Research map / Image generation / Super-resolution
Efficient Cascaded Multiscale Adaptive Network for Image Restoration
Yichen Zhou, Pan Zhou, Teck Khim Ng
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 local adaptive convolution instead of element-wise addition in LAM (ablation) | ECMA-T · GoPro deblurring (ablation) | PSNR: 32.36 -> 32.98 | Improved |
| use local adaptive convolution instead of direct concatenation (ablation) | ECMA-T · GoPro deblurring (ablation) | PSNR: 32.30 -> 32.98 | Improved |
| cascade LAMs from large to small kernel instead of small to large (ablation) | ECMA-T · GoPro deblurring (ablation) | PSNR: 32.64 -> 32.98 | Improved |
| add cascading between multiscale LAMs (ablation) | ECMA-T · GoPro deblurring (ablation) | PSNR: 32.56 -> 32.98 | Improved |
| include the large-kernel LAM (ablation) | ECMA-T · GoPro deblurring (ablation) | PSNR: 32.28 -> 32.98 | Improved |
| use different kernel sizes instead of three identical LAMs (ablation) | ECMA-T · GoPro deblurring (ablation) | PSNR: 32.26-32.51 -> 32.98 | Improved |
| use FFT for the large-kernel convolution instead of spatial domain (ablation) | ECMA-T · GoPro deblurring (ablation) | PSNR: 32.47 -> 32.98 | Improved |
| ECMA-B vs GRL-B on GoPro | ECMA-B · GoPro deblurring | PSNR: 33.93 -> 34.14 | Improved |
| ECMA-S vs NAFNet on SIDD denoising | ECMA-S · SIDD denoising | PSNR: 40.30 vs 40.28 | No clear effect |
| ECMA-T on HIDE vs Uformer-B | ECMA-T · HIDE deblurring | PSNR: 30.90 vs 30.89 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- SRFormer — SRFormer: Permuted Self-Attention for Single Image Super-Resolution (ICCV 2023) · better on 1 of 2 shared results
- Restormer — Restormer: Efficient Transformer for High-Resolution Image Restoration (CVPR 2022) · 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
- Learning Enriched Features for Fast Image Restoration and Enhancement (TPAMI 2022 · similarity 0.85)
- Learning Enriched Features for Real Image Restoration and Enhancement (ECCV 2020 · similarity 0.85)
- Scale-Wise Convolution for Image Restoration (AAAI 2020 · similarity 0.85)
- Adaptive Dynamic Filtering Network for Image Denoising (AAAI 2023 · similarity 0.85)
- Omni-Kernel Network for Image Restoration (AAAI 2024 · similarity 0.84)
- DeblurDiff: Real-Word Image Deblurring with Generative Diffusion Models (NeurIPS 2025 · 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.
