Hierarchical Semantic Segmentation on Cityscapes (val)
83.74mIoU¹HSSN (MaskFormer)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| HSSN (MaskFormer)Backbone=Swin-Small, Multi-scale inference=true2022.03 | 83.74 | 94.39 | |
| HSSN (OCRNet)Backbone=HRNet-W48, Multi-scale inference=true2022.03 | 83.37 | 93.92 | |
| HSSN (DeepLabV3+)Backbone=ResNet-101, Multi-scale inference=true2022.03 | 83.02 | 93.31 | |
| MaskFormerBackbone=Swin-Small, Multi-scale inference=true2022.03 | 82.57 | 92.96 | |
| OCRNetBackbone=HRNet-W48, Multi-scale inference=true2022.03 | 82.33 | 92.57 | |
| DeepLabV3+Backbone=ResNet-101, Multi-scale inference=true2022.03 | 82.08 | 92.16 | |
| HRNetBackbone=HRNet-W48, Multi-scale inference=true2022.03 | 81.96 | 92.12 | |
| HANetBackbone=ResNet-101, Multi-scale inference=true2022.03 | 81.82 | — | |
| AcfnetBackbone=ResNet-101, Multi-scale inference=true2022.03 | 81.6 | — | |
| DANetBackbone=ResNet-101, Multi-scale inference=true2022.03 | 81.52 | — | |
| PANBackbone=ResNet-101, Multi-scale inference=true2022.03 | 81.12 | — | |
| CCNetBackbone=ResNet-101, Multi-scale inference=true2022.03 | 81.08 | — | |
| PSANetBackbone=ResNet-101, Multi-scale inference=true2022.03 | 80.96 | — | |
| PSPNetBackbone=ResNet-101, Multi-scale inference=true2022.03 | 80.91 | — | |
| DeepLabV2Backbone=ResNet-101, Multi-scale inference=true2022.03 | 70.22 | — |