Blind Image Quality Assessment on KonIQ
0.941SRCCDR.Experts
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DR.Expertspre-training=false2026.02 | 0.941 | 0.954 | |
| DeQAMethod Category=MLLM-based: SFT training2026.06 | 0.941 | 0.953 | |
| GLIANetNumber of trainable parameters=8M2026.05 | 0.94 | 0.948 | |
| Q-AlignMethod Category=MLLM-based: SFT training2026.06 | 0.94 | 0.941 | |
| SHDIQANumber of trainable parameters=24M2026.05 | 0.937 | 0.948 | |
| QCNpre-training=false2026.02 | 0.934 | 0.945 | |
| LODApre-training=false2026.02 | 0.932 | 0.944 | |
| LoDaNumber of trainable parameters=9M2026.05 | 0.932 | 0.944 | |
| MR-IQAMethod Category=MLLM-based: RL training, Backbone=Qwen3-VL-2B, Reproduction Protocol=Controlled reproduction2026.06 | 0.931 | 0.949 | |
| MUSIQMethod Category=Deep-learning-based2026.06 | 0.929 | 0.924 | |
| LQMambapre-training=false2026.02 | 0.928 | 0.943 | |
| QMambaNumber of trainable parameters=50M2026.05 | 0.928 | 0.943 | |
| QPTpre-training=true2026.02 | 0.927 | 0.941 | |
| QFM-IQMpre-training=false2026.02 | 0.922 | 0.936 | |
| DEIQTpre-training=false2026.02 | 0.921 | 0.934 | |
| LIQENumber of trainable parameters=151M2026.05 | 0.919 | 0.908 | |
| VQ-R1 (ranking)Method Category=MLLM-based: RL training, Backbone=Qwen3-VL-2B, Reproduction Protocol=Controlled reproduction2026.06 | 0.919 | 0.886 | |
| MUSIQpre-training=false2026.02 | 0.916 | 0.928 | |
| MUSIQ2026.05 | 0.916 | 0.928 | |
| TReSpre-training=false2026.02 | 0.915 | 0.928 | |
| TReS2026.05 | 0.915 | 0.928 | |
| DAIQA2026.05 | 0.915 | 0.925 | |
| Re-IQANumber of trainable parameters=48M2026.05 | 0.914 | 0.923 | |
| C2ScoreMethod Category=MLLM-based: SFT training2026.06 | 0.91 | 0.923 | |
| HyperIQApre-training=false2026.02 | 0.906 | 0.917 | |
| HyperIQA2026.05 | 0.906 | 0.917 | |
| CLIP-IQA+Method Category=Deep-learning-based2026.06 | 0.895 | 0.909 | |
| Q-Insight (regression)Method Category=MLLM-based: RL training, Backbone=Qwen2.5-VL-7B, Reproduction Protocol=Reported2026.06 | 0.895 | 0.918 | |
| CONRTIQUEpre-training=true2026.02 | 0.894 | 0.906 | |
| MetaIQA2026.05 | 0.887 | 0.856 | |
| DB-CNNpre-training=false2026.02 | 0.878 | 0.887 | |
| DBCNN2026.05 | 0.875 | 0.884 | |
| DBCNNMethod Category=Deep-learning-based2026.06 | 0.875 | 0.884 | |
| P2P-BM2026.05 | 0.872 | 0.885 | |
| NIMAMethod Category=Deep-learning-based2026.06 | 0.859 | 0.896 | |
| MANIQAMethod Category=Deep-learning-based2026.06 | 0.834 | 0.849 | |
| WaDIQaM-NRpre-training=false2026.02 | 0.729 | 0.754 | |
| BRISQUEpre-training=false2026.02 | 0.715 | 0.702 | |
| NIQEMethod Category=Hand-crafted2026.06 | 0.53 | 0.533 | |
| ILNIQE2026.05 | 0.523 | 0.537 | |
| ILNIQEpre-training=false2026.02 | 0.503 | 0.496 | |
| BRISQUEMethod Category=Hand-crafted2026.06 | 0.226 | 0.225 |