Research map / Image generation / Image restoration
Single Image Defocus Deblurring Using Kernel-Sharing Parallel Atrous Convolutions
Hyeongseok Son, Junyong Lee, Sunghyun Cho, Seungyong Lee
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 weight sharing across atrous convolution layers (ablation) | KPAC block · DPDD | PSNR: 24.78 -> 25.21 | Improved |
| replace residual blocks with atrous convolution layers (ablation) | KPAC network · DPDD | PSNR: 24.59 -> 24.74 | Improved |
| add scale attention to atrous convolutions (ablation) | KPAC network · DPDD | PSNR: 24.74 -> 25.03 | Improved |
| add shape attention to atrous convolutions (ablation) | KPAC network · DPDD | PSNR: 24.74 -> 24.98 | Improved |
| add shape attention on top of scale attention (ablation) | KPAC network · DPDD | PSNR: 25.03 -> 25.21 | Improved |
| add scale attention on top of shape attention (ablation) | KPAC network · DPDD | PSNR: 24.98 -> 25.21 | Improved |
| use two KPAC blocks instead of one (ablation) | KPAC network · DPDD | PSNR: 24.82 -> 25.21 | Improved |
| use three KPAC blocks instead of two (ablation) | KPAC network · DPDD | PSNR: 25.21 -> 25.25 | No clear effect |
| use four KPAC blocks instead of three (ablation) | KPAC network · DPDD | PSNR: 25.25 -> 25.14 | Hurt |
| train with additive Gaussian noise augmentation (ablation) | KPAC network · DPDD noise sigma 5% | PSNR: 24.23 -> 24.85 | Improved |
| train with additive Gaussian noise augmentation (ablation) | KPAC network · DPDD noise sigma 1% | PSNR: 25.13 -> 25.06 | No clear effect |
| use 3-level encoder instead of 2-level (ablation) | KPAC network · DPDD | PSNR: 25.24 -> 25.24 | No clear effect |
Closest papers by meaning
- Prior and Prediction Inverse Kernel Transformer for Single Image Defocus Deblurring (AAAI 2024 · similarity 0.94)
- Single Image Defocus Deblurring via Implicit Neural Inverse Kernels (ICCV 2023 · similarity 0.92)
- Learnable Blur Kernel for Single-Image Defocus Deblurring in the Wild (AAAI 2023 · similarity 0.92)
- Neumann Network With Recursive Kernels for Single Image Defocus Deblurring (CVPR 2023 · similarity 0.91)
- Residual Diffusion Deblurring Model for Single Image Defocus Deblurring (AAAI 2025 · similarity 0.90)
- Quad-Pixel Image Defocus Deblurring: A New Benchmark and Model (CVPR 2025 · 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.
