Research map / Video / Video segmentation and tracking
Model-Blind Video Denoising via Frame-To-Frame Training
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 |
|---|---|---|---|
| Fine-tune DnCNN 25 offline on frame-to-frame pairs | DnCNN 25 · AWGN σ=50, four-sequence average | PSNR: 29.65 -> 29.75 | Helped |
| Fine-tune DnCNN 25 online on frame-to-frame pairs | DnCNN 25 · AWGN σ=50, four-sequence average | PSNR: 29.65 -> 29.59 | No clear effect |
| Fine-tune DnCNN 25 offline on frame-to-frame pairs | DnCNN 25 · JPEG-compressed AWGN σ=25, compression factor 10 | PSNR: 30.97 -> 31.24 | Helped |
| Fine-tune DnCNN 25 online on frame-to-frame pairs | DnCNN 25 · JPEG-compressed AWGN σ=25, compression factor 10 | PSNR: 30.97 -> 30.94 | No clear effect |
| Use a pre-trained network instead of random initialization (ablation) | Frame-to-frame fine-tuning · Denoising parameter experiment | PSNR: 2dB | Helped |
| Use more frames for fine-tuning before processing the remaining video (ablation) | Frame-to-frame fine-tuning · Denoising parameter experiment | PSNR | Helped |
| Use fewer per-frame optimization iterations (ablation) | Online fine-tuning · Denoising parameter experiment | PSNR variance | Helped |
| Use offline rather than online fine-tuning (ablation) | Frame-to-frame fine-tuning · Multiplicative Gaussian, correlated Gaussian, salt-and-pepper, and JPEG-compressed Gaussian noise | PSNR | Helped |
| Fine-tune the pre-trained network to the video noise | DnCNN 25 · Multiplicative Gaussian, correlated Gaussian, salt-and-pepper, and JPEG-compressed Gaussian noise | PSNR | Helped |
| Fine-tune DnCNN 25 to AWGN σ=25 | DnCNN 25 · AWGN σ=25 | PSNR | No clear effect |
Closest papers by meaning
- Classic Video Denoising in a Machine Learning World: Robust, Fast, and Controllable (CVPR 2025 · similarity 0.88)
- Toward Convolutional Blind Denoising of Real Photographs (CVPR 2019 · similarity 0.87)
- Unsupervised Deep Video Denoising with Untrained Network (AAAI 2023 · similarity 0.87)
- FBI-Denoiser: Fast Blind Image Denoiser for Poisson-Gaussian Noise (CVPR 2021 · similarity 0.87)
- Exploring Efficient Asymmetric Blind-Spots for Self-Supervised Denoising in Real-World Scenarios (CVPR 2024 · similarity 0.85)
- APR-RD: Complemental Two Steps for Self-Supervised Real Image Denoising (AAAI 2025 · 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.