Semantic Segmentation on Cityscapes adaptation from Synthia 1.0 (val)
82Person IoUSelf-Training + CB
Evaluation results
| Method | Links | |||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Self-Training + CBBackbone=VGG-16, Appr.=ST2018.11 | 82 | 32.4 | 69.6 | 28.7 | 69.5 | 12.1 | 0.1 | 25.4 | 11.9 | 13.6 | — | — | 81.9 | 49.1 | 14.5 | 66 | 6.6 | — | — | — | 35.4 | |
| MIC2023.11 | 81 | 74 | 86.6 | 50.5 | 89.3 | 47.9 | 7.8 | 59.4 | — | — | 87.1 | 94.6 | 58.9 | 90.1 | 61.9 | 67.1 | 64.3 | 66.7 | 63.4 | — | — | |
| Depth-aware Framework (Ours)Training with depth data=true2023.11 | 80.9 | 75.9 | 93.4 | 63.1 | 89.8 | 51.1 | 9.1 | 61.4 | — | — | 88 | 94.5 | 56.6 | 90.9 | 68.5 | 63.7 | 66.6 | 66.9 | 64 | — | — | |
| AdvEnt + MinEntBackbone=ResNet-101, Appr.=A+E2018.11 | 80.4 | 41.2 | 85.6 | 42.2 | 79.7 | 8.7 | 0.4 | 25.9 | 5.4 | 8.1 | — | — | 84.1 | 57.9 | 23.8 | 73.3 | 36.4 | — | — | — | 48 | |
| AdvEntBackbone=ResNet-101, Appr.=Adv2018.11 | 80.1 | 40.8 | 87 | 44.1 | 79.7 | 9.6 | 0.6 | 24.3 | 4.8 | 7.2 | — | — | 83.6 | 56.4 | 23.7 | 72.7 | 32.6 | — | — | — | 47.6 | |
| HRDA2023.11 | 79.4 | 72.4 | 85.2 | 47.7 | 88.8 | 49.5 | 4.8 | 57.2 | — | — | 85.3 | 92.9 | 52.8 | 89 | 64.7 | 63.9 | 64.9 | 65.7 | 60.9 | — | — | |
| Adapt-SegMap*Backbone=ResNet-101, Appr.=Adv, Note=Retrained by current paper authors2018.11 | 79.1 | 29.9 | 81.7 | 39.1 | 78.4 | 11.1 | 0.3 | 25.8 | 6.8 | 9 | — | — | 80.8 | 54.8 | 21 | 66.8 | 34.7 | — | — | — | 45.8 | |
| Adapt-SegMapBackbone=ResNet-101, Appr.=Adv2018.11 | 77.9 | — | — | — | — | — | — | — | — | 4.7 | — | — | 82.5 | 54.3 | 21 | 72.3 | 32.2 | — | — | — | 46.7 | |
| MinEntBackbone=ResNet-101, Appr.=Ent2018.11 | 76.7 | 38.1 | 73.5 | 29.2 | 77.1 | 7.7 | 0.2 | 27 | 7.1 | 11.4 | — | — | 82.1 | 57.2 | 21.3 | 69.4 | 29.2 | — | — | — | 44.2 | |
| AdvEnt + CPBackbone=VGG-16, Appr.=Adv2018.11 | 74.9 | 31.4 | 67.9 | 29.4 | 71.9 | 6.3 | 0.3 | 19.9 | 0.6 | 2.6 | — | — | 74.9 | 35.4 | 9.6 | 67.8 | 21.4 | — | — | — | 36.6 | |
| MinEnt + CPBackbone=VGG-16, Appr.=Ent2018.11 | 74.4 | 30.4 | 45.9 | 19.6 | 65.8 | 5.3 | 0.2 | 20.7 | 2.1 | 8.2 | — | — | 76.7 | 47.5 | 12.2 | 71.1 | 22.8 | — | — | — | 35.4 | |
