Referring Video Object Segmentation and Point-to-Mask Tracking on MeViS-U
37.1F-ScoreLLaVA-OV-2-8B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| LLaVA-OV-2-8BParameter count=8B, Conversion=point-to-mask via SAM 22026.05 | 37.1 | 31.7 | 45.7 | |
| Qwen3-VL-8BParameter count=8B2026.05 | 29.9 | 34.5 | 28.4 | |
| Keye-VL-1.52026.05 | 10.1 | 15.8 | 7.2 | |
| LLaVA-OV-1.52026.05 | 7.3 | 9 | 6.1 | |
| InternVL3.5-8BParameter count=8B2026.05 | 7.2 | 6.6 | 7.5 | |
| PLM-8BParameter count=8B2026.05 | 5 | 34.4 | 7.6 |