Research map / Image generation / Image restoration
MemNet: A Persistent Memory Network for Image Restoration
Ying Tai, Jian Yang, Xiaoming Liu, Chunyan Xu
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 long-term dense connections among memory blocks (ablation) | MemNet · Set5 x2 SISR | PSNR (dB): 37.68 -> 37.78 | Improved |
| add long-term dense connections among memory blocks (ablation) | MemNet · Set5 x3 SISR | PSNR (dB): 33.96 -> 34.09 | Improved |
| add long-term dense connections among memory blocks (ablation) | MemNet · Set5 x4 SISR | PSNR (dB): 31.60 -> 31.74 | Improved |
| add short-term connections from first R-1 recursions to gate unit (ablation) | MemNet · Set5 x2 SISR | PSNR (dB): 37.71 -> 37.78 | Improved |
| add short-term connections from first R-1 recursions to gate unit (ablation) | MemNet · Set5 x3 SISR | PSNR (dB): 34.00 -> 34.09 | Improved |
| add short-term connections from first R-1 recursions to gate unit (ablation) | MemNet · Set5 x4 SISR | PSNR (dB): 31.65 -> 31.74 | Improved |
| increase depth from M4R6 (54 layers) to M6R6 (80 layers) (ablation) | MemNet · Set5 x3 SISR | PSNR (dB): 34.05 -> 34.09 | Improved |
| increase depth from M6R6 (80 layers) to M6R8 (104 layers) (ablation) | MemNet · Set5 x3 SISR | PSNR (dB): 34.09 -> 34.16 | Improved |
| increase depth from M6R6 (80 layers) to M10R10 (212 layers) (ablation) | MemNet · Set5 x3 SISR | PSNR (dB): 34.09 -> 34.23 | Improved |
| use MemNet instead of RED for denoising | MemNet vs RED · BSD200 sigma=70 | PSNR (dB): 24.37 -> 24.53 | Improved |
| use MemNet instead of DnCNN for JPEG deblocking | MemNet vs DnCNN · Classic5 q=10 | PSNR (dB): 29.40 -> 29.69 | Improved |
| use MemNet instead of DRRN for SISR | MemNet vs DRRN · Urban100 x4 | PSNR (dB): 25.44 -> 25.50 | Improved |
Later papers that compared with this method
9 later accepted papers report MemNet as a baseline in their results tables; 9 of them report a better number on most of the results they share with it.
- Residual Dense Network for Image Restoration (TPAMI 2020 · better on 2 of 2 shared results)
- Stochastic Frequency Masking to Improve Super-Resolution and Denoising Networks (ECCV 2020 · better on 1 of 1 shared result)
- Self-Guided Network for Fast Image Denoising (ICCV 2019 · better on 8 of 8 shared results)
- Residual Non-local Attention Networks for Image Restoration (ICLR 2019 · better on 3 of 3 shared results)
- Denoising Prior Driven Deep Neural Network for Image Restoration (TPAMI 2018 · better on 10 of 10 shared results)
- Exploiting the Potential of Standard Convolutional Autoencoders for Image Restoration by Evolutionary Search (ICML 2018 · better on 5 of 6 shared results)
- Non-Local Recurrent Network for Image Restoration (NeurIPS 2018 · better on 6 of 6 shared results)
- Fast, Accurate, and Lightweight Super-Resolution with Cascading Residual Network (ECCV 2018 · better on 2 of 2 shared results)
1 more not listed.
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 0.88)
- Deep Image Prior (CVPR 2018 · similarity 0.87)
- Learning Enriched Features for Real Image Restoration and Enhancement (ECCV 2020 · similarity 0.87)
- A Minimalistic Unified Framework for Incremental Learning across Image Restoration Tasks (NeurIPS 2025 · similarity 0.87)
- Learning Enriched Features for Fast Image Restoration and Enhancement (TPAMI 2022 · similarity 0.87)
- Dual Residual Networks Leveraging the Potential of Paired Operations for Image Restoration (CVPR 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.
