Referring Video Object Segmentation on DAVIS RefVOS
72.5J&F ScoreUNINEXT
Evaluation results
| Method | Links | |
|---|---|---|
| UNINEXTBackbone=ViT-H, Joint Training=false, Universal=false2024.02 | 72.5 | |
| UNINEXTBackbone=ConvNeXtL, Joint Training=false, Universal=false2024.02 | 66.7 | |
| UniRefBackbone=SwinL, Joint Training=true, Universal=false2024.02 | 66.3 | |
| UNINEXTBackbone=ResNet50, Joint Training=false, Universal=false2024.02 | 63.9 | |
| UniRefBackbone=ResNet50, Joint Training=true, Universal=false2024.02 | 63.5 | |
| UniVSBackbone=SwinL, Joint Training=true, Universal=true2024.02 | 59.4 | |
| UniVSBackbone=SwinB, Joint Training=true, Universal=true2024.02 | 58.6 | |
| UniVSBackbone=SwinT, Joint Training=true, Universal=true2024.02 | 58.5 | |
| UniVSBackbone=ResNet50, Joint Training=true, Universal=true2024.02 | 57.9 |