Research map / Image generation / Image restoration
Scale-Recurrent Network for Deep Image Deblurring
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 multi-scale processing instead of a single-scale input (ablation) | SRN-DeblurNet · GOPRO test | PSNR: 28.40dB vs. 29.98dB | Helped |
| share network weights across scales instead of using independent scale-specific parameters (ablation) | multi-scale deblurring network · GOPRO test | PSNR: 29.05 vs 29.26 | Helped |
| add an explicit recurrent bottleneck module (ablation) | shared-weight multi-scale network · GOPRO test | PSNR: 29.26 -> 29.35 | Helped |
| replace vanilla RNN with ConvLSTM (ablation) | scale-recurrent network · GOPRO test | PSNR: 29.35 -> 29.98 | Helped |
| use encoder-decoder ResBlocks instead of flat convolution layers (ablation) | scale-recurrent network · GOPRO test | PSNR: 27.53 -> 29.98 | Helped |
| use encoder-decoder ResBlocks instead of same-resolution ResBlocks (ablation) | scale-recurrent network · GOPRO test | PSNR: 28.11 -> 29.98 | Helped |
| add ResBlocks to the encoder-decoder blocks (ablation) | scale-recurrent network · GOPRO test | PSNR: 29.06 -> 29.98 | Helped |
| increase the number of ResBlocks per encoder/decoder block from 1 to 2 (ablation) | scale-recurrent network · GOPRO test | PSNR: 28.60 -> 29.32 | Helped |
| increase the number of ResBlocks per encoder/decoder block from 2 to 3 (ablation) | scale-recurrent network · GOPRO test | PSNR: 29.32 -> 29.98 | Helped |
| use SRN-DeblurNet instead of Nah et al. | image deblurring · GOPRO test | PSNR: 29.08 -> 30.10 | Helped |
| use SRN-DeblurNet instead of Nah et al. | image deblurring · Köhler Dataset | MSSIM: 0.8079 -> 0.8375 | Helped |
Closest papers by meaning
- Recurrent Neural Networks With Intra-Frame Iterations for Video Deblurring (CVPR 2019 · similarity 0.90)
- Dynamic Scene Deblurring Using Spatially Variant Recurrent Neural Networks (CVPR 2018 · similarity 0.90)
- Deep Recurrent Neural Network with Multi-Scale Bi-directional Propagation for Video Deblurring (AAAI 2022 · similarity 0.90)
- Efficient Multi-scale Network with Learnable Discrete Wavelet Transform for Blind Motion Deblurring (CVPR 2024 · similarity 0.89)
- DeblurDiff: Real-Word Image Deblurring with Generative Diffusion Models (NeurIPS 2025 · similarity 0.86)
- Cross-Scale Self-Supervised Blind Image Deblurring via Implicit Neural Representation (NeurIPS 2024 · 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.