training Source Domain=Cityscapes2025.12 | 63.54 | |
| FADABackbone Category=VFM-based, Training Source Domain=GTA52025.12 | 61.94 | |
| Causal-TuneBackbone Category=VFM-based, Training Source Domain=GTA52025.12 | 61.8 | |
| SETBackbone Category=VFM-based, Training Source Domain=GTA52025.12 | 61.64 | |
| ReinBackbone Category=VFM-based, Training Source Domain=GTA52025.12 | 60.4 | |
| CMFormerBackbone Category=Transformer-based, Training Source Domain=Cityscapes2025.12 | 59.27 | |
| HGFormerBackbone Category=Transformer-based, Training Source Domain=Cityscapes2025.12 | 53.4 | |
| DIRLBackbone Category=Resnet-based, Training Source Domain=Cityscapes2025.12 | 51.8 | |
| Ours (ISW)Backbone=ResNet-50, Output Stride=16, Training Dataset=Cityscapes2021.03 | 50.73 | |
| ISWBackbone Category=Resnet-based, Training Source Domain=Cityscapes2025.12 | 50.73 | |
| CMFormerBackbone Category=Transformer-based, Training Source Domain=GTA52025.12 | 49.91 | |
| IterNormBackbone=ResNet-50, Output Stride=16, Training Dataset=Cityscapes2021.03 | 49.23 | |
| ItenormBackbone Category=Resnet-based, Training Source Domain=Cityscapes2025.12 | 49.23 | |
| Ours (IRW)Backbone=ResNet-50, Output Stride=16, Training Dataset=Cityscapes2021.03 | 48.67 | |
| IBN-NetBackbone=ResNet-50, Output Stride=16, Training Dataset=Cityscapes2021.03 | 48.56 | |
| IBNBackbone Category=Resnet-based, Training Source Domain=Cityscapes2025.12 | 48.56 | |
| SWBackbone=ResNet-50, Output Stride=16, Training Dataset=Cityscapes2021.03 | 48.49 | |
| IWBackbone Category=Resnet-based, Training Source Domain=Cityscapes2025.12 | 48.49 | |
| Ours (IW)Backbone=ResNet-50, Output Stride=16, Training Dataset=Cityscapes2021.03 | 48.19 | |
| FAMixBackbone=RN101, Source Dataset=GTAV2023.11 | 46.4 | |
| FAMixBackbone=RN50, Source Dataset=GTAV2023.11 | 45.61 | |
| BaselineBackbone=ResNet-50, Output Stride=16, Training Dataset=Cityscapes2021.03 | 44.96 | |
| TLDRBackbone=RN101, Source Dataset=GTAV, Extra-data usage=true2023.11 | 44.88 | |
| DiGA (Our Distillation)Backbone=ResNet-101, Source Dataset=GTA52023.04 | 44.42 | |
| SHADEBackbone=ResNet-101, Source Dataset=GTA52023.04 | 43.66 | |
| SHADEBackbone=RN101, Source Dataset=GTAV, Full-data training=true2023.11 | 43.66 | |
| TLDRBackbone=RN50, Source Dataset=GTAV, Extra-data usage=true2023.11 | 42.58 | |
| WildNetBackbone=RN101, Source Dataset=GTAV, Extra-data usage=true2023.11 | 41.73 | |
| SAN-SAWBackbone=ResNet-101, Source Dataset=GTA52023.04 | 41.18 | |
| SAN & SAWBackbone=RN101, Source Dataset=GTAV2023.11 | 41.18 | |
| SPC-NetBackbone=RN50, Source Dataset=GTAV2023.11 | 40.46 | |
| ISW + AdvStyleBackbone=ResNet-101, Source Domain=GTAV2022.07 | 40.32 | |
| DPCLBackbone=RN50, Source Dataset=GTAV2023.11 | 40.21 | |
