Blind Image Quality Assessment on BID
0.91SRCCDEFNet
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DEFNet2025.07 | 0.91 | 0.909 | |
| DR.Expertspre-training=false2026.02 | 0.896 | 0.919 | |
| QCNpre-training=false2026.02 | 0.892 | 0.89 | |
| QPTpre-training=true2026.02 | 0.888 | 0.911 | |
| QPT-ResNet50backbone=ResNet502023.03 | 0.8875 | 0.9109 | |
| LIQE2025.07 | 0.875 | 0.9 | |
| CDINet2025.07 | 0.874 | 0.899 | |
| DPSF2025.07 | 0.872 | 0.883 | |
| HyperIQApre-training=false2026.02 | 0.869 | 0.878 | |
| DBCNN2025.07 | 0.864 | 0.883 | |
| UNIQUE2023.03 | 0.858 | 0.873 | |
| HyperIQA2023.03 | 0.8544 | 0.8585 | |
| TreS2025.07 | 0.853 | 0.871 | |
| UNIQUE2025.07 | 0.852 | 0.875 | |
| HyperIQA2025.07 | 0.848 | 0.868 | |
| DB-CNN2023.03 | 0.845 | 0.859 | |
| DB-CNNpre-training=false2026.02 | 0.845 | 0.85 | |
| ResNet50backbone=ResNet502023.03 | 0.8423 | 0.8521 | |
| Tool-IQATraining dataset=KADID-10K, Method category=VLM-based2026.06 | 0.828 | 0.842 | |
| SFA2023.03 | 0.8202 | 0.8253 | |
| MG-IQAmulti-dataset training=true, training_data=KADID-10k + SPAQ2026.04 | 0.795 | 0.818 | |
| SynDR-IQATraining Dataset=KADID-10k2026.01 | 0.788 | 0.764 | |
| VisualQuality-R1Training dataset=KADID-10K, Method category=VLM-based2026.06 | 0.788 | 0.807 | |
| QInsightTraining dataset=KADID-10K, Method category=VLM-based2026.06 | 0.784 | 0.796 | |
| Qwen3-VL-8B-InsTraining dataset=KADID-10K, Method category=VLM-based2026.06 | 0.779 | 0.725 | |
| MG-IQATraining Dataset=KADID-10k, Evaluation=Zero-shot2026.04 | 0.778 | 0.801 | |
| VQ-R1multi-dataset training=true, training_data=KADID-10k + SPAQ2026.04 | 0.774 | 0.798 | |
| DGQATraining Dataset=KADID-10k2026.01 | 0.77 | 0.753 | |
| VQ-R1Training Dataset=KADID-10k, Evaluation=Zero-shot2026.04 | 0.758 | 0.782 | |
| Q-Insightmulti-dataset training=true, training_data=KADID-10k + SPAQ2026.04 | 0.755 | 0.78 | |
| Q-InsightTraining Dataset=KADID-10k, Evaluation=Zero-shot2026.04 | 0.751 | 0.776 | |
| DeQA-ScoreTraining Dataset=KADID-10k, Evaluation=Zero-shot2026.04 | 0.739 | 0.761 | |
| Q-AlignTraining Dataset=KADID-10k, Evaluation=Zero-shot2026.04 | 0.731 | 0.755 | |
| LIQETraining Dataset=KADID-10k, Evaluation=Zero-shot2026.04 | 0.718 | 0.743 | |
| TOPIQTraining Dataset=KADID-10k, Evaluation=Zero-shot2026.04 | 0.711 | 0.737 | |
| Qwen2.5-VL-7BTraining dataset=KADID-10K, Method category=VLM-based2026.06 | 0.711 | 0.725 | |
| MANIQATraining Dataset=KADID-10k, Evaluation=Zero-shot2026.04 | 0.702 | 0.729 | |
| DeQA-ScoreTraining dataset=KADID-10K, Method category=VLM-based2026.06 | 0.702 | 0.743 | |