| MinEntBackbone=VGG-16, Appr.=Ent2018.11 | 73.9 | 27.5 | 37.8 | 18.2 | 65.8 | 2 | 0 | 15.5 | 0 | 7.6 | — | — | 45.7 | 11.3 | 66.6 | 13.3 | 1.5 | — | — | — | 32.5 | |
| DAFormer2023.11 | 73.2 | 67.4 | 84.5 | 40.7 | 88.4 | 41.5 | 6.5 | 50 | — | — | 86 | 89.8 | 48.2 | 87.2 | 53.2 | 53.9 | 61.7 | 55 | 54.6 | — | — | |
| Adapt-SegMapBackbone=VGG-16, Appr.=Adv2018.11 | 72.6 | — | 78.9 | 29.2 | 75.5 | — | — | — | — | 4.8 | — | — | 76.7 | 43.4 | 8.8 | 71.1 | 16 | — | — | — | 37.6 | |
| Self-TrainingBackbone=VGG-16, Appr.=ST2018.11 | 72.2 | 23.9 | 0.2 | 14.5 | 53.8 | 1.6 | 0 | 18.9 | 0.9 | 7.8 | — | — | 80.3 | 48.1 | 6.3 | 67.7 | 4.7 | — | — | — | 27.8 | |
| CAMix2023.11 | 72 | 69.2 | 87.4 | 47.5 | 88.8 | — | — | — | — | — | 87 | 91.7 | 49.3 | 86.9 | 57 | 57.5 | 63.6 | 55.2 | 55.4 | — | — | |
| CorDATraining with depth data=true2023.11 | 69.7 | 62.8 | 93.3 | 61.6 | 85.3 | 19.6 | 5.1 | 37.8 | — | — | 84.9 | 90.4 | 41.8 | 85.6 | 38.4 | 32.6 | 53.9 | 36.6 | 42.8 | — | — | |
| SISC+PWLAppr.=ST2019.09 | 68.6 | 45.2 | 59.2 | 30.2 | 68.5 | 22.9 | 1 | 36.2 | — | — | 86.2 | 75.4 | 27.7 | 82.7 | 26.3 | 24.3 | 52.7 | 32.7 | 28.3 | — | 51 | |
| Hong et al.Base Model=FCN8s2019.03 | 68.5 | 41.2 | 85 | 25.8 | 73.5 | 3.4 | 3 | 31.5 | — | — | 67.4 | 69.4 | 25 | 76.5 | 41.6 | 17.9 | 29.5 | 19.5 | 21.3 | — | — | |
| CBSTAppr.=ST2019.09 | 67.2 | 42.5 | 53.6 | 23.7 | 75 | 12.5 | 0.3 | 36.4 | — | — | 84.8 | 74.7 | 17.5 | 84.5 | 28.4 | 15.2 | 55.8 | 23.5 | 26.3 | — | 48.4 | |
| SP-Adv2023.11 | 66.5 | 48.3 | 84.8 | 35.8 | 78.6 | — | — | — | — | — | 80.5 | 82 | 22.7 | 74.3 | 34.1 | 19.2 | 27.3 | 6.2 | 15.6 | — | — | |
| IAST2023.11 | 65.5 | 49.8 | 81.9 | 41.5 | 83.3 | 17.7 | 4.6 | 32.3 | — | — | 83.4 | 85 | 30.8 | 86.5 | 38.2 | 33.1 | 52.7 | 30.9 | 28.8 | — | — | |
| SISCAppr.=ST2019.09 | 65.3 | 44.4 | 73.7 | 34.4 | 78.7 | 13.7 | 2.9 | 36.6 | — | — | 86.1 | 76.8 | 20.5 | 81.7 | 31.4 | 13.9 | 47.3 | 28.2 | 22.3 | — | 50.8 | |
| PyCDABackbone=ResNet-1012019.08 | 64.1 | 46.7 | 75.5 | 30.9 | 83.3 | 20.8 | 0.7 | 32.7 | 27.3 | 33.5 | 84.7 | 85 | 25.4 | 85 | 45.2 | 21.2 | 32 | — | — | — | 53.3 | |
