RAW Image Denoising on SIDD (test)
52.55G4 PSNRADL
Evaluation results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ADLIterations=300k, Optimizer=AdamW, Learning Rate=3e-3, Patch Size=256x2562024.11 | 52.55 | 97.5 | 45.18 | 92.3 | 54.37 | 98.7 | 46.13 | 93.2 | 40.16 | 90.1 | 47.68 | 94.4 | |
| Transfer learning2024.11 | 52.28 | 97.4 | 44.96 | 92.3 | 53.04 | 98.2 | 44.77 | 92.3 | 40.1 | 89.8 | 47.03 | 94 | |
| MZSR2024.11 | 51.84 | 97.2 | 44.58 | 92.1 | 53.74 | 98.2 | 45.07 | 92.4 | 37.21 | 86.8 | 46.49 | 93.3 | |
| Blind2Unblind2024.11 | 51.78 | 97 | 44.91 | 91.9 | 54.12 | 98.5 | 46.02 | 92.8 | 38.85 | 89.2 | 47.14 | 93.9 | |
| Prabhakar et al.2024.11 | 51.76 | 97.2 | 44.68 | 91.9 | 53.82 | 98.3 | 44.92 | 92.2 | 38.67 | 87.8 | 46.34 | 93.3 | |
| Fine-tuningTraining Protocol=training on source domain data and then fine-tuning on target domain data2024.11 | 50.17 | 96.8 | 43.53 | 91.4 | 52.77 | 97.7 | 43.86 | 91.7 | 37.88 | 86.3 | 45.58 | 92.8 | |
| BM3D2024.11 | 50.08 | 96.8 | 42.14 | 90.9 | 52.39 | 97.2 | 43.4 | 91.6 | 35.52 | 85.5 | 44.71 | 92.4 | |
| ZS-N2N2024.11 | 48.86 | 94.1 | 41.54 | 90.9 | 50.06 | 96.8 | 41.88 | 91 | 35.07 | 85.6 | 43.48 | 91.7 | |
| DIP2024.11 | 46.91 | 93.1 | 39.88 | 89.6 | 48.81 | 95.5 | 41.73 | 90.6 | 35.23 | 85.5 | 42.51 | 90.9 |