Interactive Segmentation on Berkeley (test)
90IoU@1SAM
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SAMBackbone=ViT-L, Best matching selection=true, Training Dataset=SA-1B2024.05 | 90 | — | — | — | |
| GraCoBackbone=ViT-L, Granularity guidance=AGG, Training Dataset=SBD2024.05 | 89 | — | — | — | |
| SAMBackbone=ViT-B, Best matching selection=true, Training Dataset=SA-1B2024.05 | 88 | — | — | — | |
| SimpleClickBackbone=ViT-B, Fine-tuning utilizing part annotation=false, Training Dataset=SBD2024.05 | 85 | — | — | — | |
| SimpleClickBackbone=ViT-L, Fine-tuning utilizing part annotation=false, Training Dataset=SBD2024.05 | 84 | — | — | — | |
| GraCoBackbone=ViT-B, Granularity guidance=AGG, Training Dataset=SBD2024.05 | 84 | — | — | — | |
| GraCoBackbone=ViT-B, Granularity guidance=GT, Training Dataset=SBD2024.05 | 80 | — | — | — | |
| GraCoBackbone=ViT-L, Granularity guidance=GT, Training Dataset=SBD2024.05 | 76 | — | — | — | |
| SAMBackbone=ViT-L, Best matching selection=false, Training Dataset=SA-1B2024.05 | 61 | — | — | — | |
| SAMBackbone=ViT-B, Best matching selection=false, Training Dataset=SA-1B2024.05 | 56 | — | — | — | |
| SimpleClickBackbone=ViT-L, Fine-tuning utilizing part annotation=true, Training Dataset=SBD2024.05 | 46 | — | — | — | |
| SimpleClickBackbone=ViT-B, Fine-tuning utilizing part annotation=true, Training Dataset=SBD2024.05 | 43 | — | — | — | |
| CDNetBackbone=ResNet-34, Training Data=COCO + LVIS2024.04 | — | 1.47 | 2.06 | 5.42 | |
| DynaMITeBackbone=Swin-L, Training Data=COCO + LVIS2024.04 | — | 1.39 | 1.9 | — | |
| EMC-ClickBackbone=HRNet-32, Training Data=COCO + LVIS2024.04 | — | 1.48 | 2.35 | 6.95 | |
| FocalClickBackbone=SegF-B3, Training Data=COCO + LVIS2024.04 | — | 1.55 | 1.92 | 4.63 | |
| iCMFormerBackbone=ViT-B, Training Data=COCO + LVIS2024.04 | — | 1.4 | 1.86 | — | |
| MFPBackbone=ViT-B, Training Data=COCO + LVIS2024.04 | — | 1.35 | 1.9 | 4.68 | |
| PseudoClickBackbone=HRNet-32, Training Data=COCO + LVIS2024.04 | — | 1.4 | 2.08 | — | |
| RITMBackbone=HRNet-32, Training Data=COCO + LVIS2024.04 | — | 1.43 | 2.1 | 5.41 | |
| SimpleClickBackbone=ViT-B, Training Data=COCO + LVIS2024.04 | — | 1.36 | 1.97 | 5.05 |