Semantic Segmentation on South Asian Megacities unseen cities, 30% labeled source data
46.1mIoUSLUM-i
Evaluation results
| Method | Links | |
|---|---|---|
| SLUM-iBackbone=DINOv22026.02 | 46.1 | |
| SLUM-iBackbone=ResNet-1012026.02 | 45.7 | |
| UniMatchBackbone=ResNet-1012026.02 | 44.5 | |
| UniMatch-v2Backbone=DINOv22026.02 | 44.3 | |
| SupervisedBackbone=DINOv22026.02 | 41.9 | |
| FixMatchBackbone=ResNet-1012026.02 | 40.5 | |
| SupervisedBackbone=ResNet-1012026.02 | 36.8 |