Research map / Image generation / Image restoration
Adaptive Dynamic Filtering Network for Image Denoising
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 |
|---|---|---|---|
| add DCB and MDCB to the decoder (ablation) | ADFNet · Kodak24, noise level 50 | PSNR (dB): 29.61 -> 29.81 | Helped |
| add DCB and MDCB to the decoder (ablation) | ADFNet · BSD68, noise level 50 | PSNR (dB): 28.33 -> 28.43 | Helped |
| use DCB instead of a regular 3×3 convolution block (ablation) | ADFNet · Kodak24, noise level 50 | PSNR (dB): 29.69 -> 29.81 | Helped |
| use DCB instead of a regular 3×3 convolution block (ablation) | ADFNet · BSD68, noise level 50 | PSNR (dB): 28.36 -> 28.43 | Helped |
| use adaptive MDCB instead of a static multi-scale convolution block (ablation) | ADFNet · Kodak24, noise level 50 | PSNR (dB): 29.66 -> 29.81 | Helped |
| use adaptive MDCB instead of a static multi-scale convolution block (ablation) | ADFNet · BSD68, noise level 50 | PSNR (dB): 29.37 -> 28.43 | Hurt |
| add MFI feature fusion (ablation) | ADFNet · Kodak24, noise level 50 | PSNR (dB): 29.77 -> 29.81 | No clear effect |
| add MFI feature fusion (ablation) | ADFNet · BSD68, noise level 50 | PSNR (dB): 28.40 -> 28.43 | Helped |
| deploy DCB and MDCB in the decoder rather than the encoder (ablation) | ADFNet · Kodak24, noise level 50 | PSNR (dB): 29.73 -> 29.81 | Helped |
| deploy DCB and MDCB in the decoder rather than the encoder (ablation) | ADFNet · BSD68, noise level 50 | PSNR (dB): 28.38 -> 28.43 | Helped |
| replace one static 3×3 convolution with DConv | DnCNN · Color Gaussian denoising | PSNR (dB): +0.60 dB, +0.24 dB, and +0.15 dB | Helped |
| use ADFNet rather than SwinIR* | Color Gaussian denoising · Kodak24 | PSNR (dB): 29.79 -> 29.81 | Helped |
Closest papers by meaning
- Multi-Scale Adaptive Network for Single Image Denoising (NeurIPS 2022 · similarity 0.89)
- Generative Adaptive Convolutions for Real-World Noisy Image Denoising (AAAI 2022 · similarity 0.89)
- Universal Denoising Networks : A Novel CNN Architecture for Image Denoising (CVPR 2018 · similarity 0.88)
- Bi-volution: A Static and Dynamic Coupled Filter (AAAI 2022 · similarity 0.88)
- Dynamic Filter Networks (NeurIPS 2016 · similarity 0.87)
- DeblurDiff: Real-Word Image Deblurring with Generative Diffusion Models (NeurIPS 2025 · 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.