Open-vocabulary video instance segmentation on LVVIS (test)
36.3mAPSAM 3
Evaluation Results
| Method | Links | |
|---|---|---|
| SAM 32025.11 | 36.3 | |
| SAM 3 Detector + T-by-D2025.11 | 35.9 | |
| GLEEProcessing mode=all NPs at once2025.11 | 20.8 | |
| LLMDet + SAM 3 Tracker2025.11 | 15.2 | |
| GLEEProcessing mode=one NP at a time2025.11 | 9.3 |