Semantic Segmentation on Potsdam
91.74Pixel AccuracyUperNet
Evaluation results
| Method | Links | |
|---|---|---|
| UperNetPretrain=RingMo [32], Backbone=Swin-B2022.08 | 91.74 | |
| UNetFormerPretrain=IMP, Backbone=ResNet-182022.08 | 91.3 | |
| UperNetPretrain=MAE, Backbone=VITAE-B + RVSA2022.08 | 91.22 | |
| UperNetPretrain=RSP, Backbone=VITAEv2-S2022.08 | 91.21 | |
| UperNetPretrain=IMP, Backbone=Swin-T2022.08 | 91.17 | |
| UperNetPretrain=MAE, Backbone=VITAE-B + RVSA*2022.08 | 91.15 | |
| UperNetPretrain=MAE, Backbone=VITAE-B + VSA2022.08 | 91.12 | |
| UperNetPretrain=MAE, Backbone=VITAE-B2022.08 | 91.05 | |
| UperNetPretrain=MAE, Backbone=VIT-B + RVSA*2022.08 | 90.77 | |
| UperNetPretrain=IMP, Backbone=ResNet-502022.08 | 90.64 | |
| UperNetPretrain=MAE, Backbone=ViT-B + RVSA2022.08 | 90.6 | |
| UperNetPretrain=MAE, Backbone=ViT-B + VSA2022.08 | 90.54 | |
| UperNetPretrain=MAE, Backbone=ViT-B2022.08 | 90.32 | |
| DeeplabV3+Pretrain=IMP, Backbone=ResNet-502022.08 | 89.74 | |
| DANetPretrain=IMP, Backbone=ResNet-502022.08 | 89.72 | |
| PSPNetPretrain=IMP, Backbone=ResNet-502022.08 | 89.45 | |
| FCNPretrain=IMP, Backbone=VGG-162022.08 | 85.59 | |
| InfoSeg2021.10 | 57.3 | |
| AC2021.10 | 49.3 | |
| InMARS2021.10 | 47.3 | |
| IIC2021.10 | 45.4 | |
| IsolaMethodology=Clustering of features from non-segmentation method2021.10 | 44.9 | |
| DoerschMethodology=Clustering of features from non-segmentation method2021.10 | 37.2 | |
| K-Means2021.10 | 35.3 | |
| Random CNN2021.10 | 28.3 |