Semantic Segmentation on Cityscapes (val) (Metrics at 1/8, 1/4, 1/2, Full Resolution)
0.741Score (1/8 Res)Ours
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| OursModel=DeepLabv3Plus2021.04 | 0.741 | 0.778 | 0.787 | 0.787 | |
| BaselineModel=DeepLabv3Plus2021.04 | 0.689 | 0.734 | 0.769 | 0.787 | |
| OursModel=DeepLabV22021.04 | 0.676 | 0.693 | 0.707 | 0.701 | |
| ECSModel=DeepLabv3Plus2021.04 | 0.674 | 0.707 | 0.729 | 0.748 | |
| ClassMixModel=DeepLabV22021.04 | 0.614 | 0.636 | 0.663 | 0.662 | |
| BaselineModel=DeepLabV22021.04 | 0.606 | 0.667 | 0.693 | 0.701 | |
| DST-CBCModel=DeepLabV22021.04 | 0.605 | 0.644 | — | 0.669 | |
| CutMixModel=DeepLabV22021.04 | 0.603 | 0.6387 | — | 0.677 | |
| S4GAN + MLMTModel=DeepLabV22021.04 | 0.593 | 0.619 | — | 0.658 | |
| AdvSemiSegModel=DeepLabV22021.04 | 0.588 | 0.623 | 0.657 | 0.66 |