| Re-IQATraining Dataset=KADID-10k, Evaluation=Zero-shot2026.04 | 0.698 | 0.717 | |
| MUSIQTraining Dataset=KADID-10k, Evaluation=Zero-shot2026.04 | 0.694 | 0.718 | |
| UNIQUETraining Dataset=KADID-10k, Evaluation=Zero-shot2026.04 | 0.687 | 0.702 | |
| LIQETraining dataset=KADID-10K, Method category=VLM-based2026.06 | 0.677 | 0.68 | |
| FreqAlignTraining Dataset=KADID-10k2026.01 | 0.674 | 0.708 | |
| WaDIQaM-NRpre-training=false2026.02 | 0.653 | 0.636 | |
| WaDIQaM-NR2023.03 | 0.6526 | 0.6359 | |
| MUSIQpre-training=false2026.02 | 0.646 | 0.739 | |
| CLIP-IQATraining Dataset=KADID-10k, Evaluation=Zero-shot2026.04 | 0.645 | 0.671 | |
| StyleAMTraining Dataset=KADID-10k2026.01 | 0.637 | 0.567 | |
| DBCNNTraining Dataset=KADID-10k2026.01 | 0.62 | 0.609 | |
| CNNIQA2023.03 | 0.6163 | 0.6144 | |
| DANNTraining Dataset=KADID-10k2026.01 | 0.586 | 0.51 | |
| Q-AlignTraining dataset=KADID-10K, Method category=VLM-based2026.06 | 0.576 | 0.651 | |
| MUSIQTraining Dataset=KADID-10k2026.01 | 0.575 | 0.6 | |
| BRISQUEpre-training=false2026.02 | 0.574 | 0.54 | |
| BRISQUE2023.03 | 0.5736 | 0.5401 | |
| NIQE2025.07 | 0.573 | 0.618 | |
| BRISQUETraining Dataset=KADID-10k, Evaluation=Zero-shot2026.04 | 0.571 | 0.598 | |
| ILNIQE2025.07 | 0.548 | 0.494 | |
| VCRNetTraining Dataset=KADID-10k2026.01 | 0.542 | 0.545 | |
| RankDATraining Dataset=KADID-10k2026.01 | 0.535 | 0.582 | |
| DEIQTTraining dataset=KADID-10K, Method category=Discriminative Deep-Learning-based2026.06 | 0.534 | 0.539 | |
| BRISQUETraining dataset=KADID-10K, Method category=Handcrafted2026.06 | 0.522 | 0.528 | |
| BMPRI2023.03 | 0.5154 | 0.4583 | |
| NIQETraining dataset=KADID-10K, Method category=Handcrafted2026.06 | 0.515 | 0.527 | |
| RankIQATraining Dataset=KADID-10k2026.01 | 0.51 | 0.367 | |
| ILNIQEpre-training=false2026.02 | 0.495 | 0.454 | |
| ILNIQE2023.03 | 0.4946 | 0.4538 | |
| NIQE2023.03 | 0.4772 | 0.4713 | |
| CLIPIQA+Training Dataset=KADID-10k2026.01 | 0.474 | 0.442 | |
| NIQETraining Dataset=KADID-10k, Evaluation=Zero-shot2026.04 | 0.465 | 0.496 | |
| MANIQATraining dataset=KADID-10K, Method category=Discriminative Deep-Learning-based2026.06 | 0.42 | 0.512 | |
| UNIQUETraining dataset=KADID-10K, Method category=Discriminative Deep-Learning-based2026.06 | 0.412 | 0.385 | |
| HyperIQATraining Dataset=KADID-10k2026.01 | 0.379 | 0.282 | |
| Ma192025.07 | 0.373 | 0.399 | |
| UCDATraining Dataset=KADID-10k2026.01 | 0.348 | 0.391 | |
| MUSIQTraining dataset=KADID-10K, Method category=Discriminative Deep-Learning-based2026.06 | 0.327 | 0.28 | |
| dipIQ2025.07 | 0.009 | 0.346 |