Research map / Image generation / Super-resolution
Designing a Practical Degradation Model for Deep Blind Image Super-Resolution
Kai Zhang, Jingyun Liang, Luc Van Gool, Radu Timofte
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 |
|---|---|---|---|
| train with the proposed shuffled degradation model instead of bicubic degradation | ESRGAN architecture · DIV2K4D Type IV | LPIPS: 0.599 -> 0.361 | Improved |
| train with the proposed shuffled degradation model instead of bicubic degradation | ESRGAN architecture · DIV2K4D Type IV | PSNR (Y, dB): 23.68 -> 24.58 | Improved |
| train with the proposed shuffled degradation model instead of bicubic degradation | ESRGAN architecture · DIV2K4D Type II | LPIPS: 0.526 -> 0.284 | Improved |
| train with the proposed shuffled degradation model instead of bicubic degradation | ESRGAN architecture · DIV2K4D Type I (bicubic) | PSNR (Y, dB): 28.16 -> 27.30 | Hurt |
| train with the proposed shuffled degradation model instead of bicubic degradation | ESRGAN architecture · DIV2K4D Type I (bicubic) | LPIPS: 0.115 -> 0.236 | Hurt |
| train with the proposed degradation model (PSNR-oriented BSRNet vs RRDB) | RRDB architecture · DIV2K4D Type IV | PSNR (Y, dB): 24.03 -> 25.67 | Improved |
| train with the proposed degradation model (PSNR-oriented BSRNet vs RRDB) | RRDB architecture · DIV2K4D Type I (bicubic) | PSNR (Y, dB): 30.89 -> 29.07 | Hurt |
| add GAN and perceptual losses (BSRGAN vs BSRNet) | BSRNet · DIV2K4D Type IV | PSNR (Y, dB): 25.67 -> 24.58 | Hurt |
| add GAN and perceptual losses (BSRGAN vs BSRNet) | BSRNet · DIV2K4D Type IV | LPIPS: 0.506 -> 0.361 | Improved |
| train with the proposed degradation model, evaluated with no-reference IQA | ESRGAN architecture · RealSRSet | NIQE: 4.95 -> 5.60 | Hurt |
| train with the proposed degradation model, evaluated with no-reference IQA | ESRGAN architecture · RealSRSet | PI: 4.47 -> 4.72 | Hurt |
| train with the proposed degradation model, evaluated with no-reference IQA | ESRGAN architecture · RealSRSet | NRQM: 6.02 -> 6.17 | Improved |
Closest papers by meaning
- Blind Image Super-Resolution With Elaborate Degradation Modeling on Noise and Kernel (CVPR 2022 · similarity 0.92)
- Deep Plug-And-Play Super-Resolution for Arbitrary Blur Kernels (CVPR 2019 · similarity 0.91)
- Learning a Single Convolutional Super-Resolution Network for Multiple Degradations (CVPR 2018 · similarity 0.91)
- Blind Image Super-Resolution: A Survey and Beyond (TPAMI 2022 · similarity 0.91)
- Unsupervised Degradation Representation Learning for Blind Super-Resolution (CVPR 2021 · similarity 0.91)
- Unsupervised Degradation Representation Aware Transform for Real-World Blind Image Super-Resolution (AAAI 2025 · similarity 0.89)
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.