| IBN-Net + AdvStyleBackbone=ResNet-101, Source Domain=GTAV2022.07 | 39.96 | |
| Pin the MemoryBackbone=ResNet-101, Seg. model=DeepLabV22022.04 | 39.71 | |
| FSDRBackbone=ResNet-101, Seg. model=DeepLabV22022.04 | 39.66 | |
| SHADEBackbone=RN50, Source Dataset=GTAV2023.11 | 39.28 | |
| DIRLBackbone Category=Resnet-based, Training Source Domain=GTA52025.12 | 39.15 | |
| ISW + AdvStyleBackbone=ResNet-50, Source Domain=GTAV2022.07 | 38.59 | |
| ISWBackbone=ResNet-101, Source Dataset=GTA52023.04 | 38.53 | |
| WildNetBackbone=RN50, Source Dataset=GTAV, Extra-data usage=true2023.11 | 38.42 | |
| SiamDoGeBackbone=RN50, Source Dataset=GTAV2023.11 | 37.54 | |
| SAN & SAWBackbone=RN50, Source Dataset=GTAV2023.11 | 37.34 | |
| Ours-CBackbone=ResNet-50, Architecture=DeepLabV3+, Output Stride=16, Training Dataset=GTAV2023.04 | 37.03 | |
| Ours-DBackbone=ResNet-50, Architecture=DeepLabV3+, Output Stride=16, Training Dataset=GTAV2023.04 | 36.68 | |
| IBN-Net + AdvStyleBackbone=ResNet-50, Source Domain=GTAV2022.07 | 36.42 | |
| Baseline + AdvStyleBackbone=ResNet-101, Source Domain=GTAV2022.07 | 36.39 | |
| RobustNet* [4]Backbone=ResNet-50, Architecture=DeepLabV3+, Output Stride=16, Training Dataset=GTAV, Reproduced=true2023.04 | 35.61 | |
| NPBackbone=RN50, Source Dataset=GTAV2023.11 | 35.56 | |
| Baseline + AdvStyleBackbone=ResNet-50, Source Domain=GTAV2022.07 | 35.54 | |
| Ours (ISW)Backbone=ResNet-50, Output stride=16, Training Dataset=GTAV2021.03 | 35.2 | |
| RobustNetBackbone=ResNet-50, Seg. model=DeepLabV3+2022.04 | 35.2 | |
| ISWBackbone=ResNet-50, Source Domain=GTAV2022.07 | 35.2 | |
| RobustNet [4]Backbone=ResNet-50, Architecture=DeepLabV3+, Output Stride=16, Training Dataset=GTAV2023.04 | 35.2 | |
| RobustNetBackbone=RN50, Source Dataset=GTAV2023.11 | 35.2 | |
| ISWBackbone Category=Resnet-based, Training Source Domain=GTA52025.12 | 35.2 | |
| Ours-BBackbone=ResNet-50, Architecture=DeepLabV3+, Output Stride=16, Training Dataset=GTAV2023.04 | 34.95 | |
| Pin the MemoryBackbone=ResNet-50, Seg. model=DeepLabV3+2022.04 | 34.6 | |
| Pin the memoryBackbone=RN50, Source Dataset=GTAV2023.11 | 34.6 | |
| IBN-NetBackbone=ResNet-101, Source Domain=GTAV2022.07 | 34.21 | |
| Ours-ABackbone=ResNet-50, Architecture=DeepLabV3+, Output Stride=16, Training Dataset=GTAV2023.04 | 34.14 | |
| Ours (ISW)Backbone=ResNet-50, Output Stride=16, Training Set=GTAV + SYNTHIA2021.03 | 34.09 | |
| ISWBackbone=ResNet-101, Source Domain=GTAV2022.07 | 33.36 | |
| Ours (IRW)Backbone=ResNet-50, Output stride=16, Training Dataset=GTAV2021.03 | 33.18 | |
| MRFPBackbone=MobileNetv2, Framework=DeepLabV3+, Training Source=GTAV2023.11 | 33.03 | |
| Baseline + AdvStyleBackbone=MobileNetV2, Source Domain=GTAV2022.07 | 33.01 | |
