Research map / Image generation / Super-resolution
Wavelet-SRNet: A Wavelet-Based CNN for Multi-Scale Face Super Resolution
Huaibo Huang, Ran He, Zhenan Sun, Tieniu Tan
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 |
|---|---|---|---|
| predict wavelet coefficients with wavelet losses instead of bicubic interpolation | Wavelet-SRNet vs bicubic · CelebA test (16x16 input, 8x) | PSNR (dB): 24.83 -> 26.61 | Improved |
| predict wavelet coefficients instead of bicubic interpolation | Wavelet-SRNet vs bicubic · CelebA test (8x8 input, 16x) | PSNR (dB): 21.36 -> 22.65 | Improved |
| predict wavelet coefficients instead of bicubic interpolation | Wavelet-SRNet vs bicubic · Helen test (8x8 input, 16x) | PSNR (dB): 20.96 -> 21.80 | Improved |
| predict wavelet coefficients instead of bicubic interpolation | Wavelet-SRNet vs bicubic · Helen test (32x32 input, 4x) | PSNR (dB): 27.44 -> 27.94 | Improved |
| use wavelet-domain prediction instead of CBN face-prior (dense correspondence) approach | Wavelet-SRNet vs CBN · CelebA test (16x16 input, 8x) | PSNR (dB): 24.68 -> 26.61 | Improved |
| use wavelet-domain prediction instead of CBN | Wavelet-SRNet vs CBN · Helen test (8x8 input, 8x) | PSNR (dB): 19.58 -> 21.83 | Improved |
| use wavelet-domain prediction instead of URDGN adversarial training | Wavelet-SRNet vs URDGN · CelebA test (16x16 input, 8x) | SSIM: 0.8527 -> 0.8949 | Improved |
| use CNN super resolution (SRCNN) instead of bicubic interpolation | SRCNN vs bicubic · CelebA test (32x32 input, 4x) | PSNR (dB): 29.20 -> 20.61 | Hurt |
Later papers that compared with this method
6 later accepted papers report WaveletSRNet as a baseline in their results tables; 4 of them report a better number on most of the results they share with it.
- Progressive Semantic-Aware Style Transformation for Blind Face Restoration (CVPR 2021 · better on 1 of 2 shared results)
- Face Super-Resolution Guided by 3D Facial Priors (ECCV 2020 · better on 2 of 2 shared results)
- Facial Attribute Capsules for Noise Face Super Resolution (AAAI 2020 · better on 1 of 1 shared result)
- Residual Attribute Attention Network for Face Image Super-Resolution (AAAI 2019 · better on 4 of 4 shared results)
- To learn image super-resolution, use a GAN to learn how to do image degradation first (ECCV 2018 · better on 1 of 2 shared results)
- Learning Warped Guidance for Blind Face Restoration (ECCV 2018 · 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
- Face Super-resolution Guided by Facial Component Heatmaps (ECCV 2018 · similarity 0.90)
- Joint Sub-bands Learning with Clique Structures for Wavelet Domain Super-Resolution (NeurIPS 2018 · similarity 0.89)
- FSRNet: End-to-End Learning Face Super-Resolution With Facial Priors (CVPR 2018 · similarity 0.89)
- Accurate Image Super-Resolution Using Very Deep Convolutional Networks (CVPR 2016 · similarity 0.87)
- Learning with Privileged Information for Efficient Image Super-Resolution (ECCV 2020 · similarity 0.87)
- Residual Attribute Attention Network for Face Image Super-Resolution (AAAI 2019 · 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.
