Research map / Image generation / Image restoration
Neighbor2Neighbor: Self-Supervised Denoising From Single Noisy Images
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 |
|---|---|---|---|
| train with random neighbor sub-sampling and the regularized self-supervised loss | Neighbor2Neighbor · SIDD Benchmark | PSNR (dB): 50.62 -> 50.76 | Helped |
| train with random neighbor sub-sampling and the regularized self-supervised loss | Neighbor2Neighbor · SIDD Validation | PSNR (dB): 51.21 -> 51.39 | Helped |
| add the regularization term (γ=2) instead of removing it (ablation) | Neighbor2Neighbor · Kodak, Gaussian σ=25 | PSNR (dB): 31.77 -> 32.08 | Helped |
| increase regularization weight from γ=2 to γ=8 (ablation) | Neighbor2Neighbor · Kodak, Gaussian σ=25 | PSNR (dB): 32.08 -> 32.02 | Hurt |
| increase regularization weight from γ=2 to γ=20 (ablation) | Neighbor2Neighbor · Kodak, Gaussian σ=25 | PSNR (dB): 32.08 -> 31.95 | Hurt |
| use random neighbor sampling rather than fix-location sampling (ablation) | Neighbor2Neighbor · Kodak, Gaussian σ=25 | PSNR (dB): 32.05 -> 32.08 | Helped |
| use random neighbor sampling rather than fix-location sampling (ablation) | Neighbor2Neighbor · Kodak, Gaussian σ ∈ [5, 50] | PSNR (dB): 32.07 -> 32.10 | Helped |
| use random neighbor sampling rather than fix-location sampling (ablation) | Neighbor2Neighbor · Kodak, Poisson λ=30 | PSNR (dB): 31.42 -> 31.44 | Helped |
| use random neighbor sampling rather than fix-location sampling (ablation) | Neighbor2Neighbor · Kodak, Poisson λ ∈ [5, 50] | PSNR (dB): 30.83 -> 30.86 | Helped |
| use random neighbor sampling rather than fix-location sampling (ablation) | Neighbor2Neighbor · SIDD Validation | PSNR (dB): 50.92 -> 51.06 | Helped |
| replace U-Net with RRGs | Neighbor2Neighbor · SIDD Benchmark | PSNR (dB): 50.47 -> 50.76 | Helped |
| replace U-Net with RRGs | Neighbor2Neighbor · SIDD Validation | PSNR (dB): 51.06 -> 51.39 | Helped |
Closest papers by meaning
- Noise2Score: Tweedie’s Approach to Self-Supervised Image Denoising without Clean Images (NeurIPS 2021 · similarity 0.90)
- Zero-Shot Noise2Mean: Gap Minimization for Efficient Denoising from a Single Noisy Image (AAAI 2025 · similarity 0.89)
- Noise2Same: Optimizing A Self-Supervised Bound for Image Denoising (NeurIPS 2020 · similarity 0.89)
- Positive2Negative: Breaking the Information-Lossy Barrier in Self-Supervised Single Image Denoising (CVPR 2025 · similarity 0.88)
- Exploring Efficient Asymmetric Blind-Spots for Self-Supervised Denoising in Real-World Scenarios (CVPR 2024 · similarity 0.87)
- APR-RD: Complemental Two Steps for Self-Supervised Real Image Denoising (AAAI 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.