| IterNormBackbone=ResNet-50, Output stride=16, Training Dataset=GTAV2021.03 | 32.7 | |
| IterNorm [26]Backbone=ResNet-50, Architecture=DeepLabV3+, Output Stride=16, Training Dataset=GTAV2023.04 | 32.7 | |
| ItenormBackbone Category=Resnet-based, Training Source Domain=GTA52025.12 | 32.7 | |
| Ours (IW)Backbone=ResNet-50, Output stride=16, Training Dataset=GTAV2021.03 | 32.67 | |
| IBN-NetBackbone=ResNet-50, Output stride=16, Training Dataset=GTAV2021.03 | 32.3 | |
| IBN-NetBackbone=ResNet-50, Seg. model=DeepLabV3+2022.04 | 32.3 | |
| IBN-NetBackbone=ResNet-50, Source Domain=GTAV2022.07 | 32.3 | |
| IBN-Net [49]Backbone=ResNet-50, Architecture=DeepLabV3+, Output Stride=16, Training Dataset=GTAV2023.04 | 32.3 | |
| IBNBackbone Category=Resnet-based, Training Source Domain=GTA52025.12 | 32.3 | |
| IBN-NetBackbone=ResNet-50, Output Stride=16, Training Set=GTAV + SYNTHIA2021.03 | 32.18 | |
| DRPCBackbone=ResNet-50, Seg. model=FCN-8s2022.04 | 32.1 | |
| MLDGBackbone=ResNet-50, Seg. model=DeepLabV3+2022.04 | 32.1 | |
| ISW + AdvStyleBackbone=MobileNetV2, Source Domain=GTAV2022.07 | 31.84 | |
| IBN-Net + AdvStyleBackbone=MobileNetV2, Source Domain=GTAV2022.07 | 31.55 | |
| RandConv* [70]Backbone=ResNet-50, Architecture=DeepLabV3+, Output Stride=16, Training Dataset=GTAV, Reproduced=true2023.04 | 30.92 | |
| BaselineBackbone=ResNet-101, Source Domain=GTAV2022.07 | 30.77 | |
| ISWBackbone=MobileNetV2, Source Domain=GTAV2022.07 | 30.05 | |
| ISWBackbone=MobileNetv2, Framework=DeepLabV3+, Training Source=GTAV2023.11 | 30.05 | |
| IBN-NetBackbone=MobileNetV2, Source Domain=GTAV2022.07 | 27.66 | |
| IBN-NetBackbone=MobileNetv2, Framework=DeepLabV3+, Training Source=GTAV2023.11 | 27.66 | |
| SWBackbone=ResNet-50, Output stride=16, Training Dataset=GTAV2021.03 | 27.48 | |
| SW [50]Backbone=ResNet-50, Architecture=DeepLabV3+, Output Stride=16, Training Dataset=GTAV2023.04 | 27.48 | |
| IWBackbone Category=Resnet-based, Training Source Domain=GTA52025.12 | 27.48 | |
| Baseline*Backbone=ResNet-50, Architecture=DeepLabV3+, Output Stride=16, Training Dataset=GTAV, Reproduced=true2023.04 | 27.18 | |
| BaselineBackbone=MobileNetv2, Framework=DeepLabV3+, Training Source=GTAV2023.11 | 26.76 | |
| BaselineBackbone=ResNet-50, Seg. model=FCN-8s2022.04 | 26.7 | |
| BaselineBackbone=ResNet-50, Seg. model=DeepLabV3+, Comparison=with MLDG2022.04 | 26.7 | |
| BaselineBackbone=MobileNetV2, Source Domain=GTAV2022.07 | 25.73 | |
| BaselineBackbone=ResNet-50, Output stride=16, Training Dataset=GTAV2021.03 | 25.14 | |
| BaselineBackbone=ResNet-50, Source Domain=GTAV2022.07 | 25.14 | |
| Baseline [4]Backbone=ResNet-50, Architecture=DeepLabV3+, Output Stride=16, Training Dataset=GTAV2023.04 | 25.14 |