Image Denoising on Kodak Gaussian Noise (sigma=50)
29.78PSNRMPRnet
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MPRnetType=Supervised2021.11 | 29.78 | 0.817 | |
| IDRType=Unsupervised2021.11 | 29.27 | 0.803 | |
| SupervisedType=Supervised2021.11 | 29.24 | 0.803 | |
| N2N2021.11 | 29.23 | 0.803 | |
| DBSNBackbone=10x larger (97.4G FLOPS)2021.11 | 28.81 | 0.783 | |
| Nr2n2021.11 | 28.73 | 0.77 | |
| SLLearning Paradigm=Supervised2021.12 | 28.71 | — | |
| N2NLearning Paradigm=Supervised2021.12 | 28.7 | — | |
| OursLearning Paradigm=Self-Supervised2021.12 | 28.64 | — | |
| BM3D2021.11 | 28.64 | 0.772 | |
| Laine19-pmeLearning Paradigm=Self-Supervised2021.12 | 28.63 | — | |
| N2V2021.11 | 28.57 | 0.776 | |
| Nei2NeiLearning Paradigm=Self-Supervised2021.12 | 28.28 | — | |
| N2ScoreLearning Paradigm=Self-Supervised2021.12 | 28.23 | — | |
| N2SLearning Paradigm=Self-Supervised2021.12 | 28.21 | — | |
| Laine19-muLearning Paradigm=Self-Supervised2021.12 | 27.78 | — | |
| DILTraining=Multiple noise levels required2024.03 | 27.46 | 0.736 | |
| CLIPDenoisingTraining=Single noise level2024.03 | 27.39 | 0.723 | |
| BM3DLearning Paradigm=Conventional2021.12 | 27.02 | — | |
| N2VLearning Paradigm=Self-Supervised2021.12 | 25.75 | — | |
| HATTraining=Multiple noise levels required2024.03 | 21.04 | 0.39 | |
| MaskDenoising2024.03 | 20.72 | 0.368 | |
| Restormer2024.03 | 20.12 | 0.321 | |
| DnCNN2024.03 | 19.78 | 0.301 |