Semantic Segmentation on Cityscapes (val) (Adaptation Metrics)
74.76mIoUECS
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| ECSBackbone=DeepLabv3+, Labelled data ratio=Full, Resolution=Doubled2021.04 | 74.76 | — | — | — | |
| ECSBackbone=DeepLabv3+, Labelled data ratio=1/2, Resolution=Doubled2021.04 | 72.89 | — | — | 1.87 | |
| ECSBackbone=DeepLabv3+, Labelled data ratio=1/4, Resolution=Doubled2021.04 | 70.7 | — | — | 4.06 | |
| ReCoBackbone=DeepLabv3+, Labelled data ratio=1/22021.04 | 70.63 | — | — | -0.15 | |
| ReCoBackbone=DeepLabv3+, Labelled data ratio=Full2021.04 | 70.48 | — | — | — | |
| ReCoBackbone=DeepLabv2, Labelled data ratio=1/22021.04 | 68.69 | — | — | -0.09 | |
| ReCoBackbone=DeepLabv2, Labelled data ratio=Full2021.04 | 68.6 | — | — | — | |
| ReCoBackbone=DeepLabv3+, Labelled data ratio=1/42021.04 | 68.5 | — | — | 1.98 | |
| DMTBackbone=DeepLabv2, Labelled data ratio=Full2021.04 | 68.16 | — | — | — | |
| CutMixBackbone=DeepLabv2, Labelled data ratio=Full2021.04 | 67.53 | — | — | — | |
| ReCoBackbone=DeepLabv2, Labelled data ratio=1/42021.04 | 67.53 | — | — | 1.07 | |
| ECSBackbone=DeepLabv3+, Labelled data ratio=1/8, Resolution=Doubled2021.04 | 67.38 | — | — | 7.38 | |
| ReCoBackbone=DeepLabv3+, Labelled data ratio=1/82021.04 | 66.44 | — | — | 4.04 | |
| AdvSemSegBackbone=DeepLabv2, Labelled data ratio=Full2021.04 | 66.4 | — | — | — | |
| ClassMixBackbone=DeepLabv2, Labelled data ratio=1/22021.04 | 66.29 | — | — | -0.1 | |
| ClassMixBackbone=DeepLabv2, Labelled data ratio=Full2021.04 | 66.19 | — | — | — | |
| S4GANBackbone=DeepLabv2, Labelled data ratio=Full2021.04 | 65.8 | — | — | — | |
| AdvSemSegBackbone=DeepLabv2, Labelled data ratio=1/22021.04 | 65.7 | — | — | 0.7 | |
| ReCoBackbone=DeepLabv2, Labelled data ratio=1/82021.04 | 64.94 | — | — | 3.66 | |
| CutMixBackbone=DeepLabv2, Labelled data ratio=1/42021.04 | 63.87 | — | — | 3.66 | |
| ClassMixBackbone=DeepLabv2, Labelled data ratio=1/42021.04 | 63.63 | — | — | 2.56 | |
| DMTBackbone=DeepLabv2, Labelled data ratio=1/82021.04 | 63.03 | — | — | 5.13 | |
| AdvSemSegBackbone=DeepLabv2, Labelled data ratio=1/42021.04 | 62.3 | — | — | 4.1 | |
| S4GANBackbone=DeepLabv2, Labelled data ratio=1/42021.04 | 61.9 | — | — | 3.9 | |
| ClassMixBackbone=DeepLabv2, Labelled data ratio=1/82021.04 | 61.35 | — | — | 4.84 | |
| CutMixBackbone=DeepLabv2, Labelled data ratio=1/82021.04 | 60.34 | — | — | 7.19 | |
| ReCoBackbone=DeepLabv3+, Labelled data ratio=1/302021.04 | 60.28 | — | — | 10.2 | |
| S4GANBackbone=DeepLabv2, Labelled data ratio=1/82021.04 | 59.3 | — | — | 6.5 | |
| AdvSemSegBackbone=DeepLabv2, Labelled data ratio=1/82021.04 | 58.8 | — | — | 7.6 | |
| ReCoBackbone=DeepLabv2, Labelled data ratio=1/302021.04 | 56.53 | — | — | 12.07 | |
| DMTBackbone=DeepLabv2, Labelled data ratio=1/302021.04 | 54.81 | — | — | 13.36 | |
| ClassMixBackbone=DeepLabv2, Labelled data ratio=1/302021.04 | 54.07 | — | — | 12.12 | |
| CutMixBackbone=DeepLabv2, Labelled data ratio=1/302021.04 | 51.2 | — | — | 16.33 | |
| AdaptSegNet (multi-level)Baseline=ResNet-1012018.02 | — | 42.4 | 65.1 | -22.7 | |
| AdaptSegNet (single-level)Baseline=VGG-162018.02 | — | 35 | 61.8 | -25.2 | |
| CDABaseline=VGG-162018.02 | — | 28.9 | 60.3 | -31.4 | |
| CyCADA (feature)Baseline=VGG-162018.02 | — | 29.2 | 60.3 | -30.5 | |
| CyCADA (pixel)Baseline=VGG-162018.02 | — | 34.8 | 60.3 | -24.9 | |
| FCNs in the WildBaseline=VGG-162018.02 | — | 27.1 | 64.6 | -37.5 |