| Uncertainty2023.11 | 63 | 54.9 | 87.6 | 41.9 | 83.1 | 14.7 | 1.7 | 36.2 | — | — | 81.6 | 80.6 | 21.8 | 86.2 | 40.7 | 23.6 | 53.1 | 31.3 | 19.9 | — | — | |
| MRNet2023.11 | 61.3 | 50.2 | 82 | 36.5 | 80.4 | 4.2 | 0.4 | 33.7 | — | — | 81.1 | 80.8 | 21.7 | 84.4 | 32.4 | 14.8 | 45.7 | 18 | 13.4 | — | — | |
| MRKLD2023.11 | 60.8 | 50.1 | 67.7 | 32.2 | 73.9 | 10.7 | 1.6 | 37.4 | — | — | 80.8 | 80.5 | 29.1 | 82.8 | 25 | 19.4 | 45.3 | 22.2 | 31.2 | — | — | |
| CBST2023.11 | 60.6 | 48.9 | 68 | 29.9 | 76.3 | 10.8 | 1.4 | 33.9 | — | — | 77.6 | 78.3 | 28.3 | 81.6 | 23.5 | 18.8 | 39.8 | 22.8 | 29.5 | — | — | |
| DT2023.11 | 59.9 | 52.1 | 83 | 44 | 80.3 | — | — | — | — | — | 80.5 | 81.8 | 33.1 | 70.2 | 37.3 | 28.5 | 45.8 | 17.1 | 15.8 | — | — | |
| MaxSquare2023.11 | 58.5 | 45.8 | 77.4 | 34 | 78.7 | 5.6 | 0.2 | 27.7 | — | — | 80.7 | 83.2 | 20.5 | 74.1 | 32.1 | 11 | 29.9 | 5.8 | 9.8 | — | — | |
| SIBAN2023.11 | 58.3 | 46.3 | 82.5 | 24 | 79.4 | — | — | — | — | — | 79.2 | 82.8 | 18 | 79.3 | 25.3 | 17.6 | 25.9 | 16.5 | 12.7 | — | — | |
| ADVENTBackbone=ResNet-1012019.08 | 57.9 | 41.2 | 85.6 | 42.2 | 79.7 | 8.7 | 0.4 | 25.9 | 5.4 | 8.1 | 80.4 | 84.1 | 23.8 | 73.3 | 36.4 | 14.2 | 33 | — | — | — | 48 | |
| AdvEnt2023.11 | 57.9 | 48 | 85.6 | 42.2 | 79.7 | 8.7 | 0.4 | 25.9 | — | — | 80.4 | 84.1 | 23.8 | 73.3 | 36.4 | 14.2 | 33 | 5.4 | 8.1 | — | — | |
| MinEntAppr.=ST2019.09 | 57.2 | 38.1 | 73.5 | 29.2 | 77.1 | 7.7 | 0.2 | 27 | — | — | 76.7 | 82.1 | 21.3 | 69.4 | 29.2 | 12.9 | 27.9 | 7.1 | 11.4 | — | 44.2 | |
| CCM2023.11 | 56.8 | 52.9 | 79.6 | 36.4 | 80.6 | 13.3 | 0.3 | 25.5 | — | — | 81.8 | 77.4 | 25.9 | 80.7 | 45.3 | 29.9 | 52 | 22.4 | 14.9 | — | — | |
| ResNet-38Appr.=null2019.09 | 56.6 | 29.2 | 32.6 | 21.5 | 46.5 | 4.81 | 0.03 | 26.5 | — | — | 70.8 | 60.3 | 3.5 | 74.1 | 20.4 | 8.9 | 13.1 | 14.8 | 13.1 | — | 33.6 | |
| ASA2023.11 | 56.4 | 49.3 | 91.2 | 48.5 | 80.4 | 3.7 | 0.3 | 21.7 | — | — | 79.5 | 83.6 | 21 | 80.3 | 36.2 | 20 | 32.9 | 5.5 | 5.2 | — | — | |
| DISEBase Model=Deeplab v22019.03 | 55.8 | 41.5 | 91.7 | 53.5 | 77.1 | 2.5 | 0.2 | 27.1 | — | — | 78.4 | 81.2 | 19.2 | 82.3 | 30.3 | 17.1 | 34.3 | 6.2 | 7.6 | — | — | |
