Video Reasoning Segmentation on ReVOS Referring
64.3J&F ScoreRCoT-Seg-3B
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| RCoT-Seg-3B2026.05 | 64.3 | 61.6 | 66.9 | |
| Veason-R1-7BModel size=7B2025.08 | 63.6 | 60.7 | 66.5 | |
| Veason-R1Backbone=Qwen2.5VL-7B, Protocol=Fully Trained2026.05 | 63.6 | 60.7 | 66.5 | |
| VRS-HQ-13BModel size=13B2025.08 | 63.3 | 61.1 | 65.5 | |
| VRS-HQ-13BVenue=CVPR20252026.05 | 63.3 | 61.1 | 65.5 | |
| Veason-R1-3BModel size=3B2025.08 | 63 | 60.3 | 65.6 | |
| Veason-R1-3BVenue=CVPR20262026.05 | 63 | 60.3 | 65.6 | |
| VRS-HQ-7BModel size=7B2025.08 | 62.1 | 59.8 | 64.5 | |
| VRS-HQ-7BVenue=CVPR20252026.05 | 62.1 | 59.8 | 64.5 | |
| VRS-HQBackbone=ChatUniVi-7B, Protocol=Fully Trained2026.05 | 62.1 | 59.8 | 64.5 | |
| Omni-R1-8BVenue=NeurIPS20252026.05 | 61.6 | — | — | |
| SteerSegBackbone=Qwen2.5VL-7B, Training Protocol=Frozen-LVLM, Training-free=false2026.05 | 61.1 | 58.2 | 63.9 | |
| SteerSegBackbone=Qwen2VL-7B, Training Protocol=Frozen-LVLM, Training-free=false2026.05 | 59.2 | 56.2 | 62.3 | |
| DecAF*Backbone=Qwen2.5VL-7B, Training Protocol=Frozen-LVLM, Training-free=true2026.05 | 58.7 | 54.8 | 62.6 | |
| HyperSeg-3BModel size=3B2025.08 | 58.5 | 56 | 60.9 | |
| HyperSeg-3BVenue=CVPR20252026.05 | 58.5 | 56 | 60.9 | |
| GLUS-7BModel size=7B2025.08 | 58.3 | 56 | 60.7 | |
| GLUS-7BVenue=CVPR20252026.05 | 58.3 | 56 | 60.7 | |
| GLUSBackbone=LLaVA-7B, Protocol=Fully Trained2026.05 | 58.3 | 56 | 60.7 | |
| VISA-13BModel size=13B2025.08 | 57.4 | 55.6 | 59.1 | |
| VISABackbone=LLaVA-13B2026.03 | 57.4 | 55.7 | 59 | |
| VISABackbone=Chat-UniVi-13B2026.03 | 57.4 | 55.6 | 59.1 | |
| VISA-13BVenue=ECCV20242026.05 | 57.4 | 55.6 | 59.1 | |
| InstructSeg-3BModel size=3B2025.08 | 57 | 54.8 | 59.2 | |
| InstructSeg-3BVenue=ICCV20252026.05 | 57 | 54.8 | 59.2 | |
| TrajSegBackbone=LLaVA-7B2026.03 | 56 | 53.9 | 58.1 | |
| SteerSegBackbone=InternVL3-8B, Training Protocol=Frozen-LVLM, Training-free=false2026.05 | 55.6 | 51.9 | 59.3 | |
| Loc-Head*Backbone=Qwen2.5VL-7B, Training Protocol=Frozen-LVLM, Training-free=true2026.05 | 53.1 | 49.3 | 56.9 | |
| Loc-Head*Backbone=Qwen2VL-7B, Training Protocol=Frozen-LVLM, Training-free=true2026.05 | 52.7 | 49.1 | 56.2 | |
| DecAF*Backbone=Qwen2VL-7B, Training Protocol=Frozen-LVLM, Training-free=true2026.05 | 52.7 | 48.9 | 56.4 | |
| SteerSegBackbone=LLaVA-OV-7B, Training Protocol=Frozen-LVLM, Training-free=false2026.05 | 51.9 | 48.4 | 55.5 | |
| DecAF*Backbone=InternVL3-8B, Training Protocol=Frozen-LVLM, Training-free=true2026.05 | 51.7 | 47.9 | 55.5 | |
| VISABackbone=LLaVA-7B2026.03 | 51 | 49.4 | 52.6 | |
| VISA-7BModel size=7B2025.08 | 50.9 | 49.2 | 52.6 | |
| VISABackbone=Chat-UniVi-7B2026.03 | 50.9 | 49.2 | 52.6 | |
| VISA-7BVenue=ECCV20242026.05 | 50.9 | 49.2 | 52.6 | |
| VISABackbone=ChatUniVi-7B, Protocol=Fully Trained2026.05 | 50.9 | 49.2 | 52.6 | |
| TrackGPTBackbone=LLaVA-13B2026.03 | 49.5 | 48.3 | 50.6 | |
| TrackGPTBackbone=LLaVA-7B2026.03 | 48.2 | 46.7 | 49.7 | |
| Loc-Head*Backbone=InternVL3-8B, Training Protocol=Frozen-LVLM, Training-free=true2026.05 | 46.7 | 42.9 | 50.6 | |
| LISA-13BModel size=13B2025.08 | 46.6 | 45.2 | 47.9 | |
| LISA-7BModel size=7B2025.08 | 45.7 | 44.3 | 47.1 | |
| LISABackbone=LLaVA-7B, Protocol=Fully Trained2026.05 | 45.7 | 44.3 | 47.1 | |
| DecAF*Backbone=LLaVA-OV-7B, Training Protocol=Frozen-LVLM, Training-free=true2026.05 | 43.4 | 39.1 | 47.6 | |
| Loc-Head*Backbone=LLaVA-7B, Training Protocol=Frozen-LVLM, Training-free=true2026.05 | 36.9 | 32.5 | 41.3 | |
| LMPM2025.08 | 34.1 | 29 | 39.1 | |
| LMPMBackbone=Swin-T2026.03 | 34.1 | 29 | 39.1 | |
| ReferFormerBackbone=Video-Swin-B2026.03 | 32.7 | 31.2 | 34.3 | |
| MTTRBackbone=Video-Swin-T2026.03 | 30 | 29.8 | 30.2 |