Image Quality Assessment on KonIQ-10k (test)
0.949SRCCEffnet-2C-MLSP
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Effnet-2C-MLSPTraining Dataset=KonIQ-10k, Backbone=EfficientNet-B7, Column Architecture=Two-column2022.12 | 0.949 | 0.9596 | |
| NTPCoT training=true2025.03 | 0.948 | 0.959 | |
| RvTCinput_modality=image-only2025.07 | 0.942 | 0.952 | |
| Q-Aligninput_modality=image-only2025.07 | 0.94 | 0.941 | |
| Q-ALIGNTraining Set=KonIQ, #Training=7K (70%), Evaluation Protocol=Intra-set2023.12 | 0.94 | 0.941 | |
| Q-AlignCoT training=false2025.03 | 0.94 | 0.941 | |
| NTPCoT training=false2025.03 | 0.939 | 0.949 | |
| Norm-in-NormLoss=l + 0.1l'2020.08 | 0.938 | 0.947 | |
| Norm-in-NormLoss=l2020.08 | 0.937 | 0.947 | |
| QCNCoT training=false2025.03 | 0.934 | 0.945 | |
| LoDaCoT training=false2025.03 | 0.932 | 0.944 | |
| LinearityIQATraining Dataset=KonIQ-10k2022.12 | 0.9299 | 0.9415 | |
| MUSIQTraining Set=KonIQ, #Training=7K (70%), Evaluation Protocol=Intra-set2023.12 | 0.929 | 0.924 | |
| LIQETraining Set=KonIQ, #Training=7K (70%), Evaluation Protocol=Intra-set2023.12 | 0.928 | 0.912 | |
| LIQECoT training=false2025.03 | 0.928 | 0.912 | |
| MUSIQ2023.03 | 0.924 | 0.937 | |
| KonCept512Loss=MSE loss2020.08 | 0.921 | 0.937 | |
| KonCept5122023.03 | 0.921 | 0.937 | |
| LIQEinput_modality=image-only2025.07 | 0.919 | 0.908 | |
| VILA-R2023.03 | 0.919 | 0.932 | |
| MUSIQinput_modality=image-only2025.07 | 0.916 | 0.928 | |
| MUSIQScale=Multi-scale2021.08 | 0.916 | 0.928 | |
| DSMixCoT training=false2025.03 | 0.915 | 0.925 | |
| Kimi-VL-A3B-InstructRatio=1:9, IQARAG integration=true2026.01 | 0.911 | 0.9121 | |
| Kimi-VL-A3B-InstructRatio=1:4, IQARAG integration=false2026.01 | 0.9098 | 0.8943 | |
| Kimi-VL-A3B-InstructRatio=3:7, IQARAG integration=false2026.01 | 0.9098 | 0.892 | |
| Kimi-VL-A3B-InstructRatio=1:9, IQARAG integration=false2026.01 | 0.9094 | 0.8938 | |
| Kimi-VL-A3B-InstructRatio=1:4, IQARAG integration=true2026.01 | 0.9084 | 0.9055 | |
| Kimi-VL-A3B-InstructRatio=3:7, IQARAG integration=true2026.01 | 0.9077 | 0.9058 | |
| BIQA2023.03 | 0.906 | 0.917 | |
| HyperIQATraining Set=KonIQ, #Training=7K (70%), Evaluation Protocol=Intra-set2023.12 | 0.906 | 0.917 | |
| HyperIQACoT training=false2025.03 | 0.906 | 0.917 | |
| BIQANumber of crops=25 crops2021.08 | 0.906 | 0.917 | |
| MUSIQ-singleScale=Single-scale2021.08 | 0.905 | 0.919 | |
| FEWSHOT-Q-ALIGNTraining Set=KonIQ, #Training=2K (20%), Evaluation Protocol=Intra-set2023.12 | 0.903 | 0.901 | |
| CLIP-IQA+2023.03 | 0.895 | 0.909 | |
| CLIP-IQA+Training Set=KonIQ, #Training=7K (70%), Evaluation Protocol=Intra-set2023.12 | 0.895 | 0.909 | |
| CLIP-IQA+CoT training=false2025.03 | 0.895 | 0.909 | |
| PQR2023.03 | 0.88 | 0.884 | |
| PQR2021.08 | 0.88 | 0.884 | |
| DBCNN2023.03 | 0.875 | 0.884 | |
| DBCNNTraining Set=KonIQ, #Training=7K (70%), Evaluation Protocol=Intra-set2023.12 | 0.875 | 0.884 | |
| DBCNN2021.08 | 0.875 | 0.884 | |
| DeepBIQLoss=SVR loss2020.08 | 0.872 | 0.886 | |
| DeepRNLoss=Huber loss2020.08 | 0.867 | 0.88 | |
