Research map / Image generation / Super-resolution
Bayesian Image Super-Resolution With Deep Modeling of Image Statistics
Shangqi Gao, Xiahai Zhuang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add batch normalization (ablation) | BayeSR (ResNet, depth 2) · DIV2K realistic SISR x4 (supervised) | PSNR: 23.84 -> 23.83 | No clear effect |
| use RCAB instead of ResBlock (ablation) | BayeSR (ResNet, depth 2) · DIV2K realistic SISR x4 (supervised) | PSNR: 23.84 -> 23.83 | No clear effect |
| increase CNN depth from (2,2,2) to (8,8,8) (ablation) | BayeSR (ResNet RCAB) · DIV2K realistic SISR x4 (supervised) | PSNR: 23.83 -> 24.10 | Improved |
| use ResNet instead of U-Net (ablation) | BayeSR · DIV2K realistic SISR x4 (supervised) | PSNR: 23.92 -> 24.10 | Improved |
| use ResNet instead of U-Net (ablation) | BayeSR · DIV2K realistic SISR x4 (unsupervised) | PSNR: 23.01 -> 23.43 | Improved |
| add GAN adversarial learning (ablation) | BayeSR (ResNet) · DIV2K realistic SISR x4 (unsupervised) | PSNR: 23.43 -> 23.67 | Improved |
| use WGAN instead of GAN (ablation) | BayeSR (ResNet) · DIV2K realistic SISR x4 (unsupervised) | NIQE: 7.40 -> 5.06 | Mixed |
| add discriminative learning to generative learning (ablation) | BayeSR (ResNet) · DIV2K realistic SISR x4 (unsupervised) | PSNR: 21.74 -> 23.43 | Improved |
| add generative learning to discriminative learning (ablation) | BayeSR (ResNet) · DIV2K realistic SISR x4 (unsupervised) | PSNR: 22.20 -> 23.43 | Improved |
| add explicit prior modeling (variational loss) (ablation) | BayeSR auto-encoder · BSD68 with AWGN noise 0 to 20 | PSNR and SSIM | Mixed |
| use S-BayeSR instead of S-Baseline under noise | S-BayeSR · Set5 x4, noise sigma 10 | PSNR: 27.27 -> 28.19 | Improved |
| use S-BayeSR instead of S-Baseline without noise | S-BayeSR · Set5 x4, noise sigma 0 | PSNR: 32.07 -> 31.50 | Hurt |
Closest papers by meaning
- BUFF: Bayesian Uncertainty Guided Diffusion Probabilistic Model for Single Image Super-Resolution (AAAI 2025 · similarity 0.88)
- Stochastic Deep Restoration Priors for Imaging Inverse Problems (ICML 2025 · similarity 0.88)
- Noise-Free Optimization in Early Training Steps for Image Super-resolution (AAAI 2024 · similarity 0.88)
- A Restoration Network as an Implicit Prior (ICLR 2024 · similarity 0.88)
- ResDiff: Combining CNN and Diffusion Model for Image Super-resolution (AAAI 2024 · similarity 0.88)
- Universal Robustness via Median Randomized Smoothing for Real-World Super-Resolution (CVPR 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
