Referring Image Segmentation on OmniRef Visual (test)
76.63cIoUOmniSegNet
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| OmniSegNetBackbone=Swin-B, Text Encoder=BERT2025.12 | 76.63 | 68.87 | 90.81 | |
| DCAMABackbone=Swin-B2025.12 | 60.23 | 49.91 | 80.46 | |
| DCAMA+ReLABackbone=Swin-B, Text Encoder=BERT2025.12 | 60.23 | 49.91 | 80.46 | |
| VRP-SAMBackbone=ViT-L2025.12 | 55.7 | 52.74 | 78.63 | |
| VRP-SAM+ReLABackbone=ViT-L, Text Encoder=BERT2025.12 | 55.7 | 52.74 | 78.63 | |
| MatcherBackbone=ViT-L2025.12 | 41.29 | 42.04 | — | |
| GF-SAMBackbone=ViT-L2025.12 | 40.66 | 42.46 | — |