Motion Amplitude Assessment on UltraVQA
91.4Accuracy@0.5ASO
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ASOModel Category=Ours, Protocol=Analytic Score Optimization2026.02 | 91.4 | 0.287 | 0.926 | 0.916 | |
| GRPOModel Category=Ours, Protocol=Group Relative Policy Optimization2026.02 | 90.4 | 0.296 | 0.921 | 0.907 | |
| SFTModel Category=Ours, Protocol=Supervised Fine-Tuning2026.02 | 80.3 | 0.419 | 0.851 | 0.847 | |
| InternVL-3.5Model Category=Open-source VLMs2026.02 | 71.7 | 0.564 | 0.771 | 0.743 | |
| VideoScoreV2Model Category=VQA Baselines2026.02 | 70.5 | 0.505 | 0.812 | 0.801 | |
| FineVQModel Category=VQA Baselines2026.02 | 69.2 | 0.824 | 0.705 | 0.792 | |
| GPT-4.1Model Category=VLM-APIs2026.02 | 66.8 | 0.595 | 0.697 | 0.693 | |
| MiniCPMV-4.5Model Category=Open-source VLMs2026.02 | 62.2 | 0.704 | 0.651 | 0.648 | |
| Gemini-2.5ProModel Category=VLM-APIs2026.02 | 49.7 | 0.888 | 0.744 | 0.723 | |
| InternVideo-2.5Model Category=Open-source VLMs2026.02 | 49.5 | 0.848 | 0.43 | 0.428 | |
| Qwen2.5-VLModel Category=Open-source VLMs2026.02 | 44.2 | 1.042 | 0.749 | 0.733 | |
| VideoChat-FlashModel Category=Open-source VLMs2026.02 | 44.1 | 1.132 | 0.154 | 0.117 | |
| Q-AlignModel Category=VQA Baselines2026.02 | 43.2 | 1.073 | 0.199 | 0.183 | |
| VideoLLaMA3Model Category=Open-source VLMs2026.02 | 38.8 | 1.14 | 0.641 | 0.615 |