No-Reference Image Quality Assessment on Average across datasets
0.946PLCCMAMIQA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MAMIQA2026.02 | 0.946 | 0.936 | |
| MS-SCANet2026.02 | 0.928 | 0.923 | |
| TreS2026.02 | 0.926 | 0.913 | |
| DBCNN2026.02 | 0.921 | 0.914 | |
| HyperIQA2026.02 | 0.909 | 0.9 | |
| TRIQ2026.02 | 0.892 | 0.879 | |
| DeQAMethod Category=MLLM-based: SFT training2026.06 | 0.838 | 0.813 | |
| MR-IQAMethod Category=MLLM-based: RL training, Backbone=Qwen3-VL-2B, Reproduction Protocol=Controlled reproduction2026.06 | 0.831 | 0.81 | |
| Q-Insight (regression)Method Category=MLLM-based: RL training, Backbone=Qwen2.5-VL-7B, Reproduction Protocol=Reported2026.06 | 0.816 | 0.791 | |
| MEON2026.02 | 0.809 | 0.795 | |
| Q-AlignMethod Category=MLLM-based: SFT training2026.06 | 0.8 | 0.807 | |
| CLIP-IQA+Method Category=Deep-learning-based2026.06 | 0.795 | 0.77 | |
| VQ-R1 (ranking)Method Category=MLLM-based: RL training, Backbone=Qwen3-VL-2B, Reproduction Protocol=Controlled reproduction2026.06 | 0.776 | 0.792 | |
| MUSIQMethod Category=Deep-learning-based2026.06 | 0.775 | 0.753 | |
| C2ScoreMethod Category=MLLM-based: SFT training2026.06 | 0.765 | 0.729 | |
| BRISQUE2026.02 | 0.751 | 0.763 | |
| NIMAMethod Category=Deep-learning-based2026.06 | 0.748 | 0.721 | |
| MANIQAMethod Category=Deep-learning-based2026.06 | 0.719 | 0.692 | |
| DIIVINE2026.02 | 0.708 | 0.708 | |
| DBCNNMethod Category=Deep-learning-based2026.06 | 0.699 | 0.686 | |
| NIQEMethod Category=Hand-crafted2026.06 | 0.575 | 0.535 | |
| BRISQUEMethod Category=Hand-crafted2026.06 | 0.464 | 0.392 |