| All StructureAppr.=Adv2019.09 | 55.8 | 41.5 | 91.7 | 53.5 | 77.1 | 2.5 | 0.2 | 27.1 | — | — | 78.4 | 81.2 | 19.2 | 82.3 | 30.3 | 17.1 | 34.3 | 6.2 | 7.6 | — | 48.7 | |
| Source onlyBackbone=ResNet-1012019.03 | 55.3 | 38.6 | 55.6 | 23.8 | 74.6 | — | — | — | 6.1 | 12.1 | 74.8 | 79 | 19.1 | 39.6 | 23.3 | 13.7 | 25 | — | — | — | — | |
| Source onlyBackbone=ResNet-101, Protocol=Direct transfer2018.09 | 55.3 | 38.6 | 55.6 | 23.8 | 74.6 | — | — | — | 6.1 | 12.1 | 74.8 | 79 | 19.1 | 39.6 | 23.3 | 13.7 | 25 | — | — | — | — | |
| Source onlyBackbone=ResNet-1012019.08 | 55.3 | — | 55.6 | 23.8 | 74.6 | — | — | — | 6.1 | 12.1 | 74.8 | 79 | 19.1 | 39.6 | 23.3 | 13.7 | 25 | — | — | — | 38.6 | |
| AdaptSetNetAppr.=Adv2019.09 | 54.8 | 39.6 | 81.7 | 39.1 | 78.4 | 11.1 | 0.3 | 25.8 | — | — | 79.1 | 80.8 | 21 | 66.8 | 34.7 | 13.8 | 29.9 | 6.8 | 9 | — | 45.8 | |
| Source only (ours)Backbone=ResNet-1012019.08 | 54.7 | 33 | 55.6 | 22.7 | 68.6 | 4.3 | 0.1 | 23 | 5.6 | 9.1 | 77.2 | 75.9 | 8.7 | 81.5 | 23.9 | 8.4 | 8.8 | — | — | — | 38.5 | |
| DADA2023.11 | 54.7 | 49.8 | 89.2 | 44.8 | 81.4 | 6.8 | 0.3 | 26.2 | — | — | 81.8 | 84 | 19.3 | 79.7 | 40.7 | 14 | 38.8 | 8.6 | 11.1 | — | — | |
| Source OnlyBackbone=DRN-1052017.12 | 54.3 | 23.4 | 14.9 | 11.4 | 58.7 | 1.9 | 0 | 24.1 | 1.2 | 6 | 68.8 | 76 | 7.1 | 34.2 | 15 | 0.8 | 0 | — | — | — | — | |
| Tsai et al.Base Model=Deeplab v22019.03 | 54.3 | 40 | 84.3 | 42.7 | 77.5 | 9.3 | 0.2 | 22.9 | — | — | 77.9 | 82.5 | 21 | 72.3 | 32.2 | 18.9 | 32.3 | 4.7 | 7 | — | — | |
| OutputAdaptBackbone=ResNet-1012019.08 | 54.3 | — | 84.3 | 42.7 | 77.5 | — | — | — | 4.7 | 7 | 77.9 | 82.5 | 21 | 72.3 | 32.2 | 18.9 | 32.3 | — | — | — | 46.7 | |
| AdaptSegNet2023.11 | 54.3 | 46.7 | 84.3 | 42.7 | 77.5 | — | — | — | — | — | 77.9 | 82.5 | 21 | 72.3 | 32.2 | 18.9 | 32.3 | 4.7 | 7 | — | — | |
| BL2023.11 | 54.1 | 51.4 | 86 | 46.7 | 80.3 | — | — | — | — | — | 79.2 | 81.3 | 27.9 | 73.7 | 42.2 | 25.7 | 45.3 | 14.1 | 11.6 | — | — | |
| AdaSegNetBackbone=ResNet-1012019.03 | 53.5 | 45.9 | 79.2 | 37.2 | 78.8 | — | — | — | 9.9 | 10.5 | 78.2 | 80.5 | 19.6 | 67 | 29.5 | 21.6 | 31.3 | — | — | — | — | |
| Baseline (TAN)Backbone=ResNet-101, Protocol=Adversarial Learning2018.09 | 53.5 | 45.9 | 79.2 | 37.2 | 78.8 | — | — | — | 9.9 | 10.5 | 78.2 | 80.5 | 19.6 | 67 | 29.5 | 21.6 | 31.3 | — | — | 7.3 | — | |
