Research map / Image generation / Super-resolution
Gated Fusion Network for Degraded Image Super Resolution
Xinyi Zhang, Hang Dong, Zhe Hu, Wei‐Sheng Lai, Fei Wang, Ming–Hsuan Yang
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 restoration module to baseline (restoration first) (ablation) | GFN baseline · LR-GOPRO 4x (blurry SR) | PSNR (dB): 26.20 -> 26.82 | Improved |
| add restoration module to baseline (restoration first) (ablation) | GFN baseline · LR-RESIDE 4x (hazy SR) | PSNR (dB): 23.58 -> 24.82 | Improved |
| apply SR first then restoration instead of restoration first (ablation) | GFN Model-1 · LR-GOPRO 4x (blurry SR) | PSNR (dB): 26.82 -> 27.00; time 0.09 -> 0.57 | Mixed |
| use dual-branch with addition fusion instead of sequential restoration then SR (ablation) | GFN Model-1 · LR-GOPRO 4x (blurry SR) | PSNR (dB): 26.82 -> 27.01 | Improved |
| use dual-branch with addition fusion instead of sequential restoration then SR (ablation) | GFN Model-1 · LR-RESIDE 4x (hazy SR) | PSNR (dB): 24.82 -> 25.10 | Improved |
| add feature-level fusion with restoration-branch loss (ablation) | GFN Model-3 · LR-GOPRO 4x (blurry SR) | PSNR (dB): 27.01 -> 27.53 | Improved |
| add feature-level fusion with restoration-branch loss (ablation) | GFN Model-3 · LR-RESIDE 4x (hazy SR) | PSNR (dB): 25.10 -> 25.21 | Improved |
| add gate module with 3 recursive blocks (ablation) | GFN Model-4 · LR-GOPRO 4x (blurry SR) | PSNR (dB): 27.53 -> 27.91 | Improved |
| add gate module with 3 recursive blocks (ablation) | GFN Model-4 · LR-RESIDE 4x (hazy SR) | PSNR (dB): 25.21 -> 25.77 | Improved |
| use 4 recursive blocks in gate module instead of 3 (ablation) | GFN · LR-GOPRO 4x (blurry SR) | PSNR (dB): 27.91 -> 27.86 | No clear effect |
| replace ResBlock with Residual Dense Block (ablation) | GFN · LR-RESIDE 4x (hazy SR) | PSNR (dB): 25.77 -> 25.88 | Improved |
| replace restoration module with GCANet dilation architecture (ablation) | GFN · LR-RESIDE 4x (hazy SR) | PSNR (dB): 25.77 -> 25.83 | Improved |
Closest papers by meaning
- Learning a Single Convolutional Super-Resolution Network for Multiple Degradations (CVPR 2018 · similarity 0.93)
- Learning with Privileged Information for Efficient Image Super-Resolution (ECCV 2020 · similarity 0.90)
- Zero-Shot Image Super-Resolution with Depth Guided Internal Degradation Learning (ECCV 2020 · similarity 0.90)
- Image Super-Resolution Using Deep Convolutional Networks (TPAMI 2016 · similarity 0.89)
- Unsupervised Degradation Representation Learning for Blind Super-Resolution (CVPR 2021 · similarity 0.89)
- Finding Discriminative Filters for Specific Degradations in Blind Super-Resolution (NeurIPS 2021 · similarity 0.89)
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.
