No-Reference Image Quality Assessment on KonIQ (test)
0.906SROCCHyperIQA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HyperIQAModel Type=Supervised pretraining and supervised fine-tuning2021.10 | 0.906 | 0.917 | |
| CONTRIQUEModel Type=Unsupervised pretraining and Linear Regression2021.10 | 0.894 | 0.906 | |
| ResNet-50Model Type=Supervised pretraining and Linear Regression2021.10 | 0.888 | 0.904 | |
| PQRModel Type=Supervised pretraining and supervised fine-tuning2021.10 | 0.88 | 0.884 | |
| DB-CNNModel Type=Supervised pretraining and supervised fine-tuning2021.10 | 0.875 | 0.884 | |
| PaQ-2-PiQModel Type=Supervised pretraining and supervised fine-tuning2021.10 | 0.87 | 0.88 | |
| HOSAModel Type=Codebook-based Features2021.10 | 0.805 | 0.813 | |
| CORNIAModel Type=Codebook-based Features2021.10 | 0.78 | 0.795 | |
| HyperIQATrain on dataset=CLIVE, Cross-dataset evaluation protocol=true2021.08 | 0.772 | — | |
| HyperIQATrain on dataset=LIVEFB, Cross-dataset evaluation protocol=true2021.08 | 0.758 | — | |
| P2P-BMTrain on dataset=LIVEFB, Cross-dataset evaluation protocol=true2021.08 | 0.755 | — | |
| DBCNNTrain on dataset=CLIVE, Cross-dataset evaluation protocol=true2021.08 | 0.754 | — | |
| P2P-BMTrain on dataset=CLIVE, Cross-dataset evaluation protocol=true2021.08 | 0.74 | — | |
| TReSTrain on dataset=CLIVE, Cross-dataset evaluation protocol=true2021.08 | 0.733 | — | |
| DBCNNTrain on dataset=LIVEFB, Cross-dataset evaluation protocol=true2021.08 | 0.716 | — | |
| TReSTrain on dataset=LIVEFB, Cross-dataset evaluation protocol=true2021.08 | 0.713 | — | |
| WaDIQaMTrain on dataset=CLIVE, Cross-dataset evaluation protocol=true2021.08 | 0.711 | — | |
| WaDIQaMTrain on dataset=LIVEFB, Cross-dataset evaluation protocol=true2021.08 | 0.708 | — | |
| BRISQUEModel Type=Traditional/Handcrafted Features2021.10 | 0.665 | 0.681 | |
| NIQEModel Type=Traditional/Handcrafted Features2021.10 | 0.531 | 0.538 |