| PatchAlign2023.11 | 53.5 | 46.5 | 82.4 | 38 | 78.6 | 8.7 | 0.6 | 26 | — | — | 75.5 | 84.6 | 21.6 | 71.4 | 32.6 | 19.3 | 31.7 | 3.9 | 11.1 | — | — | |
| CLANBackbone=ResNet-101, Protocol=Adversarial Learning2018.09 | 53.4 | 47.8 | 81.3 | 37 | 80.1 | — | — | — | 16.1 | 13.7 | 78.2 | 81.5 | 21.2 | 73 | 32.9 | 22.6 | 30.7 | — | — | 9.2 | — | |
| CLANBackbone=ResNet-1012019.08 | 53.4 | — | 81.3 | 37 | 80.1 | — | — | — | 16.1 | 13.7 | 78.2 | 81.5 | 21.2 | 73 | 32.9 | 22.6 | 30.7 | — | — | — | 47.8 | |
| CLANAppr.=Adv2019.09 | 53.4 | — | 81.3 | 37 | 80.1 | — | — | 16.1 | — | — | 78.2 | 81.5 | 21.2 | 73 | 32.9 | 22.6 | 30.7 | 13.7 | 7.6 | — | 47.8 | |
| CLAN2023.11 | 53.4 | 47.8 | 81.3 | 37 | 80.1 | — | — | — | — | — | 78.2 | 81.5 | 21.2 | 73 | 32.9 | 22.6 | 30.7 | 16.1 | 13.7 | — | — | |
| Maximum Classifier DiscrepancyBackbone=DRN-105, k=22017.12 | 51.5 | 36.3 | 83.5 | 40.9 | 77.6 | 6 | 0.1 | 27.9 | 6.2 | 6 | 83.1 | 83.5 | 11.8 | 78.9 | 19.8 | 4.6 | 0 | — | — | — | — | |
| Wu et al.Base Model=FCN8s2019.03 | 51.5 | 36.5 | 81.5 | 33.4 | 72.4 | 7.9 | 0.2 | 20 | — | — | 71 | 68.7 | 18.7 | 75.3 | 22.7 | 12.8 | 28.1 | 8.6 | 10.5 | — | — | |
| Maximum Classifier DiscrepancyBackbone=DRN-105, k=42017.12 | 51.3 | 37.2 | 88.1 | 43.2 | 79.1 | 2.4 | 0.1 | 27.3 | 7.4 | 4.9 | 83.4 | 81.1 | 10.9 | 82.1 | 29 | 5.7 | 0 | — | — | — | — | |
| FCN WldBackbone=VGG-162017.12 | 51.2 | 20.2 | 11.5 | 19.6 | 30.8 | 4.4 | 0 | 20.3 | 0.1 | 11.7 | 42.3 | 68.7 | 3.8 | 54 | 0.2 | 0.6 | 8.2 | — | — | — | — | |
| FCN WldBackbone=FCN-8s, Reference=[27]2017.07 | 51.2 | 20.2 | 11.5 | 19.6 | 30.8 | 4.4 | 0 | 20.3 | 0.1 | 11.7 | 42.3 | 68.7 | 3.8 | 54 | 3.2 | 0.2 | 0.6 | — | — | — | — | |
| FCN Wldsource=Reported in [10]2018.12 | 51.2 | 20.2 | 11.5 | 19.6 | 30.8 | 4.4 | 0 | 20.3 | 0.1 | 11.7 | 42.3 | 68.7 | 3.8 | 54 | 3.2 | 0.2 | 0.6 | — | — | — | — | |
| FCN WldBackbone=VGG-162019.03 | 51.2 | 22.9 | 11.5 | 19.6 | 30.8 | — | — | — | 0.1 | 11.7 | 42.3 | 68.7 | 3.8 | 54 | 3.2 | 0.2 | 0.6 | — | — | — | — | |
