Research map / Image generation / Super-resolution
Not All Degradations Are Equal: A Targeted Feature Denoising Framework for Generalizable Image Super-Resolution
Hongjun Wang, Jiyuan Chen, Zhengwei Yin, Xuan Song, Yinqiang Zheng
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 noise detection module (ablation) | TFD on SRResNet · Set5 clean | PSNR: 26.32 -> 26.71 | Improved |
| add noise detection module (ablation) | TFD on SwinIR · Set5 clean | PSNR: 26.58 -> 26.92 | Improved |
| add spatial denoising path (ablation) | TFD on SRResNet · Set5 clean | PSNR: 26.28 -> 26.71 | Improved |
| add frequency-domain denoising path (ablation) | TFD on SRResNet · Set5 clean | PSNR: 25.41 -> 26.71 | Improved |
| add frequency-domain denoising path (ablation) | TFD on SwinIR · Set5 clean | PSNR: 26.42 -> 26.92 | Improved |
| use Full TFD instead of Heavy configuration (ablation) | TFD on SwinIR · Urban100 | PSNR: 22.60 -> 22.63 | No clear effect |
| use Full TFD instead of Light configuration (ablation) | TFD on SwinIR · Urban100 | PSNR: 22.38 -> 22.63 | Improved |
| use lambda_cls 0.10, lambda_denoise 0.01 instead of 0.20/0.020 (ablation) | TFD on SRResNet · Set5 | PSNR: 26.63 -> 26.71 | Improved |
| add TFD to SRResNet | SRResNet · Set5 blind SR average | PSNR: 23.68 -> 24.46 | Improved |
| add TFD to HAT | HAT · Urban100 blind SR average | PSNR: 21.12 -> 21.50 | Improved |
| use TFD instead of Dropout | SRResNet · Canon real-world | PSNR: 24.86 -> 25.72 | Improved |
| use TFD instead of feature alignment | SRResNet · Canon real-world | PSNR: 25.13 -> 25.72 | Improved |
Closest papers by meaning
- Crafting Training Degradation Distribution for the Accuracy-Generalization Trade-off in Real-World Super-Resolution (ICML 2023 · similarity 0.90)
- NegVSR: Augmenting Negatives for Generalized Noise Modeling in Real-world Video Super-Resolution (AAAI 2024 · similarity 0.89)
- Stochastic Frequency Masking to Improve Super-Resolution and Denoising Networks (ECCV 2020 · similarity 0.88)
- Unsupervised Diffusion-Based Degradation Modeling for Real-World Super-Resolution (AAAI 2025 · similarity 0.88)
- Navigating Beyond Dropout: An Intriguing Solution towards Generalizable Image Super Resolution (CVPR 2024 · similarity 0.88)
- Unsupervised Degradation Representation Aware Transform for Real-World Blind Image Super-Resolution (AAAI 2025 · 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.
