Exposure Correction on Introduced +0, +1, and +1.5 relative EVs 1.0 (test)
18.976PSNR (Expert A)coarse-to-fine DNN
Evaluation results
| Method | Links | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| coarse-to-fine DNNAdversarial Loss=w/o Ladv2020.03 | 18.976 | 74.3 | 19.767 | 73.1 | 19.98 | 76.8 | 18.966 | 71.6 | 19.056 | 72.7 | 19.349 | 73.7 | 2.189 | |
| coarse-to-fine DNNAdversarial Loss=w/ Ladv2020.03 | 18.874 | 73.8 | 19.569 | 71.8 | 19.788 | 76 | 18.823 | 70.5 | 18.936 | 71.9 | 19.198 | 72.8 | 2.183 | |
| DPEDModel Backbone=Sony2020.03 | 18.114 | 68.3 | 18.682 | 72 | 18.513 | 68.7 | 18.814 | 70.1 | 19.297 | 71.1 | 18.685 | 70 | 2.74 | |
| HEnon-deep learning=true2020.03 | 16.14 | 68.6 | 16.277 | 67.2 | 16.531 | 69.9 | 16.643 | 66.9 | 17.321 | 69.1 | 16.582 | 68.3 | 2.351 | |
| DPEDModel Backbone=BlackBerry2020.03 | 15.17 | 62.1 | 16.193 | 69.1 | 15.781 | 64.2 | 17.042 | 67.7 | 18.035 | 67.8 | 16.444 | 66.2 | 2.518 | |
| HDR CNNRefining Module=PS2020.03 | 14.804 | 65.1 | 15.622 | 68.9 | 15.348 | 67 | 16.583 | 68.5 | 18.022 | 70.3 | 16.076 | 68 | 2.248 | |
| DPEVariant=S-FiveK, Supervision=Supervised2020.03 | 14.786 | 63.8 | 15.519 | 64.9 | 15.625 | 66.8 | 16.586 | 66.4 | 17.661 | 68.4 | 16.035 | 66.1 | 2.621 | |
| DPEVariant=HDR2020.03 | 14.399 | 57.2 | 15.219 | 57.3 | 15.091 | 59.3 | 15.692 | 58.1 | 16.64 | 62.6 | 15.408 | 58.9 | 2.417 | |
| DPEVariant=U-FiveK, Supervision=Unsupervised2020.03 | 14.314 | 61.5 | 14.958 | 62.8 | 15.075 | 64.5 | 15.987 | 64.7 | 16.931 | 66.7 | 15.453 | 64 | 2.63 | |
| CLAHEnon-deep learning=true2020.03 | 13.934 | 56.8 | 14.689 | 58.6 | 14.453 | 58.4 | 15.116 | 59.3 | 15.85 | 61.2 | 14.808 | 58.9 | 2.27 | |
| HDR CNNRefining Module=RHT2020.03 | 13.151 | 47.5 | 13.637 | 47.8 | 13.622 | 49.7 | 14.177 | 47.9 | 14.625 | 50.3 | 13.842 | 48.6 | 4.284 | |
| DPEDModel Backbone=iPhone2020.03 | 12.68 | 56.2 | 13.422 | 58.6 | 13.135 | 58.1 | 14.477 | 59.6 | 15.702 | 63 | 13.883 | 59.1 | 2.909 | |
| WVMnon-deep learning=true2020.03 | 12.355 | 62.4 | 13.147 | 65.6 | 12.748 | 64.5 | 14.059 | 66.9 | 15.207 | 69 | 13.503 | 65.7 | 2.342 | |
| HQECnon-deep learning=true2020.03 | 11.775 | 60.7 | 12.536 | 63.1 | 12.127 | 62.7 | 13.424 | 65.2 | 14.511 | 67.5 | 12.875 | 63.8 | 2.387 | |
| RetinexNetoptimized for=underexposure correction2020.03 | 10.149 | 57 | 10.88 | 58.6 | 10.471 | 59.5 | 11.498 | 61.3 | 12.295 | 63.5 | 11.059 | 60 | 2.933 | |
| Zero-DCEoptimized for=underexposure correction2020.03 | 10.116 | 50.3 | 10.767 | 50.2 | 10.395 | 51.4 | 11.471 | 52.2 | 12.354 | 55.7 | 11.026 | 51.96 | 2.774 | |
| Deep UPEoptimized for=underexposure correction2020.03 | 10.047 | 53.2 | 10.462 | 56.8 | 10.307 | 55.7 | 11.583 | 59.1 | 12.639 | 61.9 | 11.008 | 57.3 | 2.428 | |
| LIMEnon-deep learning=true, optimized for=underexposure correction2020.03 | 9.627 | 54.9 | 10.096 | 56.9 | 9.875 | 57 | 10.936 | 59.7 | 11.903 | 62.6 | 10.487 | 58.2 | 2.412 |