| FCNs in the wildBackbone=VGG16-FCN8s, Protocol=Adversarial Learning2018.09 | 51.2 | 22.9 | 11.5 | 19.6 | 30.8 | — | — | — | 0.1 | 11.7 | 42.3 | 68.7 | 3.8 | 54 | 3.2 | 0.2 | 0.6 | — | — | 2.7 | — | |
| NoAdaptBackbone=FCN-8s, Reference=[27]2017.07 | 51.1 | 17.4 | 6.4 | 17.7 | 29.7 | 1.2 | 0 | 15.1 | 0 | 7.3 | 30.3 | 66.8 | 1.5 | 47.3 | 2.9 | 0.1 | 0 | — | — | — | — | |
| NoAdaptsource=Reported in [10]2018.12 | 51.1 | 17.4 | 6.4 | 17.7 | 29.7 | 1.2 | 0 | 15.1 | 0 | 7.2 | 30.3 | 66.8 | 1.5 | 47.3 | 3.9 | 0.1 | 0 | — | — | — | — | |
| Source onlyBackbone=VGG-162019.03 | 51.1 | 20.2 | 6.4 | 17.7 | 29.7 | — | — | — | 0 | 7.2 | 30.3 | 66.8 | 1.5 | 47.3 | 3.9 | 0.1 | 0 | — | — | — | — | |
| Wu et al.Base Model=PSPNet2019.03 | 51.1 | 38.4 | 82.8 | 36.4 | 75.7 | 5.1 | 0.1 | 25.8 | — | — | 74.7 | 76.9 | 15.9 | 77.7 | 24.8 | 4.1 | 37.3 | 8.04 | 18.7 | — | — | |
| Source onlyBackbone=VGG16-FCN8s, Protocol=Direct transfer (Baseline for AT)2018.09 | 51.1 | 20.2 | 6.4 | 17.7 | 29.7 | — | — | — | 0 | 7.2 | 30.3 | 66.8 | 1.5 | 47.3 | 3.9 | 0.1 | 0 | — | — | — | — | |
| Maximum Classifier DiscrepancyBackbone=DRN-105, k=32017.12 | 51 | 37.3 | 84.8 | 43.6 | 79 | 3.9 | 0.2 | 29.1 | 7.2 | 5.5 | 83.8 | 83.1 | 11.7 | 79.9 | 27.2 | 6.2 | 0 | — | — | — | — | |
| Source onlyBackbone=VGG-162019.08 | 49.7 | 25.4 | 4.7 | 11.6 | 62.3 | 10.7 | 0 | 22.8 | 4.3 | 15.3 | 68 | 70.8 | 6.4 | 60.5 | 11.8 | 2.6 | 4.3 | — | — | — | 28.7 | |
| CBSTBackbone=VGG-162019.03 | 49.1 | 36.1 | 69.6 | 28.7 | 69.5 | — | — | — | 11.9 | 13.6 | 82 | 81.9 | 14.5 | 66 | 6.6 | 3.7 | 32.4 | — | — | — | — | |
| CBSTBackbone=VGG16-FCN8s, Protocol=Self-training2018.09 | 49.1 | 36.1 | 69.6 | 28.7 | 69.5 | — | — | — | 11.9 | 13.6 | 82 | 81.9 | 14.5 | 66 | 6.6 | 3.7 | 32.4 | — | — | 9.9 | — | |
| CBSTBackbone=VGG-162019.08 | 49.1 | 35.4 | 69.6 | 28.7 | 69.5 | 12.1 | 0.1 | 25.4 | 11.9 | 13.6 | 82 | 81.9 | 14.5 | 66 | 6.6 | 3.7 | 32.4 | — | — | — | 40.4 | |
| DANNBackbone=DRN-1052017.12 | 48.3 | 32.5 | 67 | 29.1 | 71.5 | 14.3 | 0.1 | 28.1 | 12.6 | 10.3 | 72.7 | 76.7 | 12.7 | 62.5 | 11.3 | 2.7 | 0 | — | — | — | — | |
| Sankaranarayanan et al.Base Model=FCN8s2019.03 | 48.2 | 36.1 | 80.1 | 29.1 | 77.5 | 2.8 | 0.4 | 26.8 | — | — | 78.1 | 76.7 | 15.2 | 70.5 | 17.4 | 8.7 | 16.7 | 11.1 | 18 | — | — | |
