Research map / Image generation / Image restoration
Exploiting the Potential of Standard Convolutional Autoencoders for Image Restoration by Evolutionary Search
Masanori Suganuma, Mete Ozay, Takayuki Okatani
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 |
|---|---|---|---|
| Use evolutionary-search-designed standard CAEs instead of the strongest reported baseline | Image inpainting · CelebA, Cars, and SVHN; center, pixel, and half masks | PSNR: CelebA: 28.5 -> 29.9 (center), 22.9 -> 27.8 (pixel), 19.9 -> 21.1 (half); Cars: 19.6 -> 20.9 (center), 18.9 -> 19.5 (pix | Improved |
| Use evolutionary-search-designed standard CAEs instead of the strongest baseline | Image denoising · BSD200, σ=30, 50, and 70 | PSNR: 27.95 -> 28.23; 25.86 -> 26.17; 24.53 -> 24.83 | Improved |
| Use evolutionary-search-designed standard CAEs instead of the strongest baseline | Image denoising · BSD200, σ=30 | SSIM: 0.8053 -> 0.8047 | Hurt |
| Use evolutionary-search-designed standard CAEs instead of the strongest baseline | Image denoising · BSD200, σ=50 and 70 | SSIM: 0.7202 -> 0.7255; 0.6608 -> 0.6636 | Improved |
| Reduce mutation probability; the tested setting with the best reported outcome is r=0.1 (ablation) | Evolutionary search for E-CAE · CelebA center-mask inpainting | PSNR | Improved |
| Increase the number of children per generation; λ=4 gives the best reported performance (ablation) | Evolutionary search for E-CAE · CelebA center-mask inpainting | PSNR: 29.80 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MemNet — MemNet: A Persistent Memory Network for Image Restoration (ICCV 2017) · better on 5 of 6 shared results
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
- Revitalizing Convolutional Network for Image Restoration (TPAMI 2024 · similarity 0.91)
- On-Demand Learning for Deep Image Restoration (ICCV 2017 · similarity 0.88)
- Re-Boosting Self-Collaboration Parallel Prompt GAN for Unsupervised Image Restoration (TPAMI 2025 · similarity 0.88)
- Image Restoration Using Very Deep Convolutional Encoder-Decoder Networks with Symmetric Skip Connections (NeurIPS 2016 · similarity 0.87)
- Deep Image Prior (CVPR 2018 · similarity 0.87)
- A Restoration Network as an Implicit Prior (ICLR 2024 · similarity 0.86)
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.
