Gaussian Deblurring on ImageNet 256 x 256
28.06PSNRDDNM
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DDNM2026.06 | 28.06 | 70.3 | 0.278 | 81.43 | |
| DDRM2026.06 | 27.81 | 63.9 | 0.333 | 124.55 | |
| SOCS-Nonlinear-γ2026.06 | 26.67 | 74 | 0.282 | 110.69 | |
| DAPS2026.06 | 26.37 | 69.7 | 0.297 | 109.61 | |
| SOCS-Nonlinear2026.06 | 26.32 | 70.8 | 0.313 | 106 | |
| DAPS2026.06 | 26.15 | 68.4 | 0.253 | 75.68 | |
| Robust-GDrho (contamination fraction)=0.02, sigma (Gaussian noise level)=0.05, Problem type=Linear IPs2026.05 | 25.96 | 67.9 | 0.424 | 159.84 | |
| EPSNFE=1†, Architecture=EDM-DDPM++, Pre-training=Trained from scratch2026.06 | 25.9 | 70.4 | 0.249 | 100.1 | |
| SOCS-Linear2026.06 | 25.38 | 72 | 0.275 | 111.49 | |
| DCDP2026.06 | 25.12 | 69 | 0.429 | 142.46 | |
| EPSNFE=20, Architecture=EDM-DDPM++, Pre-training=Trained from scratch2026.06 | 24.92 | 66.6 | 0.222 | 71.31 | |
| Robust-GDrho (contamination fraction)=0.10, sigma (Gaussian noise level)=0.05, Problem type=Linear IPs2026.05 | 24.9 | 66 | 0.453 | 171.81 | |
| Robust-CGrho (contamination fraction)=0.02, sigma (Gaussian noise level)=0.05, Problem type=Linear IPs2026.05 | 24.79 | 67.3 | 0.458 | 169.11 | |
| Robust-CGrho (contamination fraction)=0.10, sigma (Gaussian noise level)=0.05, Problem type=Linear IPs2026.05 | 24.73 | 67.2 | 0.45 | 167.56 | |
| EPSNFE=100, Architecture=EDM-DDPM++, Pre-training=Trained from scratch2026.06 | 24.34 | 64 | 0.224 | 72.38 | |
| RED-diff2026.06 | 24.24 | 58.9 | 0.292 | 108 | |
| FPS-SMC2026.06 | 23.91 | 60.1 | 0.387 | 91.72 | |
| DDNM2026.06 | 23.82 | 65.7 | 0.323 | 211.84 | |
| RED-diffrho (contamination fraction)=0.02, sigma (Gaussian noise level)=0.05, Problem type=Linear IPs2026.05 | 23.8 | 60.3 | 0.558 | 169.24 | |
| DiffPIR2026.06 | 23.25 | 64.6 | 0.34 | 195.75 | |
| FPS - SMC2026.06 | 23.09 | 58.8 | 0.417 | 220.59 | |
| DPS2026.06 | 22.84 | 58.7 | 0.312 | 115.02 | |
| DiffPIR2026.06 | 22.8 | — | 0.355 | 93.36 | |
| DPSrho (contamination fraction)=0.02, sigma (Gaussian noise level)=0.05, Problem type=Linear IPs2026.05 | 21.82 | 53.2 | 0.358 | 139.39 | |
| DDRM2026.06 | 21.26 | 56.4 | 0.443 | 146.89 | |
| RED-diffrho (contamination fraction)=0.10, sigma (Gaussian noise level)=0.05, Problem type=Linear IPs2026.05 | 20.76 | 49.7 | 0.645 | 219.67 | |
| DAPSrho (contamination fraction)=0.02, sigma (Gaussian noise level)=0.05, Problem type=Linear IPs2026.05 | 20.35 | 45 | 0.619 | 251.12 | |
| DPS2026.06 | 20.31 | 59.8 | 0.397 | 116.42 | |
| DiffPIRrho (contamination fraction)=0.02, sigma (Gaussian noise level)=0.05, Problem type=Linear IPs2026.05 | 19.85 | 36.7 | 0.676 | 250.67 | |
| DPSrho (contamination fraction)=0.10, sigma (Gaussian noise level)=0.05, Problem type=Linear IPs2026.05 | 19.7 | 49 | 0.405 | 160.82 | |
| DCPSrho (contamination fraction)=0.02, sigma (Gaussian noise level)=0.05, Problem type=Linear IPs2026.05 | 18.49 | 35.6 | 0.593 | 224.9 | |
| DiffPIRrho (contamination fraction)=0.10, sigma (Gaussian noise level)=0.05, Problem type=Linear IPs2026.05 | 16.6 | 27.4 | 0.74 | 292.01 | |
| DAPSrho (contamination fraction)=0.10, sigma (Gaussian noise level)=0.05, Problem type=Linear IPs2026.05 | 14.17 | 25.3 | 0.72 | 308.08 | |
| DCPSrho (contamination fraction)=0.10, sigma (Gaussian noise level)=0.05, Problem type=Linear IPs2026.05 | 12.89 | 19.3 | 0.702 | 305.81 |