| STBackbone=VGG-162019.08 | 48.1 | 23.9 | 0.2 | 14.5 | 53.8 | 1.6 | 0 | 18.9 | 0.9 | 7.8 | 72.2 | 80.3 | 6.3 | 67.7 | 4.7 | 0.2 | 4.5 | — | — | — | 27.8 | |
| PyCDABackbone=VGG-162019.08 | 48 | 35.9 | 80.6 | 26.6 | 74.5 | 2 | 0.1 | 18.1 | 13.7 | 14.2 | 80.8 | 71 | 19 | 72.3 | 22.5 | 12.1 | 18.1 | — | — | — | 42.6 | |
| Maximum Classifier DiscrepancyBackbone=DRN-105, Training=GRL2017.12 | 47.8 | 34.8 | 74.7 | 35.5 | 75.9 | 6.2 | 0.1 | 29 | 7.4 | 6.1 | 82.9 | 83.4 | 9.2 | 71.7 | 19.3 | 7 | 0 | — | — | — | — | |
| SWDBackbone=VGG-162019.03 | 47.4 | 43.5 | 83.3 | 35.4 | 82.1 | — | — | — | 12.2 | 12.6 | 83.8 | 76.5 | 12 | 71.5 | 17.9 | 1.6 | 29.7 | — | — | — | — | |
| Curriculum Domain Adaptation (SP)Backbone=FCN-8s, Supervision=Landmark superpixel label distributions2017.07 | 47.3 | 28.1 | 61.8 | 23.9 | 74.6 | 0.1 | 0.4 | 8.1 | 3.7 | 2.7 | 75.5 | 68.7 | 7.9 | 42.9 | 21.4 | 0.8 | 10.2 | — | — | — | — | |
| Curriculum Domain AdaptationSuperpixel information=true2018.12 | 47.3 | 28.1 | 61.8 | 23.9 | 74.6 | 0.1 | 0.4 | 8.1 | 3.7 | 2.7 | 75.5 | 68.7 | 7.9 | 42.9 | 21.4 | 0.8 | 10.2 | — | — | — | — | |
| CDA (I+SP)Backbone=VGG-162017.12 | 47.1 | 29 | 65.2 | 26.1 | 74.9 | 0.1 | 0.5 | 10.7 | 3.7 | 3 | 76.1 | 70.6 | 7.9 | 43.2 | 20.7 | 0.7 | 13.1 | — | — | — | — | |
| Curriculum Domain Adaptation (I+SP)Backbone=FCN-8s, Supervision=Global + Landmark superpixel label distributions2017.07 | 47.1 | 29 | 65.2 | 26.1 | 74.9 | 0.1 | 0.5 | 10.7 | 3.5 | 3 | 76.1 | 70.6 | 8.2 | 43.2 | 20.7 | 0.7 | 13.1 | — | — | — | — | |
| Curriculum Domain AdaptationIntensity information=true, Superpixel information=true2018.12 | 47.1 | 29 | 65.2 | 26.1 | 74.9 | 0.1 | 0.5 | 10.7 | 3.7 | 3 | 76.1 | 70.6 | 8.2 | 43.2 | 20.7 | 0.7 | 13.1 | — | — | — | — | |
| CLANBackbone=VGG16-FCN8s, Protocol=Adversarial Learning2018.09 | 46.5 | 39.3 | 80.4 | 30.7 | 74.7 | — | — | — | 1.4 | 8 | 77.1 | 79 | 8.9 | 73.8 | 18.2 | 2.2 | 9.9 | — | — | 19.1 | — | |
| CLANBackbone=VGG-162019.08 | 46.5 | — | 80.4 | 30.7 | 74.7 | — | — | — | 1.4 | 8 | 77.1 | 79 | 8.9 | 73.8 | 18.2 | 2.2 | 9.9 | — | — | — | 39.3 | |
