Research map / Image generation / Image restoration
Noisier2Noise: Learning to Denoise From Unpaired Noisy Data
Check 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 α=0.5 rather than α=1 for synthetic Gaussian noise (ablation) | Noisier2Noise · KODAK, Gaussian noise σ=0.05 | PSNR: 33.98 -> 35.09 | Helped |
| use α=0.5 rather than α=1 for synthetic Gaussian noise (ablation) | Noisier2Noise · KODAK, Gaussian noise σ=0.1 | PSNR: 30.80 -> 31.80 | Helped |
| use α=0.5 rather than α=1 for synthetic Gaussian noise (ablation) | Noisier2Noise · KODAK, Gaussian noise σ=0.2 | PSNR: 27.92 -> 28.73 | Helped |
| use α=0.25 rather than α=0.5 for synthetic Gaussian noise (ablation) | Noisier2Noise · KODAK, Gaussian noise σ=0.05 | PSNR: 35.09 -> 35.18 | No clear effect |
| use α=0.25 rather than α=0.5 for synthetic Gaussian noise (ablation) | Noisier2Noise · KODAK, Gaussian noise σ=0.1 | PSNR: 31.80 -> 31.79 | Hurt |
| use α=0.25 rather than α=0.5 for synthetic Gaussian noise (ablation) | Noisier2Noise · KODAK, Gaussian noise σ=0.2 | PSNR: 28.73 -> 28.30 | Hurt |
| feed singly-noisy inputs at test time instead of adding synthetic noise (ablation) | Noisier2Noise, α=1 · KODAK, Gaussian noise σ=0.05 | PSNR: 33.98 -> 32.85 | Hurt |
| feed singly-noisy inputs at test time instead of adding synthetic noise (ablation) | Noisier2Noise, α=1 · KODAK, Gaussian noise σ=0.1 | PSNR: 30.80 -> 30.42 | Hurt |
| feed singly-noisy inputs at test time instead of adding synthetic noise (ablation) | Noisier2Noise, α=1 · KODAK, Gaussian noise σ=0.2 | PSNR: 27.92 -> 27.17 | Hurt |
| feed singly-noisy inputs at test time instead of adding synthetic noise (ablation) | Noisier2Noise, α=0.5 · KODAK, Gaussian noise σ=0.05 | PSNR: 35.09 -> 35.42 | Helped |
| feed singly-noisy inputs at test time instead of adding synthetic noise (ablation) | Noisier2Noise, α=0.5 · KODAK, Gaussian noise σ=0.1 | PSNR: 31.80 -> 31.96 | Helped |
| feed singly-noisy inputs at test time instead of adding synthetic noise (ablation) | Noisier2Noise, α=0.5 · KODAK, Gaussian noise σ=0.2 | PSNR: 28.73 -> 28.69 | Hurt |
Closest papers by meaning
- GAN2GAN: Generative Noise Learning for Blind Denoising with Single Noisy Images (ICLR 2021 · similarity 0.91)
- Noise2Score: Tweedie’s Approach to Self-Supervised Image Denoising without Clean Images (NeurIPS 2021 · similarity 0.91)
- IDR: Self-Supervised Image Denoising via Iterative Data Refinement (CVPR 2022 · similarity 0.90)
- Noise2Void - Learning Denoising From Single Noisy Images (CVPR 2019 · similarity 0.90)
- Is Noise Conditioning Necessary for Denoising Generative Models? (ICML 2025 · similarity 0.89)
- Zero-Shot Noise2Mean: Gap Minimization for Efficient Denoising from a Single Noisy Image (AAAI 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.