Research map / Image generation / Fast diffusion sampling
Real-Time Controllable Denoising for Image and Video
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 multi-level noise-map control with RCD to NAFNet | NAFNet · Gaussian image denoising, CBSD68, Kodak24, McMaster, and Urban100 | PSNR: 34.11 -> 34.13 (CBSD68, σ=15); 35.14 -> 35.15 (Kodak24, σ=15); 35.07 -> 35.11 (McMaster, σ=15); 34.41 -> 34.45 (Urban100 | No clear effect |
| Add RCD to NAFNet on real SIDD data | NAFNet · SIDD real noise test set | PSNR: 43.22 -> 42.91 | Hurt |
| Add RCD to NAFNet on synthetic-noise SIDD | NAFNet · SIDD synthetic noise | PSNR: 38.85 -> 39.14 | Helped |
| Add RCD to FastDVD | FastDVD · DAVIS and Set 8 video denoising, 1-frame and 5-frame settings | PSNR: 34.17 -> 34.21 (DAVIS, σ=20, 1 frame); 35.69 -> 35.65 (DAVIS, σ=20, 5 frames); 29.41 -> 29.60 (Set 8, σ=50, 5 frames) | Mixed |
| Use RCD with the tiny NAFNet backbone (ablation) | NAFNet-tiny · CBSD68, σ=15 | PSNR: 33.58 -> 33.71 | Helped |
| Use 12 predefined noise levels instead of 1 (ablation) | FastDVD-RCD · Set8, σ=20 | PSNR: 31.87 -> 32.01 | Helped |
| Increase the number of predefined noise levels from 12 to 30 (ablation) | FastDVD-RCD · Set8, σ=20 | PSNR: 32.01 -> 31.39 | Hurt |
| Increase the number of predefined noise levels from 12 to 60 (ablation) | FastDVD-RCD · Set8, σ=20 | PSNR: 32.01 -> 31.07 | Hurt |
| Replace repeated network inference for control with network-free noise-map editing | NAFNet-RCD · One image, 1000 control settings | latency: 81.03s -> 0.04s (AdaFM-Net); 128.08s -> 0.04s (CResMDNet) | Helped |
| Add RCD to NAFNet on Nam real-world data | NAFNet · Nam | PSNR: 38.93 -> 39.03; 39.54 -> 39.67 | Helped |
| Add RCD to Restormer | Restormer · SIDD synthetic noise | PSNR: 41.31 -> 41.79 | Helped |
Closest papers by meaning
- Generative Adaptive Convolutions for Real-World Noisy Image Denoising (AAAI 2022 · similarity 0.90)
- Classic Video Denoising in a Machine Learning World: Robust, Fast, and Controllable (CVPR 2025 · similarity 0.90)
- Toward Convolutional Blind Denoising of Real Photographs (CVPR 2019 · similarity 0.89)
- APR-RD: Complemental Two Steps for Self-Supervised Real Image Denoising (AAAI 2025 · similarity 0.87)
- Zero-Shot Noise2Mean: Gap Minimization for Efficient Denoising from a Single Noisy Image (AAAI 2025 · similarity 0.87)
- LAN: Learning to Adapt Noise for Image Denoising (CVPR 2024 · similarity 0.87)
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.