Research map / Label-efficient and robust learning / Self-supervised and contrastive
AP-BSN: Self-Supervised Denoising for Real-World Images via Asymmetric PD and Blind-Spot Network
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 |
|---|---|---|---|
| Add random-replacing refinement to AP-BSN (ablation) | AP-BSN · SIDD benchmark | PSNR (dB): 34.90 -> 35.97 | Helped |
| Add random-replacing refinement to AP-BSN (ablation) | AP-BSN · SIDD benchmark | SSIM: 0.900 -> 0.925 | Helped |
| Add random-replacing refinement to AP-BSN (ablation) | AP-BSN · DND benchmark | PSNR (dB): 37.46 -> 38.09 | Helped |
| Add random-replacing refinement to AP-BSN (ablation) | AP-BSN · DND benchmark | SSIM: 0.924 -> 0.937 | Helped |
| Train and evaluate on SIDD benchmark images rather than the SIDD-Medium training set (ablation) | AP-BSN + R³ · SIDD benchmark | PSNR (dB): 35.97 -> 36.91 | Helped |
| Use training stride 5 and inference stride 2 instead of the same PD stride (ablation) | PD-BSN · SIDD real-world denoising | PSNR | Helped |
| Increase training stride from 5 to 6 (ablation) | AP-BSN · SIDD | PSNR | Hurt |
| Increase training stride from 5 to 6 (ablation) | AP-BSN · NIND | PSNR | Mixed |
| Increase inference stride above 2 (ablation) | AP-BSN · SIDD real-world denoising | PSNR | Hurt |
| Use inference stride 1, without PD (ablation) | AP-BSN · SIDD real-world denoising | PSNR | Hurt |
| Use fixed, structured replacement masks instead of randomized masks (ablation) | R³ refinement · Real-world image denoising | PSNR | Hurt |
| Increase replacement probability above approximately 0.16 (ablation) | R³ refinement · Real-world image denoising | PSNR | Hurt |
Closest papers by meaning
- Exploring Efficient Asymmetric Blind-Spots for Self-Supervised Denoising in Real-World Scenarios (CVPR 2024 · similarity 0.93)
- APR-RD: Complemental Two Steps for Self-Supervised Real Image Denoising (AAAI 2025 · similarity 0.91)
- Pseudo-Siamese Blind-spot Transformers for Self-Supervised Real-World Denoising (NeurIPS 2024 · similarity 0.89)
- Rethinking Transformer-Based Blind-Spot Network for Self-Supervised Image Denoising (AAAI 2025 · similarity 0.89)
- Masked and Shuffled Blind Spot Denoising for Real-World Images (CVPR 2024 · similarity 0.87)
- Zero-Shot Blind-spot Image Denoising via Implicit Neural Sampling (CVPR 2025 · 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.