Semantic Segmentation on Cityscapes fine (test)
81.9mIoUCCNet
Evaluation Results
| Method | Links | |
|---|---|---|
| CCNetBackbone=ResNet-1012020.12 | 81.9 | |
| SETR-PUPBackbone=T-Large, Training data annotation=fine and coarse2020.12 | 81.64 | |
| DANetBackbone=ResNet-1012020.12 | 81.5 | |
| SETR-PUPBackbone=T-Large, Training iterations=100k2020.12 | 81.08 | |
| DenseASPPBackbone=DenseNet-1612020.12 | 80.6 | |
| PSANetBackbone=ResNet-1012020.12 | 80.1 | |
| OCNetBackbone=ResNet-1012020.12 | 80.1 | |
| Axial-DeepLab-XLBackbone=Axial-ResNet-XL2020.12 | 79.9 | |
| Axial-DeepLab-LBackbone=Axial-ResNet-L2020.12 | 79.5 | |
| BiSeNetBaseModel=Res1012018.08 | 78.9 | |
| BiSeNetBackbone=ResNet-1012020.12 | 78.9 | |
| PSPNetBaseModel=Res1012018.08 | 78.4 | |
| PSPNetBackbone=ResNet-1012020.12 | 78.4 | |
| BiSeNetBaseModel=Res182018.08 | 77.7 | |
| DUCBaseModel=Res1522018.08 | 76.1 | |
| SwiftNetPretrained on ImageNet=true, Number of Parameters=11.8M, GFLOPs=104, Resolution=1024 x 20482021.04 | 75.5 | |
| Lite-HRNet-30Pretrained on ImageNet=false, Number of Parameters=1.8M, GFLOPs=3.02, Resolution=512 x 10242021.04 | 75.3 | |
| BiSeNetV1 BPretrained on ImageNet=true, Number of Parameters=49.0M, GFLOPs=55.3, Resolution=768 x 15362021.04 | 74.7 | |
| RefineNetBaseModel=Res1012018.08 | 73.6 | |
| Lite-HRNet-18Pretrained on ImageNet=false, Number of Parameters=1.1M, GFLOPs=1.95, Resolution=512 x 10242021.04 | 72.8 | |
| MobileNetV3Backbone=V3, OS=16, MAdds (f)=9.74B, MAdds (h)=2.48B, CPU (f)=2.47s, CPU (h)=657ms2019.05 | 72.6 | |
| BiSeNetV2 SmallPretrained on ImageNet=false, GFLOPs=21.15, Resolution=512 x 10242021.04 | 72.6 | |
| MobileNetV3Pretrained on ImageNet=true, Number of Parameters=1.5M, GFLOPs=9.1, Resolution=1024 x 20482021.04 | 72.6 | |
| MobileNetV3Backbone=V3, OS=32, MAdds (f)=7.74B, MAdds (h)=1.98B, CPU (f)=2.06s, CPU (h)=534ms2019.05 | 72 | |
| FasterSegPretrained on ImageNet=true, Number of Parameters=4.4M, GFLOPs=28.2, Resolution=1024 x 20482021.04 | 71.5 | |
| BiSeNetBaseModel=Xception392018.08 | 71.4 | |
| DF1-Seg-d8Pretrained on ImageNet=true, Resolution=1024 x 20482021.04 | 71.4 | |
| CASPretrained on ImageNet=true, Resolution=768 x 15362021.04 | 70.5 | |
| DeepLab-v2+CRFBaseModel=Res1012018.08 | 70.4 | |
| DFANet A'Pretrained on ImageNet=true, Number of Parameters=7.8M, GFLOPs=1.7, Resolution=512 x 10242021.04 | 70.3 | |
| LRRBaseModel=VGG162018.08 | 69.7 | |
| ICNetPretrained on ImageNet=true, GFLOPs=28.3, Resolution=1024 x 20482021.04 | 69.5 | |
| MobileNetV3-SmallBackbone=V3-Small, OS=16, MAdds (f)=2.90B, MAdds (h)=0.74B, CPU (f)=1.21s, CPU (h)=327ms2019.05 | 69.4 | |
| MobileNet V3-SmallPretrained on ImageNet=true, Number of Parameters=0.5M, GFLOPs=2.7, Resolution=512 x 10242021.04 | 69.4 | |
| BiSeNetV1 APretrained on ImageNet=true, Number of Parameters=5.8M, GFLOPs=14.8, Resolution=768 x 15362021.04 | 68.4 | |
| MobileNetV3-SmallBackbone=V3-Small, OS=32, MAdds (f)=2.06B, MAdds (h)=0.53B, CPU (f)=1.03s, CPU (h)=275ms2019.05 | 68.3 | |
| Fast-SCNNPretrained on ImageNet=false, Resolution=1024 x 20482021.04 | 68 | |
| Dilation10BaseModel=VGG162018.08 | 67.1 | |
| AdelaideBaseModel=VGG162018.08 | 66.4 | |
| ESPNetv2MAdds (h)=2.7B2019.05 | 66.2 | |
| FCN-8sBaseModel=VGG162018.08 | 65.3 | |
| DeepLabBaseModel=VGG162018.08 | 63.1 | |
| CCC2MAdds (h)=3.15B2019.05 | 62 | |
| ESPNetv1MAdds (h)=4.5B2019.05 | 60.3 |