Table 1 Quantitative comparison with state-of-the-art methods on MIT-Adobe FiveK dataset.
From: Dual-color space network with global priors for photo retouching
Method | PSNR \(\uparrow\) | SSIM \(\uparrow\) | \(\triangle {E}^{*}\) \(\downarrow\) | #params |
|---|---|---|---|---|
White-Box5 | 18.59 | 0.797 | 13.24 | 8,561,762 |
Distort-and-Recover6 | 19.54 | 0.800 | 12.91 | 259,263,320 |
DUPE26 | 20.22 | 0.829 | 13.38 | 998,816 |
Pix2Pix27 | 22.05 | 0.788 | 11.88 | 11,383,427 |
HDRNet4 | 22.65 | 0.880 | 11.64 | 482,080 |
CSRNet23 | 23.62 | 0.894 | 10.97 | 36,489 |
NeurOp24 | 23.87 | 0.899 | 10.48 | 28,108 |
Ours | 24.12 | 0.900 | 10.23 | 92,882 |
