Low-light image enhancement on LOL and SICE Averaged v1, v2 (test)
0.988Content SimilarityNeural Preset
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Neural Preset2026.05 | 0.988 | 0.942 | 0.179 | 0.176 | 0.41 | 55.61 | |
| SA-LUT2026.05 | 0.879 | 0.671 | 0.413 | 0.386 | 0.325 | 45.78 | |
| OursBackbone=MSCU-Net2026.05 | 0.864 | 0.743 | 0.065 | 0.079 | 0.453 | 63.34 | |
| PhotoWCT22026.05 | 0.857 | 0.778 | 0.077 | 0.114 | 0.429 | 59.29 | |
| SwinFusionFusion Framework=w. our framework2026.05 | 0.844 | 0.696 | 0.052 | 0.06 | 0.43 | 63.31 | |
| CDDFuseFusion Framework=w. our framework2026.05 | 0.83 | 0.65 | 0.069 | 0.078 | 0.45 | 63.18 | |
| CDDFuseFusion Framework=w. fusion losses2026.05 | 0.654 | 0.42 | 0.085 | 0.092 | 0.488 | 66.24 | |
| SwinFusionFusion Framework=w. fusion losses2026.05 | 0.652 | 0.412 | 0.147 | 0.144 | 0.469 | 65.94 | |
| Input/GenAI output2026.05 | 0.593 | 0.313 | 0.336 | 0.311 | 0.418 | 66.64 |