Research map / Image generation / Super-resolution
Structure-Preserving Image Super-Resolution
Cheng Ma, Yongming Rao, Jiwen Lu, Jie Zhou
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 gradient loss without gradient branch (ablation) | ESRGAN · Urban100 | LPIPS: 0.1229 -> 0.1188 | Improved |
| add gradient branch (ablation) | SPSR-G · Urban100 | LPIPS: 0.1188 -> 0.1184 | Improved |
| add gradient loss (ablation) | SPSR-G · Urban100 | LPIPS: 0.1195 -> 0.1184 | Improved |
| add pixelwise gradient loss (ablation) | SPSR-G · Urban100 | LPIPS: 0.1222 -> 0.1184 | Improved |
| add adversarial structure loss (ablation) | SPSR-P · Urban100 | LPIPS: 0.1171 -> 0.1146 | Improved |
| add adversarial structure loss (ablation) | SPSR-P · Urban100 | PSNR: 25.585 -> 25.228 | Hurt |
| add pixelwise structure loss (ablation) | SPSR-P · Urban100 | PSNR: 25.163 -> 25.228 | Improved |
| add pixelwise structure loss (ablation) | SPSR-P · Urban100 | LPIPS: 0.1140 -> 0.1146 | Hurt |
| add gradient branch (ablation) | SPSR-P · Urban100 | LPIPS: 0.1173 -> 0.1146 | Improved |
| add whole structure loss (ablation) | SPSR-P · Urban100 | LPIPS: 0.1221 -> 0.1146 | Improved |
| use learned NSE instead of pretrained VGG19 (ablation) | SPSR · Urban100 | LPIPS: 0.1359 -> 0.1146 | Improved |
| use learned NSE instead of pretrained ResNet18 (ablation) | SPSR · Urban100 | LPIPS: 0.1271 -> 0.1146 | Improved |
Closest papers by meaning
- Structure-Preserving Super Resolution With Gradient Guidance (CVPR 2020 · similarity 0.97)
- Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network (CVPR 2017 · similarity 0.91)
- GSDD: Generative Space Dataset Distillation for Image Super-resolution (AAAI 2024 · similarity 0.89)
- Noise-Started One-Step Image Super-Resolution via LR-Conditioned SplitMeanFlow and GAN Refinement (NeurIPS 2026 · similarity 0.88)
- Training Generative Image Super-Resolution Models by Wavelet-Domain Losses Enables Better Control of Artifacts (CVPR 2024 · similarity 0.88)
- SeD: Semantic-Aware Discriminator for Image Super-Resolution (CVPR 2024 · similarity 0.88)
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.
