Research map / Image generation / Super-resolution
Image Super-Resolution Using Deep Convolutional Networks
Chao Dong, Chen Change Loy, Kaiming He, Xiaoou Tang
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 on ImageNet instead of 91 images (ablation) | SRCNN 9-1-5 · Set5 x3 | PSNR (dB): 32.39 -> 32.52 | Improved |
| double filter numbers (n1=128, n2=64) vs 64/32 (ablation) | SRCNN · Set5 x3 | PSNR (dB): 32.52 -> 32.60 | Improved |
| halve filter numbers (n1=32, n2=16) (ablation) | SRCNN · Set5 x3 | PSNR (dB): 32.52 -> 32.26 | Hurt |
| enlarge first/last filters to 11-1-7 (ablation) | SRCNN 9-1-5 · Set5 x3 | PSNR (dB): 32.52 -> 32.57 | Improved |
| enlarge second-layer filter to 3 (9-3-5) (ablation) | SRCNN 9-1-5 · Set5 x3 | PSNR (dB): 32.52 -> 32.66 | Improved |
| enlarge second-layer filter to 5 (9-5-5) (ablation) | SRCNN 9-3-5 · Set5 x3 | PSNR (dB): 32.66 -> 32.75 | Improved |
| add an extra nonlinear mapping layer (4-layer, e.g. 9-1-1-5) (ablation) | SRCNN · Set5 x3 | PSNR (dB) | No clear effect |
| add extra layer with 32 filters, or two extra layers (ablation) | SRCNN 9-1-5 · Set5 x3 | PSNR (dB) | Hurt |
| train YCbCr channels jointly instead of Y only (ablation) | SRCNN · Set5 x3 color | RGB PSNR (dB): 36.37 -> 33.47 | Hurt |
| train on RGB channels instead of Y only (ablation) | SRCNN · Set5 x3 color | RGB PSNR (dB): 36.37 -> 36.44 | Improved |
| Y pre-train then fine-tune all channels instead of Y only (ablation) | SRCNN · Set5 x3 color | RGB PSNR (dB): 36.37 -> 36.32 | Hurt |
| use SRCNN instead of A+ (scale 3) | A+ · Set5 x3 | PSNR (dB): 32.59 -> 32.75 | Improved |
Closest papers by meaning
- Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network (CVPR 2016 · similarity 0.94)
- Accurate Image Super-Resolution Using Very Deep Convolutional Networks (CVPR 2016 · similarity 0.94)
- Accelerating the Super-Resolution Convolutional Neural Network (ECCV 2016 · similarity 0.93)
- Deep Learning for Image Super-Resolution: A Survey (TPAMI 2020 · similarity 0.93)
- Fast and Accurate Single Image Super-Resolution via Information Distillation Network (CVPR 2018 · similarity 0.93)
- Learning with Privileged Information for Efficient Image Super-Resolution (ECCV 2020 · similarity 0.92)
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.
