Zero-Shot Noise2Noise: Efficient Image Denoising Without Any Data

Abstract

Recently, self-supervised neural networks have shown excellent image denoising performance. However, current dataset free methods are either computationally expensive, require a noise model, or have inadequate image quality. In this work we show that a simple 2-layer network, without any training data or knowledge of the noise distribution, can enable high-quality image denoising at low computational cost. Our approach is motivated by Noise2Noise and Neighbor2Neighbor and works well for denoising pixel-wise independent noise. Our experiments on artificial, real-world camera, and microscope noise show that our method termed ZS-N2N (Zero Shot Noise2Noise) often outperforms existing dataset-free methods at a reduced cost, making it suitable for use cases with scarce data availability and limited compute.

Cite

Text

Mansour and Heckel. "Zero-Shot Noise2Noise: Efficient Image Denoising Without Any Data." Conference on Computer Vision and Pattern Recognition, 2023. doi:10.1109/CVPR52729.2023.01347

Markdown

[Mansour and Heckel. "Zero-Shot Noise2Noise: Efficient Image Denoising Without Any Data." Conference on Computer Vision and Pattern Recognition, 2023.](https://mlanthology.org/cvpr/2023/mansour2023cvpr-zeroshot/) doi:10.1109/CVPR52729.2023.01347

BibTeX

@inproceedings{mansour2023cvpr-zeroshot,
  title     = {{Zero-Shot Noise2Noise: Efficient Image Denoising Without Any Data}},
  author    = {Mansour, Youssef and Heckel, Reinhard},
  booktitle = {Conference on Computer Vision and Pattern Recognition},
  year      = {2023},
  pages     = {14018-14027},
  doi       = {10.1109/CVPR52729.2023.01347},
  url       = {https://mlanthology.org/cvpr/2023/mansour2023cvpr-zeroshot/}
}