Gaussian Denoising on BSD68 (test)
34.48PSNRMambaIR
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| MambaIRnoise level (σ)=152024.02 | 34.48 | — | — | — | — | |
| SwinIRnoise level (σ)=152024.02 | 34.42 | — | — | — | — | |
| Restormernoise level (σ)=152024.02 | 34.4 | — | — | — | — | |
| DRUNetnoise level (σ)=152024.02 | 34.3 | — | — | — | — | |
| DnCNNnoise level (σ)=152024.02 | 33.9 | — | — | — | — | |
| FFDNetnoise level (σ)=152024.02 | 33.87 | — | — | — | — | |
| IRCNNnoise level (σ)=152024.02 | 33.86 | — | — | — | — | |
| MambaIRnoise level (σ)=252024.02 | 32.24 | — | — | — | — | |
| feature space deep residual learningNoise (σ)=152016.11 | 31.8607 | 89.41 | — | — | — | |
| Restormernoise level (σ)=252024.02 | 31.79 | — | — | — | — | |
| SwinIRnoise level (σ)=252024.02 | 31.78 | — | — | — | — | |
| DnCNN-SNoise (σ)=152016.11 | 31.7202 | 89.01 | — | — | — | |
| DRUNetnoise level (σ)=252024.02 | 31.69 | — | — | — | — | |
| MEASNet2026.03 | 31.257 | 87.7 | — | — | — | |
| DnCNNnoise level (σ)=252024.02 | 31.24 | — | — | — | — | |
| MoCE-IR2026.03 | 31.22 | 87.3 | — | — | — | |
| FFDNetnoise level (σ)=252024.02 | 31.21 | — | — | — | — | |
| IRCNNnoise level (σ)=252024.02 | 31.16 | — | — | — | — | |
| PromptIR2026.03 | 31.117 | 87.3 | — | — | — | |
| BM3DNoise (σ)=152016.11 | 31.0761 | 87.22 | — | — | — | |
| AirNet2026.03 | 31.06 | 87.3 | — | — | — | |
| DaAIR2026.03 | 31.06 | 86.9 | — | — | — | |
| DaR-NET2026.03 | 30.743 | 87.5 | — | — | — | |
| Restormer2026.03 | 29.363 | — | — | — | — | |
| Supervised methodsigma=252026.05 | 29.2 | — | — | — | — | |
| Refined BM3dsigma=252026.05 | 29.09 | — | — | — | — | |
| Neighbour2Neighboursigma=252026.05 | 28.87 | — | — | — | — | |
| LGSRsigma=252026.05 | 28.79 | — | — | — | — | |
| MambaIRnoise level (σ)=502024.02 | 28.66 | — | — | — | — | |
| BM3Dsigma=252026.05 | 28.62 | — | — | — | — | |
| Restormernoise level (σ)=502024.02 | 28.6 | — | — | — | — | |
| SwinIRnoise level (σ)=502024.02 | 28.56 | — | — | — | — | |
| feature space deep residual learningNoise (σ)=302016.11 | 28.5599 | 80.92 | — | — | — | |
| DRUNetnoise level (σ)=502024.02 | 28.51 | — | — | — | — | |
| DnCNN-SNoise (σ)=302016.11 | 28.3324 | 80.03 | — | — | — | |
| FFDNetnoise level (σ)=502024.02 | 27.96 | — | — | — | — | |
| DnCNNnoise level (σ)=502024.02 | 27.95 | — | — | — | — | |
| Noise2Fastsigma=252026.05 | 27.87 | — | — | — | — | |
| IRCNNnoise level (σ)=502024.02 | 27.86 | — | — | — | — | |
| BM3DNoise (σ)=302016.11 | 27.7492 | 77.35 | — | — | — | |
| MMESsigma=252026.05 | 27.21 | — | — | — | — | |
| Noise2Selfsigma=252026.05 | 27.06 | — | — | — | — | |
| feature space deep residual learningNoise (σ)=502016.11 | 26.3577 | 72.7 | — | — | — | |
| Supervised methodsigma=502026.05 | 26.24 | — | — | — | — | |
| DnCNN-SNoise (σ)=502016.11 | 26.2275 | 71.63 | — | — | — | |
| Refined BM3dsigma=502026.05 | 26.11 | — | — | — | — | |
| Neighbour2Neighboursigma=502026.05 | 25.85 | — | — | — | — | |
| LGSRsigma=502026.05 | 25.84 | — | — | — | — | |
| BM3Dsigma=502026.05 | 25.69 | — | — | — | — | |
| BM3DNoise (σ)=502016.11 | 25.6103 | 68.68 | — | — | — | |
| Noise2Selfsigma=502026.05 | 25.14 | — | — | — | — | |
| MMESsigma=502026.05 | 24.62 | — | — | — | — | |
| Noise2Fastsigma=502026.05 | 24.55 | — | — | — | — | |
| LF (Linear filters)sigma=252026.05 | 24.49 | — | — | — | — | |
| LF (Linear filters)sigma=502026.05 | 23.08 | — | — | — | — | |
| DAGLModel Strategy=Separate model for each noise level2021.11 | — | — | 31.93 | 29.46 | 26.51 | |
| DeamNetModel Strategy=Separate model for each noise level2021.11 | — | — | 31.91 | 29.44 | 26.54 | |
| DnCNNModel Strategy=Single model for various noise levels2021.11 | — | — | 31.62 | 29.16 | 26.23 | |
| DRUNetModel Strategy=Single model for various noise levels2021.11 | — | — | 31.91 | 29.48 | 26.59 | |
| FFDNetModel Strategy=Single model for various noise levels2021.11 | — | — | 31.63 | 29.19 | 26.29 | |
| FOCNetModel Strategy=Separate model for each noise level2021.11 | — | — | 31.83 | 29.38 | 26.5 | |
| IRCNNModel Strategy=Single model for various noise levels2021.11 | — | — | 31.63 | 29.15 | 26.19 | |
| MWCNNModel Strategy=Separate model for each noise level2021.11 | — | — | 31.86 | 29.41 | 26.53 | |
| NLRNModel Strategy=Separate model for each noise level2021.11 | — | — | 31.88 | 29.41 | 26.47 | |
| RestormerModel Strategy=Single model for various noise levels2021.11 | — | — | 31.95 | 29.51 | 26.62 | |
| RestormerModel Strategy=Separate model for each noise level2021.11 | — | — | 31.96 | 29.52 | 26.62 | |
| RNANModel Strategy=Separate model for each noise level2021.11 | — | — | — | — | 26.48 | |
| SwinIRModel Strategy=Separate model for each noise level2021.11 | — | — | 31.97 | 29.5 | 26.58 |