Research map / Image generation / Image restoration
Efficient Deep Image Denoising via Class Specific Convolution
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 |
|---|---|---|---|
| Replace selected EDSR convolutions with class-specific filters | EDSR16 denoiser · BSD68, σ=50 | PSNR: 26.77 -> 27.31 | Helped |
| Replace selected EDSR convolutions with class-specific filters | EDSR16 denoiser · DIV2K, σ=50 | PSNR: 28.51 -> 29.17 | Helped |
| Replace selected EDSR convolutions with class-specific filters | EDSR16 denoiser · Urban100, σ=25 | PSNR: 29.96 -> 30.17 | Helped |
| Replace selected EDSR convolutions with class-specific filters | EDSR16 denoiser · Urban100, σ=50 | PSNR: 26.44 -> 27.07 | Helped |
| Use PCN to classify pixels rather than applying RAISR eigenanalysis directly to noisy images (ablation) | CS-EDSR · BSD68, σ=25 | PSNR: 29.42 -> 29.76 | Helped |
| Use PCN to classify pixels rather than applying RAISR eigenanalysis directly to noisy images (ablation) | CS-EDSR · Urban100, σ=25 | PSNR: 29.36 -> 30.17 | Helped |
| Use PCN to classify pixels rather than applying RAISR eigenanalysis directly to noisy images (ablation) | CS-EDSR · DIV2K, σ=25 | PSNR: 31.28 -> 31.74 | Helped |
| Use PCN to classify pixels rather than applying RAISR eigenanalysis directly to noisy images (ablation) | CS-EDSR · Set12, σ=25 | PSNR: 30.40 -> 30.97 | Helped |
| Use noisy-image PCN classification instead of clean-image RAISR classification (ablation) | CS-EDSR · Set12, σ=25 | PSNR: 31.39 -> 30.97 | Hurt |
| Use 4/2/2 rather than 8/3/3 orientation/strength/coherence classes (ablation) | CS-EDSR · BSD68, σ=25 | PSNR: 29.76 -> 29.67 | Hurt |
| Use 16/6/6 rather than 8/3/3 orientation/strength/coherence classes (ablation) | CS-EDSR · BSD68, σ=25 | PSNR: 29.76 -> 28.74 | Hurt |
| Use 4/2/2 rather than 8/3/3 orientation/strength/coherence classes (ablation) | CS-EDSR · Set12, σ=25 | PSNR: 30.97 -> 30.66 | Hurt |
Closest papers by meaning
- Universal Denoising Networks : A Novel CNN Architecture for Image Denoising (CVPR 2018 · similarity 0.91)
- Robust And Interpretable Blind Image Denoising Via Bias-Free Convolutional Neural Networks (ICLR 2020 · similarity 0.89)
- Non-Local Color Image Denoising With Convolutional Neural Networks (CVPR 2017 · similarity 0.89)
- Toward Convolutional Blind Denoising of Real Photographs (CVPR 2019 · similarity 0.88)
- High-Quality Self-Supervised Deep Image Denoising (NeurIPS 2019 · similarity 0.88)
- Real-World Mobile Image Denoising Dataset with Efficient Baselines (CVPR 2024 · similarity 0.86)
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.