Research map / Image generation / Super-resolution
Stochastic Frequency Masking to Improve Super-Resolution and Denoising Networks
Majed El Helou, Ruofan Zhou, Sabine Süsstrunk
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 50% SFM during training | RCAN · DIV2K val x4, Gaussian g6.5 test kernel | PSNR (dB): 21.72 -> 21.79 | Improved |
| add 50% SFM during training | RRDB · DIV2K val x4, bicubic | PSNR (dB): 28.79 -> 29.10 | Improved |
| add 50% SFM during training | ESRGAN · DIV2K val x4, Gaussian g6.5 | PSNR (dB): 19.42 -> 19.77 | Improved |
| add 50% SFM during training | IKC · DIV2K val x4, unseen kernels | PSNR (dB): 22.08 -> 22.35 | Improved |
| add 50% SFM during training | IKC · DIV2K val x4, bicubic | PSNR (dB): 27.81 -> 27.78 | No clear effect |
| add 50% SFM during training | KMSR · RealSR x4 | PSNR (dB): 28.27 -> 28.50 | Improved |
| add 50% SFM during training | RCAN · SR-RAW x8 | PSNR (dB): 24.18 -> 24.50 | Improved |
| add 50% SFM during training | MemNet · BSD68 AWGN noise level 100 | PSNR (dB): 17.16 -> 19.46 | Improved |
| add 50% SFM during training | RIDNet · BSD68 AWGN noise level 10 | PSNR (dB): 33.65 -> 33.43 | Hurt |
| add 50% SFM during training | N3Net · BSD68 AWGN noise level 10 | PSNR (dB): 33.53 -> 33.41 | Hurt |
| add 50% SFM during training | DnCNN-B · BSD68 AWGN noise level 70 | PSNR (dB): 22.06 -> 22.80 | Improved |
| add 50% SFM during training | Noise2Noise · Two-photon microscopy, 1 raw image | PSNR (dB): 32.70 -> 33.05 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Noise2Noise — Noise2Noise: Learning Image Restoration without Clean Data (ICML 2018) · better on 1 of 1 shared result
- MemNet — MemNet: A Persistent Memory Network for Image Restoration (ICCV 2017) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Blind Image Super-Resolution With Elaborate Degradation Modeling on Noise and Kernel (CVPR 2022 · similarity 0.90)
- Not All Degradations Are Equal: A Targeted Feature Denoising Framework for Generalizable Image Super-Resolution (ICCV 2025 · similarity 0.88)
- Spectrum-to-Kernel Translation for Accurate Blind Image Super-Resolution (NeurIPS 2021 · similarity 0.88)
- Finding Discriminative Filters for Specific Degradations in Blind Super-Resolution (NeurIPS 2021 · similarity 0.88)
- Stochastic Deep Restoration Priors for Imaging Inverse Problems (ICML 2025 · similarity 0.87)
- One-Step Effective Diffusion Network for Real-World Image Super-Resolution (NeurIPS 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.
