Semi-supervised Video Object Segmentation on 17 video datasets (test)
79.3J&F AccuracySAM 2
Evaluation Results
| Method | Links | |
|---|---|---|
| SAM 2Prompt type=ground-truth mask, Zero-shot=true2024.08 | 79.3 | |
| SAM 2Prompt type=5-click, Zero-shot=true2024.08 | 77.6 | |
| SAM 2Prompt type=3-click, Zero-shot=true2024.08 | 75.3 | |
| SAM 2Prompt type=bounding box, Zero-shot=true2024.08 | 74.4 | |
| SAM+CutiePrompt type=ground-truth mask, Zero-shot=true, Input protocol=Direct mask input (without SAM)2024.08 | 74.1 | |
| SAM+XMem++Prompt type=ground-truth mask, Zero-shot=true, Input protocol=Direct mask input (without SAM)2024.08 | 72.7 | |
| SAM+CutiePrompt type=5-click, Zero-shot=true2024.08 | 72.2 | |
| SAM+XMem++Prompt type=5-click, Zero-shot=true2024.08 | 70.6 | |
| SAM+CutiePrompt type=3-click, Zero-shot=true2024.08 | 70.1 | |
| SAM+CutiePrompt type=bounding box, Zero-shot=true2024.08 | 69.4 | |
| SAM+XMem++Prompt type=3-click, Zero-shot=true2024.08 | 68.4 | |
| SAM+XMem++Prompt type=bounding box, Zero-shot=true2024.08 | 67.6 | |
| SAM 2Prompt type=1-click, Zero-shot=true2024.08 | 64.7 | |
| SAM+XMem++Prompt type=1-click, Zero-shot=true2024.08 | 56.9 | |
| SAM+CutiePrompt type=1-click, Zero-shot=true2024.08 | 56.7 |