Research map / Image generation / Image restoration
Joint Demosaicking and Denoising by Fine-Tuning of Bursts of Raw Images
Thibaud Ehret, Axel Davy, Pablo Arias, Gabriele Facciolo
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 without RGB ground truth using mosaic-to-mosaic loss instead of supervised ground truth | Section 2 demosaicking network · Kodak demosaicking | PSNR (dB): 41.2 -> 41.3 | Improved |
| fine-tune on a single image's burst (ablation) | Section 2 network · Kodak kodim19 | PSNR (dB): 42.1 -> 44.4 | Improved |
| fine-tune on kodim19 and evaluate on whole dataset (ablation) | Section 2 network · Kodak (full dataset) | PSNR (dB): 41.3 -> 40.4 | Hurt |
| add fine-tuning to Gharbi et al. network | Gharbi et al. · Kodak all sigma=5 | PSNR (dB): 36.2 -> 37.6 | Improved |
| add fine-tuning to Gharbi et al. network | Gharbi et al. · Kodak 19 sigma=10 | PSNR (dB): 33.2 -> 34.0 | Improved |
| fine-tune the noise-free Section 2 network (never saw noise) | Section 2 network · Kodak all sigma=5 | PSNR (dB): 36.0 vs Kokkinos 35.2, Gharbi 36.2 | Mixed |
| lengthen burst from 10 to 20 images (ablation) | Section 2 network + fine-tuning · Kodak 19 sigma=10 | PSNR (dB): 32.9 -> 33.2 | Improved |
| lengthen burst from 10 to 20 images (ablation) | Gharbi et al. + fine-tuning · Kodak 19 sigma=10 | PSNR (dB): 34.0 -> 34.3 | Improved |
| use L1 instead of L2 loss for fine-tuning on clipped noise (ablation) | DnCNN fine-tuned on burst · saturated-region grayscale burst | PSNR | Improved |
| fine-tune DnCNN on a burst | DnCNN sigma=25 · 10-frame noisy grayscale burst | PSNR | Improved |
| fine-tune on a self-similar stripes image vs a binary noise image (ablation) | DnCNN fine-tuning · synthetic stripes vs binary noise | PSNR | Mixed |
Closest papers by meaning
- Deep Image Demosaicking using a Cascade of Convolutional Residual Denoising Networks (ECCV 2018 · similarity 0.92)
- Joint Demosaicking and Denoising in the Wild: The Case of Training Under Ground Truth Uncertainty (AAAI 2021 · similarity 0.91)
- Joint Demosaicing and Denoising With Self Guidance (CVPR 2020 · similarity 0.89)
- End-to-End Learning for Joint Image Demosaicing, Denoising and Super-Resolution (CVPR 2021 · similarity 0.89)
- Not All Noise Is Harmful: Towards Perception Aware and Controllable RAW Image Joint Denoising and Demosaicing (NeurIPS 2026 · similarity 0.88)
- Joint Demosaicing and Denoising for Spike Camera (AAAI 2024 · similarity 0.85)
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.
