Grayscale Image Denoising on Urban100 (test)
33.85PSNRRestormer-Local
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Restormer-LocalModel Training=separate model for each noise level2021.12 | 33.85 | — | — | — | — | |
| DAGLModel Training=separate model for each noise level2021.12 | 33.79 | — | — | — | — | |
| RestormerModel Training=separate model for each noise level2021.12 | 33.79 | — | — | — | — | |
| Restormer-LocalModel Training=single model for various noise levels2021.12 | 33.73 | — | — | — | — | |
| SwinIRModel Training=separate model for each noise level2021.12 | 33.7 | — | — | — | — | |
| RestormerModel Training=single model for various noise levels2021.12 | 33.67 | — | — | — | — | |
| GCDNParameters=6M, Noise Level (σ)=15, Inference Time (s)=15802023.06 | 33.47 | 93.58 | — | — | — | |
| NLRNModel Training=separate model for each noise level2021.12 | 33.45 | — | — | — | — | |
| DRUNetModel Training=single model for various noise levels2021.12 | 33.44 | — | — | — | — | |
| NLRNParameters=340k, Noise Level (σ)=15, Inference Time (s)=135.82023.06 | 33.42 | 93.48 | — | — | — | |
| DeamNetModel Training=separate model for each noise level2021.12 | 33.37 | — | — | — | — | |
| MWCNNModel Training=separate model for each noise level2021.12 | 33.17 | — | — | — | — | |
| GroupCDL-SParameters=550k, Noise Level (σ)=15, Inference Time (s)=3.562023.06 | 33.07 | 93.4 | — | — | — | |
| GroupSCParameters=68k, Noise Level (σ)=15, Inference Time (s)=93.332023.06 | 32.72 | 93.08 | — | — | — | |
| DnCNNParameters=556k, Noise Level (σ)=15, Inference Time (s)=0.0962023.06 | 32.68 | 92.55 | — | — | — | |
| CDLNetParameters=507k, Noise Level (σ)=15, Inference Time (s)=0.092023.06 | 32.59 | 92.85 | — | — | — | |
| IRCNNModel Training=single model for various noise levels2021.12 | 32.46 | — | — | — | — | |
| FFDNetModel Training=single model for various noise levels2021.12 | 32.4 | — | — | — | — | |
| BM3DNoise Level (σ)=15, Inference Time (s)=0.032023.06 | 32.35 | 92.2 | — | — | — | |
| DnCNNModel Training=single model for various noise levels2021.12 | 32.28 | — | — | — | — | |
| GCDNParameters=6M, Noise Level (σ)=25, Inference Time (s)=15802023.06 | 30.95 | 90.2 | — | — | — | |
| NLRNParameters=340k, Noise Level (σ)=25, Inference Time (s)=135.82023.06 | 30.88 | 90.03 | — | — | — | |
| GroupCDL-SParameters=550k, Noise Level (σ)=25, Inference Time (s)=3.562023.06 | 30.61 | 90.03 | — | — | — | |
| GroupSCParameters=68k, Noise Level (σ)=25, Inference Time (s)=93.332023.06 | 30.05 | 89.12 | — | — | — | |
| CDLNetParameters=507k, Noise Level (σ)=25, Inference Time (s)=0.092023.06 | 30.03 | 89 | — | — | — | |
| DnCNNParameters=556k, Noise Level (σ)=25, Inference Time (s)=0.0962023.06 | 29.92 | 87.97 | — | — | — | |
| BM3DNoise Level (σ)=25, Inference Time (s)=0.032023.06 | 29.7 | 87.77 | — | — | — | |
| GCDNParameters=6M, Noise Level (σ)=50, Inference Time (s)=15802023.06 | 27.41 | 81.6 | — | — | — | |
