Referring Image Segmentation on G-Ref hard UMD (val)
67.79oIoUMagNet+
Evaluation results
| Method | Links | |
|---|---|---|
| MagNet+Backbone=Swin-B, Training Setting=Multiple Datasets2023.12 | 67.79 | |
| PolyFormerBackbone=Swin-B, Training Setting=Multiple Datasets2023.12 | 67.76 | |
| LISA-7BBackbone=SAM-H, Training Setting=Multiple Datasets2023.12 | 66.4 | |
| SEEMBackbone=Focal-T, Training Setting=Extra Datasets2023.12 | 65.7 | |
| MagNetBackbone=Swin-B, Training Setting=Single Dataset2023.12 | 65.36 | |
| ReLABackbone=Swin-B, Training Setting=Single Dataset2023.12 | 65 | |
| SADLRBackbone=Swin-B, Training Setting=Single Dataset2023.12 | 63.6 | |
| MCRESBackbone=Swin-B, Training Setting=Single Dataset2023.12 | 63.51 | |
| PVDBackbone=Swin-B, Training Setting=Single Dataset2023.12 | 63.13 | |
| VPDBackbone=Swin-B, Training Setting=Single Dataset2023.12 | 63.12 | |
| CoupAlignBackbone=Swin-B, Training Setting=Single Dataset2023.12 | 62.84 | |
| LAVTBackbone=Swin-B, Training Setting=Single Dataset2023.12 | 61.24 | |
| CRISBackbone=ResNet-101, Training Setting=Single Dataset2023.12 | 59.87 | |
| VLTBackbone=Darknet-53, Training Setting=Single Dataset2023.12 | 52.99 |