Image Denoising on Kodak Gaussian σ=25 (test)
33.04PSNRRestormer
Evaluation results
| Method | Links | ||
|---|---|---|---|
| RestormerLearning Paradigm=Supervised, Training Noise=Gaussian σ=252024.01 | 33.04 | 0.897 | |
| SwinIRLearning Paradigm=Supervised, Training Noise=Gaussian σ=252024.01 | 32.89 | 0.895 | |
| Honzatko-pmeSupervision=Self-supervised, train_passes=1, test_passes=1, assumed_known_noise_model=true2023.05 | 32.45 | — | |
| Noise2CleanSupervision=Supervised, train_passes=1, test_passes=12023.05 | 32.43 | 0.884 | |
| Noise2NoiseSupervision=Supervised, train_passes=1, test_passes=12023.05 | 32.41 | 0.884 | |
| Laine19-pmeSupervision=Self-supervised, train_passes=4, test_passes=4, assumed_known_noise_model=true2023.05 | 32.4 | 0.883 | |
| B2UBSupervision=Self-supervised, train_passes=17, test_passes=12023.05 | 32.27 | 0.88 | |
| DCD-NetSupervision=Self-supervised, train_passes=3, test_passes=12023.05 | 32.27 | 0.881 | |
| B2ULearning Paradigm=Unsupervised, Training Noise=Gaussian σ=252024.01 | 32.26 | 0.88 | |
| R2RSupervision=Self-supervised, train_passes=1, test_passes=502023.05 | 32.25 | 0.88 | |
| NBR2NBRSupervision=Self-supervised, train_passes=2, test_passes=12023.05 | 32.08 | 0.879 | |
| Nb2NbLearning Paradigm=Unsupervised, Training Noise=Gaussian σ=252024.01 | 32.06 | 0.88 | |
| CBM3DSupervision=Traditional2023.05 | 31.87 | 0.868 | |
| MPILearning Paradigm=Zero-shot2024.01 | 31.78 | 0.865 | |
| DBSNSupervision=Self-supervised, train_passes=2, test_passes=12023.05 | 31.64 | 0.856 | |
| SwinIASupervision=Self-supervised (True BSN), train_passes=1, test_passes=1, self_implemented=true2023.05 | 31.43 | 0.863 | |
| Self2SelfSupervision=Self-supervised, train_passes=1, test_passes=502023.05 | 31.28 | 0.864 | |
| MPI (faster)Learning Paradigm=Zero-shot2024.01 | 30.94 | 0.848 | |
| Noise2SameSupervision=Self-supervised, train_passes=2, test_passes=1, self_implemented=true2023.05 | 30.77 | 0.841 | |
| Noisier2NoiseSupervision=Self-supervised, train_passes=1, test_passes=1, assumed_known_noise_model=true2023.05 | 30.7 | 0.845 | |
| Laine19Supervision=Self-supervised (True BSN), train_passes=4, test_passes=42023.05 | 30.62 | 0.84 | |
| Noise2VoidSupervision=Self-supervised, train_passes=1, test_passes=12023.05 | 30.32 | 0.821 | |
| DIPLearning Paradigm=Unsupervised, Training Noise=Gaussian σ=252024.01 | 30.05 | 0.806 | |
| ZS-N2NLearning Paradigm=Zero-shot2024.01 | 29.46 | 0.775 |