Research map / Video / Video segmentation and tracking
Temporal As a Plugin: Unsupervised Video Denoising with Pre-Trained Image Denoisers
Zixuan Fu, Lanqing Guo, Chong Wang, Yufei Wang, Zhihao Li, Bihan Wen
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 |
|---|---|---|---|
| fine-tune only temporal modules instead of all parameters (ablation) | TAP · DAVIS sigma=30 | PSNR (dB): 34.20 dB -> 34.84 dB | Improved |
| fine-tune temporal modules progressively instead of jointly (ablation) | TAP · DAVIS sigma=30 | PSNR (dB): 34.55 dB -> 34.84 dB | Improved |
| repeat the whole fine-tuning process twice (ablation) | TAP · DAVIS sigma=30 | PSNR (dB): 34.84 dB -> 34.86 dB | No clear effect |
| add progressive fine-tuned temporal modules to pre-trained NAFNet | NAFNet image denoiser · DAVIS (Gaussian average) | PSNR (dB): 34.20 -> 35.48 | Improved |
| add progressive fine-tuned temporal modules to pre-trained NAFNet | NAFNet image denoiser · Set8 (Gaussian average) | PSNR (dB): 32.43 -> 33.95 | Improved |
| TAP compared with unsupervised RDRF on CRVD indoor | TAP · CRVD indoor raw | PSNR (dB): 45.09 -> 45.15 | Improved |
| add temporal modules to NAFNet | NAFNet image denoiser · Static toy dataset temporal coherence | MAE of adjacent frames (x10^-3): 16.5 -> 5.8 | Improved |
Closest papers by meaning
- Unsupervised Deep Video Denoising with Untrained Network (AAAI 2023 · similarity 0.91)
- Classic Video Denoising in a Machine Learning World: Robust, Fast, and Controllable (CVPR 2025 · similarity 0.89)
- Unsupervised Deep Video Denoising (ICCV 2021 · similarity 0.89)
- TempFormer: Temporally Consistent Transformer for Video Denoising (ECCV 2022 · similarity 0.89)
- Supervised Raw Video Denoising With a Benchmark Dataset on Dynamic Scenes (CVPR 2020 · similarity 0.88)
- TRIP: Temporal Residual Learning with Image Noise Prior for Image-to-Video Diffusion Models (CVPR 2024 · 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.
