Table 1 Quantitative comparison results on the LOL-V125, LOL-V226and mixed datasets. Bold indicates the best results, and italic indicates the second-best results. LIME18represents a traditional method, ZeroDec++7and PairLIE54 represent unsupervised learning methods, and the others represent supervised learning methods.
Methods | LOL-V1 | LOL-V2 | Mix Data | ||||
|---|---|---|---|---|---|---|---|
PSNR \(\uparrow\) | SSIM\(\uparrow\) | NIQE \(\downarrow\) | PSNR\(\uparrow\) | SSIM\(\uparrow\) | NIQE\(\downarrow\) | NIQE\(\downarrow\) | |
LIME18 | 17.22 | 0.50 | 5.32 | 15.77 | 0.46 | 5.37 | 4.57 |
ZeroDec++7 | 15.35 | 0.57 | 7.86 | 18.49 | 0.58 | 8.05 | 4.53 |
PairLIE54 | 18.47 | 0.75 | 4.25 | 19.88 | 0.78 | 4.34 | 3.90 |
RetinexNet25 | 17.86 | 0.78 | 6.37 | 17.37 | 0.76 | 9.09 | 5.68 |
MBLLEN8 | 17.90 | 0.70 | 2.50 | 18.00 | 0.72 | 3.11 | 3.32 |
KIND9 | 20.38 | 0.83 | 5.45 | 23.78 | 0.88 | 4.96 | 3.87 |
KIND++55 | 21.80 | 0.84 | 5.17 | 22.21 | 0.84 | 4.89 | 3.74 |
IAT56 | 23.38 | 0.81 | 3.92 | 23.50 | 0.82 | 4.19 | 4.71 |
DecNet57 | 22.49 | 0.82 | 4.51 | 22.56 | 0.84 | 4.83 | 4.26 |
FLW-Net51 | 23.84 | 0.83 | 4.22 | 25.71 | 0.87 | 4.09 | 3.93 |
BiEnNet (LOL-V1) | 25.88 | 0.86 | 2.75 | 29.06 | 0.89 | 3.08 | 3.49 |
BiEnNet (LOL-V2) | 26.44 | 0.87 | 2.63 | 29.66 | 0.90 | 3.01 | 3.48 |
