Open-vocabulary video instance segmentation on SA-Co/VEval YT-Temporal-1B (val)
71.3cgF1Human
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Human2025.11 | 71.3 | 98 | 73.1 | 78.3 | 68.6 | 89.6 | 83.4 | |
| SAM 32025.11 | 50.2 | 88 | 57.2 | 70.5 | 60.5 | 82.7 | 70.8 | |
| SAM 3temporal disambiguation=false2025.11 | 47 | 84 | 55.8 | 68.6 | 57.8 | 82 | 70.1 | |
| SAM 3 Detector + T-by-D as tracker2025.11 | 46 | 90 | 51.3 | 68.5 | 61 | 77.6 | 70.4 | |
| LLMDet + SAM 3 TrackerZero-shot=true2025.11 | 10.5 | 39 | 26.8 | 39.6 | 20.1 | 78.3 | 37 | |
| GLEEPrompting strategy=one NP at a time, Zero-shot=true2025.11 | 2.3 | 23 | 10.3 | 19.5 | 9.2 | 42 | 23.2 | |
| GLEEPrompting strategy=all NPs at once, Zero-shot=true2025.11 | 1.8 | 26 | 7 | 17.1 | 7.3 | 42 | 16.9 |