Video Semantic Segmentation on VIPSeg (val)
64.9mIoUTube-Link
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Tube-LinkBackbone=Swin-large2023.03 | 64.9 | 92.4 | 89.9 | |
| Tube-LinkBackbone=Swin-base2023.03 | 62.3 | 91.4 | 89.3 | |
| Video K-Net+Backbone=Swin-base2023.03 | 57.2 | 90.1 | 87.8 | |
| Tube-LinkBackbone=ResNet502023.03 | 43.4 | 89.2 | 85.4 | |
| Video K-Net (PSPNet)Backbone=ResNet1012023.03 | 39 | 88.2 | 84.2 | |
| Mask2FormerBackbone=ResNet502023.03 | 38.4 | 87.5 | 82.5 | |
| Video K-Net (Deeplabv3+)Backbone=ResNet1012023.03 | 38.3 | 88 | 83.1 |