| IRN-1C-MLSPTraining Dataset=KonIQ-10k, Backbone=InceptionResNetV2, Column Architecture=One-column2022.12 | 0.8601 | 0.8932 | |
| NIMAinput_modality=image-only2025.07 | 0.859 | 0.896 | |
| NIMATraining Set=KonIQ, #Training=7K (70%), Evaluation Protocol=Intra-set2023.12 | 0.859 | 0.896 | |
| SFA2023.03 | 0.856 | 0.872 | |
| SFA2021.08 | 0.856 | 0.872 | |
| MetaIQA2023.03 | 0.85 | 0.887 | |
| MetaIQA2021.08 | 0.85 | 0.887 | |
| Q-ALIGNTraining Set=SPAQ, #Training=8.8K (80%), Evaluation Protocol=Cross-set2023.12 | 0.848 | 0.879 | |
| LIQETraining Set=SPAQ, #Training=8.8K (80%), Evaluation Protocol=Cross-set2023.12 | 0.826 | 0.847 | |
| Qwen3-VL-8B-InstructRatio=1:9, IQARAG integration=true2026.01 | 0.8071 | 0.8036 | |
| HOSALoss=SVR loss2020.08 | 0.805 | 0.828 | |
| Qwen3-VL-8B-InstructRatio=3:7, IQARAG integration=true2026.01 | 0.7985 | 0.7976 | |
| WaDIQaM2023.03 | 0.797 | 0.805 | |
| WaDIQaM2021.08 | 0.797 | 0.805 | |
| FEWSHOT-Q-ALIGNTraining Set=SPAQ, #Training=2.2K (20%), Evaluation Protocol=Cross-set2023.12 | 0.792 | 0.826 | |
| Qwen3-VL-8B-InstructRatio=1:4, IQARAG integration=true2026.01 | 0.79 | 0.7963 | |
| CORNIALoss=SVR loss2020.08 | 0.78 | 0.808 | |
| HyperIQATrain Set=CLIVE2023.03 | 0.772 | — | |
| Qwen3-VL-8B-InstructRatio=1:9, IQARAG integration=false2026.01 | 0.7712 | 0.809 | |
| Qwen3-VL-8B-InstructRatio=1:4, IQARAG integration=false2026.01 | 0.7692 | 0.8081 | |
| Qwen3-VL-8B-InstructRatio=3:7, IQARAG integration=false2026.01 | 0.7671 | 0.8048 | |
| Effnet-NIMATraining Dataset=KonIQ-10k, Backbone=EfficientNet2022.12 | 0.7635 | 0.7788 | |
| DBCNNTrain Set=CLIVE2023.03 | 0.754 | — | |
| MUSIQTraining Set=SPAQ, #Training=8.8K (80%), Evaluation Protocol=Cross-set2023.12 | 0.753 | 0.68 | |
| CLIP-IQA+Training Set=SPAQ, #Training=8.8K (80%), Evaluation Protocol=Cross-set2023.12 | 0.753 | 0.777 | |
| QPTTrain Set=CLIVE2023.03 | 0.7494 | — | |
| NIMATraining Set=SPAQ, #Training=8.8K (80%), Evaluation Protocol=Cross-set2023.12 | 0.733 | 0.788 | |
| DBCNNTraining Set=SPAQ, #Training=8.8K (80%), Evaluation Protocol=Cross-set2023.12 | 0.731 | 0.758 | |
| Fang et al.Training Set=SPAQ, #Training=8.8K (80%), Evaluation Protocol=Cross-set2023.12 | 0.714 | 0.742 | |
| BRISQUELoss=SVR loss2020.08 | 0.705 | 0.707 | |
| InternVL3.5-8BRatio=1:4, IQARAG integration=true2026.01 | 0.685 | 0.7381 | |
| CONTRIQUETrain Set=CLIVE2023.03 | 0.676 | — | |
| HOSA2023.03 | 0.671 | 0.694 | |
| HOSA2021.08 | 0.671 | 0.694 | |
| BRISQUE2023.03 | 0.665 | 0.681 | |
| BRISQUE2021.08 | 0.665 | 0.681 | |
| InternVL3.5-8BRatio=3:7, IQARAG integration=true2026.01 | 0.6517 | 0.684 | |
| InternVL3.5-8BRatio=1:9, IQARAG integration=true2026.01 | 0.6268 | 0.6656 | |
| BIECON2023.03 | 0.618 | 0.651 | |
| BIECON2021.08 | 0.618 | 0.651 | |
| InternVL3.5-8BRatio=1:4, IQARAG integration=false2026.01 | 0.6056 | 0.6506 | |
| InternVL3.5-8BRatio=1:9, IQARAG integration=false2026.01 | 0.6046 | 0.6518 | |
| InternVL3.5-8BRatio=3:7, IQARAG integration=false2026.01 | 0.5982 | 0.6378 | |
| CNNIQALoss=MAE loss2020.08 | 0.572 | 0.584 | |
| ILNIQE2023.03 | 0.507 | 0.523 | |
| ILNIQE2021.08 | 0.507 | 0.523 |