Research map / Image generation / Image restoration
Simple Baselines for Image Restoration
Liangyu Chen, Xiaojie Chu, Xiangyu Zhang, Jian Sun
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 LayerNorm (allowing 10x larger lr) (ablation) | PlainNet · GoPro | PSNR: 28.51 -> 31.90 | Improved |
| add LayerNorm (allowing 10x larger lr) (ablation) | PlainNet · SIDD | PSNR: 39.29 -> 39.73 | Improved |
| replace ReLU with GELU (ablation) | PlainNet+LN · SIDD | PSNR: 39.73 -> 39.71 | No clear effect |
| replace ReLU with GELU (ablation) | PlainNet+LN · GoPro | PSNR: 31.90 -> 32.11 | Improved |
| add channel attention (ablation) | PlainNet+LN+GELU · SIDD | PSNR: 39.71 -> 39.85 | Improved |
| add channel attention (ablation) | PlainNet+LN+GELU · GoPro | PSNR: 32.11 -> 32.35 | Improved |
| replace GELU with SimpleGate (ablation) | Baseline · GoPro | PSNR: 32.35 -> 32.76 | Improved |
| replace GELU with SimpleGate (ablation) | Baseline · SIDD | PSNR: 39.85 -> 39.93 | Improved |
| replace Channel Attention with Simplified Channel Attention (ablation) | NAFNet baseline · GoPro | PSNR: 32.76 -> 32.85 | Improved |
| replace Channel Attention with Simplified Channel Attention (ablation) | NAFNet baseline · SIDD | PSNR: 39.93 -> 39.96 | No clear effect |
| increase blocks from 36 to 72 (ablation) | NAFNet · GoPro | PSNR: 32.85 -> 32.88 | No clear effect |
| increase blocks from 9 to 36 (ablation) | NAFNet · GoPro | PSNR: 31.79 -> 32.85 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- HINet — HiNet: Deep Image Hiding by Invertible Network (ICCV 2021) · better on 3 of 3 shared results
- Restormer — Restormer: Efficient Transformer for High-Resolution Image Restoration (CVPR 2022) · better on 2 of 2 shared results
- NBNet — NBNet: Noise Basis Learning for Image Denoising With Subspace Projection (CVPR 2021) · better on 1 of 1 shared result
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
- xUnit: Learning a Spatial Activation Function for Efficient Image Restoration (CVPR 2018 · similarity 0.87)
- Functional Neural Networks for Parametric Image Restoration Problems (NeurIPS 2021 · similarity 0.87)
- MR-VNet: Media Restoration using Volterra Networks (CVPR 2024 · similarity 0.85)
- A Restoration Network as an Implicit Prior (ICLR 2024 · similarity 0.85)
- Image Restoration by Denoising Diffusion Models with Iteratively Preconditioned Guidance (CVPR 2024 · similarity 0.85)
- Zero-Shot Image Restoration Using Few-Step Guidance of Consistency Models (and Beyond) (CVPR 2025 · similarity 0.85)
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.