| NLRNParameters=340k, Noise Level (σ)=50, Inference Time (s)=135.82023.06 | 27.4 | 82.44 | — | — | — | |
| GroupCDL-SParameters=550k, Noise Level (σ)=50, Inference Time (s)=3.562023.06 | 27.29 | 83.05 | — | — | — | |
| CDLNetParameters=507k, Noise Level (σ)=50, Inference Time (s)=0.092023.06 | 26.66 | 81.11 | — | — | — | |
| GroupSCParameters=68k, Noise Level (σ)=50, Inference Time (s)=93.332023.06 | 26.43 | 80.02 | — | — | — | |
| DnCNNParameters=556k, Noise Level (σ)=50, Inference Time (s)=0.0962023.06 | 26.28 | 78.74 | — | — | — | |
| BM3DNoise Level (σ)=50, Inference Time (s)=0.032023.06 | 25.95 | 77.91 | — | — | — | |
| DAGLModel Strategy=Separate model for each noise level2021.11 | — | — | 33.79 | 31.39 | 27.97 | |
| DeamNetModel Strategy=Separate model for each noise level2021.11 | — | — | 33.37 | 30.85 | 27.53 | |
| DnCNNModel Strategy=Single model for various noise levels2021.11 | — | — | 32.28 | 29.8 | 26.35 | |
| DnCNNParams [M]=0.562023.03 | — | — | 32.64 | 29.95 | 26.26 | |
| DnCNNParams [M]=0.562024.11 | — | — | 32.64 | 29.95 | 26.26 | |
| DRUNetModel Strategy=Single model for various noise levels2021.11 | — | — | 33.44 | 31.11 | 27.96 | |
| DRUNetParams [M]=32.642023.03 | — | — | 33.44 | 31.11 | 27.96 | |
| DRUNetParams [M]=32.642024.11 | — | — | 33.44 | 31.11 | 27.96 | |
| FFDNetModel Strategy=Single model for various noise levels2021.11 | — | — | 32.4 | 29.9 | 26.5 | |
| FOCNetModel Strategy=Separate model for each noise level2021.11 | — | — | 33.15 | 30.64 | 27.4 | |
| GRL-BParams [M]=19.812023.03 | — | — | 34.09 | 31.8 | 28.59 | |
| GRL-SParams [M]=3.122023.03 | — | — | 33.84 | 31.49 | 28.24 | |
| GRL-TParams [M]=0.882023.03 | — | — | 33.66 | 31.23 | 27.89 | |
| Hi-IRParams [M]=22.332024.11 | — | — | 34.11 | 31.92 | 28.91 | |
| IPTParams [M]=115.332024.11 | — | — | — | — | 29.71 | |
| IRCNNModel Strategy=Single model for various noise levels2021.11 | — | — | 32.46 | 29.8 | 26.22 | |
| MWCNNModel Strategy=Separate model for each noise level2021.11 | — | — | 33.17 | 30.66 | 27.42 | |
| NLRNModel Strategy=Separate model for each noise level2021.11 | — | — | 33.45 | 30.94 | 27.49 | |
| RestormerModel Strategy=Single model for various noise levels2021.11 | — | — | 33.67 | 31.39 | 28.33 | |
| RestormerModel Strategy=Separate model for each noise level2021.11 | — | — | 33.79 | 31.46 | 28.29 | |
| RestormerParams [M]=26.132023.03 | — | — | 33.79 | 31.46 | 28.29 | |
| RestormerParams [M]=26.132024.11 | — | — | 33.79 | 31.46 | 28.29 | |
| RNANModel Strategy=Separate model for each noise level2021.11 | — | — | — | — | 27.65 | |
| RNANParams [M]=8.962023.03 | — | — | — | — | 27.65 | |
| RNANParams [M]=8.962024.11 | — | — | — | — | 27.65 | |
| SwinIRModel Strategy=Separate model for each noise level2021.11 | — | — | 33.7 | 31.3 | 27.98 | |
| SwinIRParams [M]=11.752023.03 | — | — | 33.7 | 31.3 | 27.98 | |
| SwinIRParams [M]=11.752024.11 | — | — | 33.7 | 31.3 | 27.98 | |
| XformerParams [M]=21.232024.11 | — | — | 33.98 | 31.78 | 28.71 |