| Curriculum Domain AdaptationColor Constancy=true, Intensity information=true, Superpixel information=true2018.12 | 45 | 29.7 | 57.4 | 23.1 | 74.7 | 0.5 | 0.6 | 14 | 5.3 | 4.3 | 77.8 | 73.7 | 11 | 44.8 | 21.2 | 1.9 | 20.3 | — | — | — | — | |
| CDABackbone=VGG-162019.08 | 45 | 29.7 | 57.4 | 23.1 | 74.7 | 0.5 | 0.6 | 14 | 5.3 | 4.3 | 77.8 | 73.7 | 11 | 44.8 | 21.2 | 1.9 | 20.3 | — | — | — | 35.4 | |
| Chen et al.Base Model=Deeplab v22019.03 | 44.5 | 36.2 | 77.7 | 30 | 77.5 | 9.6 | 0.3 | 25.8 | — | — | 77.6 | 79.8 | 16.6 | 67.8 | 14.5 | 7 | 23.8 | 10.3 | 15.6 | — | — | |
| ROADBackbone=VGG-162019.08 | 44.5 | 36.2 | 77.7 | 30 | 77.5 | 9.6 | 0.3 | 25.8 | 10.3 | 15.6 | 77.6 | 79.8 | 16.6 | 67.8 | 14.5 | 7 | 23.8 | — | — | — | 41.8 | |
| RoadAppr.=Adv2019.09 | 44.5 | 36.2 | 77.7 | 30 | 77.5 | 9.6 | 0.3 | 25.8 | — | — | 77.6 | 79.8 | 16.6 | 67.8 | 14.5 | 7 | 23.8 | 10.3 | 15.6 | — | 41.8 | |
| Curriculum Domain AdaptationColor Constancy=true, Superpixel information=true2018.12 | 44.4 | 28.9 | 59.6 | 22.4 | 74.5 | 0.5 | 0.5 | 10.9 | 5.4 | 3.4 | 76.8 | 73.4 | 9.9 | 43.7 | 17.6 | 1.8 | 17.7 | — | — | — | — | |
| SWDBackbone=ResNet-1012019.03 | 44 | 48.1 | 82.4 | 33.2 | 82.5 | — | — | — | 22.6 | 19.7 | 83.7 | 78.8 | 17.9 | 75.4 | 30.2 | 14.4 | 39.9 | — | — | — | — | |
| Source only (ours)Backbone=VGG-162019.08 | 43.9 | 22.4 | 50.1 | 20 | 49.4 | 0 | 0 | 16.3 | 0 | 0 | 69.9 | 54.2 | 4.7 | 43.1 | 6.1 | 0.1 | 0.1 | — | — | — | 26.3 | |
| AdaSegNetBackbone=VGG-162019.03 | 43.4 | 37.6 | 78.9 | 29.2 | 75.5 | — | — | — | 0.1 | 4.8 | 72.6 | 76.7 | 8.8 | 71.1 | 16 | 3.6 | 8.4 | — | — | — | — | |
| Baseline (TAN)Backbone=VGG16-FCN8s, Protocol=Adversarial Learning2018.09 | 43.4 | 37.6 | 78.9 | 29.2 | 75.5 | — | — | — | 0.1 | 4.8 | 72.6 | 76.7 | 8.8 | 71.1 | 16 | 3.6 | 8.4 | — | — | 17.4 | — | |
| FCNs in the WildBackbone=VGG-16, Appr.=Adv2018.11 | 42.3 | 20.2 | 11.5 | 19.6 | 30.8 | 4.4 | 0 | 20.3 | 0.1 | 11.7 | — | — | 68.7 | 51.2 | 3.8 | 54 | 3.2 | — | — | — | 22.1 | |
| Source onlyBackbone=VGG16-FCN8s, Protocol=Direct transfer2018.09 | 37.6 | 26.2 | 17.2 | 19.7 | 47.3 | — | — | — | 3 | 9.1 | 71.8 | 78.3 | 4.7 | 42.2 | 9 | 0.1 | 0.9 | — | — | — | — |