Semantic Segmentation on HCMSSD–World (val)
76.3mIoUMask2Former–Swin-L
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Mask2Former–Swin-Ltraining_mode=transfer2026.03 | 76.3 | 88.4 | 92.6 | 90.5 | |
| Mask2Former–Swin-Ltraining_mode=few-shots2026.03 | 74.4 | 81.7 | 88.6 | 85.2 | |
| Mask2Former–Swin-Ltraining_mode=retrained2026.03 | 74.2 | 86.8 | 84.1 | 85.4 | |
| UNet–ResNet101architecture=UNet–ResNet1012026.03 | 45.2 | 64.9 | 62.6 | 63.8 | |
| UNet–Transformerarchitecture=UNet–Transformer2026.03 | 28.1 | 43.9 | 45.2 | 44.6 |