Semi-supervised Semantic Segmentation on Cityscapes (val)
80.93mIoUDiGA
Evaluation results
| Method | Links | |
|---|---|---|
| DiGABackbone=HRNet, Partition protocol=1/2 (1488)2023.04 | 80.93 | |
| CPSBackbone=HRNet, Partition protocol=1/2 (1488)2023.04 | 80.67 | |
| DiGABackbone=HRNet, Partition protocol=1/4 (744)2023.04 | 80.01 | |
| CPSBackbone=HRNet, Partition protocol=1/4 (744)2023.04 | 79.24 | |
| DiGABackbone=HRNet, Partition protocol=1/8 (372)2023.04 | 78.51 | |
| CPSBackbone=HRNet, Partition protocol=1/8 (372)2023.04 | 77.92 | |
| DiGABackbone=HRNet, Partition protocol=1/16 (186)2023.04 | 76.86 | |
| CPSBackbone=HRNet, Partition protocol=1/16 (186)2023.04 | 75.09 | |
| SemiSeg-ContrastiveLabeled Ratio=1/3, Backbone=Deeplabv2 with ResNet-101, Training=ImageNet pre-trained, Domain Adaptation=no adaptation2021.04 | 65.1 | |
| SemiSeg-ContrastiveLabeled Ratio=1/6, Backbone=Deeplabv2 with ResNet-101, Training=ImageNet pre-trained, Domain Adaptation=no adaptation2021.04 | 63.7 | |
| SemiSeg-ContrastiveLabeled Ratio=1/15, Backbone=Deeplabv2 with ResNet-101, Training=ImageNet pre-trained, Domain Adaptation=no adaptation2021.04 | 59.9 | |
| SemiSeg-ContrastiveLabeled Ratio=1/30, Backbone=Deeplabv2 with ResNet-101, Training=ImageNet pre-trained, Domain Adaptation=no adaptation2021.04 | 58 |