Video Object Segmentation on ReVOS (Overall)
61.3J&F ScoreRefer-Agent
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Refer-AgentMLLM=Ovis2.5-9B, Protocol=Zero-shot2026.02 | 61.3 | 58.3 | 64.3 | 28.6 | |
| Veason-R1MLLM=Qwen2.5VL-7B, Training Protocol=Training-based2025.10 | 61.3 | 58.2 | 64.4 | — | |
| VRS-HQMLLM=Chat-UniVi-13B, Protocol=SFT-based2026.02 | 60 | 57.6 | 62.5 | 18.9 | |
| VRS-HQMLLM=ChatUniVi-7B, Training Protocol=Training-based2025.10 | 59.1 | 56.6 | 61.6 | — | |
| RGA3MLLM=Qwen2.5-VL-7B, Protocol=SFT-based2026.02 | 58 | 55.9 | 60 | 28.6 | |
| ViLLaMLLM=InternVideo2-6B, Protocol=SFT-based2026.02 | 57 | 54.9 | 59.1 | — | |
| CoT-RVSMLLM=GPT-4o, Protocol=Zero-shot2026.02 | 55.9 | 52.8 | 59 | — | |
| HyperSegMLLM=Mipha-3B, Protocol=SFT-based2026.02 | 55.7 | 53.1 | 58.4 | — | |
| GLUSMLLM=LISA-7B, Protocol=SFT-based2026.02 | 54.9 | 52.4 | 57.3 | 17.9 | |
| GLUSMLLM=LLaVA-7B, Training Protocol=Training-based2025.10 | 54.9 | 52.4 | 57.3 | — | |
| InstructSegMLLM=Mipha-3B, Protocol=SFT-based2026.02 | 54.5 | 52 | 56.9 | — | |
| DecAFMLLM=Qwen2.5VL-7B, Training Protocol=Training-free2025.10 | 54.2 | 50.1 | 58.2 | — | |
| VISAMLLM=Chat-UniVi-13B, Protocol=SFT-based2026.02 | 50.9 | 48.8 | 52.9 | 14.5 | |
| DecAFMLLM=InternVL3-8B, Training Protocol=Training-free2025.10 | 47.4 | 43.7 | 51.2 | — | |
| CoT-RVSMLLM=Gemma3-12B, Protocol=Zero-shot2026.02 | 47.1 | 43.4 | 50.9 | — | |
| Loc-HeadMLLM=Qwen2.5VL-7B, Training Protocol=Training-free2025.10 | 47 | 43.3 | 50.7 | — | |
| VISAMLLM=ChatUniVi-7B, Training Protocol=Training-based2025.10 | 46.9 | 44.9 | 49 | — | |
| DecAFMLLM=Qwen2VL-7B, Training Protocol=Training-free2025.10 | 45.3 | 41.6 | 49 | — | |
| Loc-HeadMLLM=Qwen2VL-7B, Training Protocol=Training-free2025.10 | 44 | 40.8 | 47.2 | — | |
| Loc-HeadMLLM=InternVL3-8B, Training Protocol=Training-free2025.10 | 43.7 | 39.9 | 47.5 | — | |
| LISAMLLM=LLaVA-7B, Training Protocol=Training-based2025.10 | 40.9 | 39.1 | 42.7 | — | |
| DecAFMLLM=LLaVA-OV-7B, Training Protocol=Training-free2025.10 | 40 | 35.8 | 44.1 | — | |
| Loc-HeadMLLM=LLaVA-7B, Training Protocol=Training-free2025.10 | 35.3 | 31.1 | 39.6 | — | |
| Loc-HeadMLLM=LLaVA-OV-7B, Training Protocol=Training-free2025.10 | 31.7 | 28.3 | 35.1 | — |