Research map / Image generation / Super-resolution
Rank-One Network: An Effective Framework for Image Restoration
Shangqi Gao, Xiahai Zhuang
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 3 rank-one projection units instead of 1 (ablation) | Realistic image SR, DIV2K RONet · Realistic SR ablation | PSNR: 24.31 -> 24.33 | Improved |
| use 6 rank-one projection units instead of 3 (ablation) | RONet · Realistic SR ablation | PSNR: 24.33 -> 24.21 | Hurt |
| use L1 loss instead of L2 (ablation) | RONet · Realistic SR ablation | PSNR: 24.33 -> 24.50 | Improved |
| use more residual blocks (12,24,3) or (8,16,8) instead of (3,6,3) (ablation) | RONet · Realistic SR ablation | PSNR: 24.50 -> 24.53 | No clear effect |
| use Xavier uniform instead of MSRA normal initialization (ablation) | RONet · Realistic SR ablation | PSNR: 24.46 -> 24.50 | Improved |
| add batch normalization (ablation) | RONet · Realistic SR ablation | SSIM: 0.5802 -> 0.5814 | Improved |
| multiply residual branch by 0.1 (residual scaling) (ablation) | RONet · Realistic SR ablation | PSNR: 24.46 -> 24.42 | Hurt |
| use unsupervised instead of supervised RO decomposition (ablation) | RONet · Realistic SR ablation | PSNR: 24.50 -> 24.50 | No clear effect |
| use learned RO decomposition instead of SVD (ablation) | RONet · Realistic SR ablation | training time: 99.4h -> 17.0h | Improved |
| train end-to-end instead of pretraining RODec (ablation) | RONet · Realistic SR ablation | PSNR: 24.50 -> 24.47 | No clear effect |
| add deep supervision lambda=0.001 versus lambda=0 (ablation) | RONet-R · Realistic SR DIV2K val | PSNR: 24.50 -> 24.53 | Improved |
| use RONet-R instead of EDSR/WDSR | realistic SR · Realistic SR DIV2K val | PSNR: 24.42 -> 24.53 | Improved |
Closest papers by meaning
- A Restoration Network as an Implicit Prior (ICLR 2024 · similarity 0.85)
- Omni-Kernel Network for Image Restoration (AAAI 2024 · similarity 0.85)
- IR-CM: The Fast and General-purpose Image Restoration Method Based on Consistency Model (NeurIPS 2024 · similarity 0.84)
- APR-RD: Complemental Two Steps for Self-Supervised Real Image Denoising (AAAI 2025 · similarity 0.84)
- Image Restoration by Denoising Diffusion Models with Iteratively Preconditioned Guidance (CVPR 2024 · similarity 0.84)
- One-Step Effective Diffusion Network for Real-World Image Super-Resolution (NeurIPS 2024 · similarity 0.83)
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.
