Perception Visual Reasoning on SVU-31K Fine-grained Perception Visual Reasoning 1.0 (test)
18.9BLEU-4MedScope
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MedScopeTool Use=true, Reasoning=true, Zero-shot=true2026.02 | 18.9 | 10.93 | 21.62 | 41.54 | |
| SurgVidLMTool Use=false, Reasoning=false, Zero-shot=true2026.02 | 16.71 | 9.76 | 18.27 | 37.51 | |
| Video-LLaVA-7BTool Use=false, Reasoning=false, Zero-shot=true2026.02 | 10.76 | 5.48 | 13.52 | 25.93 | |
| VideoLLaMA3-7BTool Use=false, Reasoning=false, Zero-shot=true2026.02 | 9.57 | 3.97 | 12.63 | 24.21 | |
| VideoR1-7BTool Use=false, Reasoning=true, Zero-shot=true2026.02 | 7.45 | 5.13 | 13.91 | 23.34 | |
| VideoRFT-7BTool Use=true, Reasoning=true, Zero-shot=true2026.02 | 6.46 | 3.61 | 16.9 | 23.78 | |
| VideoChat-R1-7BTool Use=true, Reasoning=true, Zero-shot=true2026.02 | 5.9 | 1.93 | 17.62 | 24.54 | |
| LongVT-7B-RFTTool Use=true, Reasoning=true, Zero-shot=true2026.02 | 5.41 | 2 | 19.53 | 25.59 | |
| GroundingGPT-7BTool Use=false, Reasoning=false, Zero-shot=true2026.02 | 4.47 | 1.51 | 11.96 | 21.34 | |
| Qwen2.5-VL-7BTool Use=false, Reasoning=false, Zero-shot=true2026.02 | 3.19 | 2.18 | 15.04 | 25.55 | |
| InternVL3-8BTool Use=false, Reasoning=false, Zero-shot=true2026.02 | 2.93 | 1.89 | 12.2 | 26.72 |