Clarity Quality Assessment on UltraVQA
86.7Acc@0.5ASO
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ASOModel Category=Ours, Protocol=Analytic Score Optimization2026.02 | 86.7 | 0.387 | 0.898 | 0.885 | |
| GRPOModel Category=Ours, Protocol=Group Relative Policy Optimization2026.02 | 85 | 0.369 | 0.895 | 0.881 | |
| SFTModel Category=Ours, Protocol=Supervised Fine-Tuning2026.02 | 83.4 | 0.428 | 0.867 | 0.859 | |
| GPT-4.1Model Category=VLM-APIs2026.02 | 75.3 | 0.514 | 0.848 | 0.839 | |
| VideoScoreV2Model Category=VQA Baselines2026.02 | 75.3 | 0.881 | 0.791 | 0.776 | |
| FineVQModel Category=VQA Baselines2026.02 | 74.1 | 0.998 | 0.782 | 0.767 | |
| Gemini-2.5ProModel Category=VLM-APIs2026.02 | 65.8 | 0.611 | 0.85 | 0.824 | |
| Qwen2.5-VLModel Category=Open-source VLMs2026.02 | 54.4 | 0.786 | 0.793 | 0.799 | |
| InternVideo-2.5Model Category=Open-source VLMs2026.02 | 50.7 | 0.835 | 0.685 | 0.685 | |
| MiniCPMV-4.5Model Category=Open-source VLMs2026.02 | 47.1 | 0.9 | 0.775 | 0.774 | |
| VideoLLaMA3Model Category=Open-source VLMs2026.02 | 41.4 | 1.01 | 0.755 | 0.771 | |
| InternVL-3.5Model Category=Open-source VLMs2026.02 | 40.9 | 0.974 | 0.817 | 0.86 | |
| Q-AlignModel Category=VQA Baselines2026.02 | 36.5 | 0.942 | 0.876 | 0.871 | |
| VideoChat-FlashModel Category=Open-source VLMs2026.02 | 36.4 | 1.082 | 0.643 | 0.574 |