Noise2Noise: Learning Image Restoration without Clean Data

Jaakko Lehtinen, Jacob Munkberg, Jon Hasselgren, Samuli Laine, Tero Karras, Miika Aittala, Timo Aila
Proceedings of the 35th International Conference on Machine Learning, PMLR 80:2965-2974, 2018.

Abstract

We apply basic statistical reasoning to signal reconstruction by machine learning - learning to map corrupted observations to clean signals - with a simple and powerful conclusion: it is possible to learn to restore images by only looking at corrupted examples, at performance at and sometimes exceeding training using clean data, without explicit image priors or likelihood models of the corruption. In practice, we show that a single model learns photographic noise removal, denoising synthetic Monte Carlo images, and reconstruction of undersampled MRI scans - all corrupted by different processes - based on noisy data only.

Cite this Paper


BibTeX
@InProceedings{pmlr-v80-lehtinen18a, title = {{N}oise2{N}oise: Learning Image Restoration without Clean Data}, author = {Lehtinen, Jaakko and Munkberg, Jacob and Hasselgren, Jon and Laine, Samuli and Karras, Tero and Aittala, Miika and Aila, Timo}, booktitle = {Proceedings of the 35th International Conference on Machine Learning}, pages = {2965--2974}, year = {2018}, editor = {Dy, Jennifer and Krause, Andreas}, volume = {80}, series = {Proceedings of Machine Learning Research}, month = {10--15 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v80/lehtinen18a/lehtinen18a.pdf}, url = {https://proceedings.mlr.press/v80/lehtinen18a.html}, abstract = {We apply basic statistical reasoning to signal reconstruction by machine learning - learning to map corrupted observations to clean signals - with a simple and powerful conclusion: it is possible to learn to restore images by only looking at corrupted examples, at performance at and sometimes exceeding training using clean data, without explicit image priors or likelihood models of the corruption. In practice, we show that a single model learns photographic noise removal, denoising synthetic Monte Carlo images, and reconstruction of undersampled MRI scans - all corrupted by different processes - based on noisy data only.} }
Endnote
%0 Conference Paper %T Noise2Noise: Learning Image Restoration without Clean Data %A Jaakko Lehtinen %A Jacob Munkberg %A Jon Hasselgren %A Samuli Laine %A Tero Karras %A Miika Aittala %A Timo Aila %B Proceedings of the 35th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2018 %E Jennifer Dy %E Andreas Krause %F pmlr-v80-lehtinen18a %I PMLR %P 2965--2974 %U https://proceedings.mlr.press/v80/lehtinen18a.html %V 80 %X We apply basic statistical reasoning to signal reconstruction by machine learning - learning to map corrupted observations to clean signals - with a simple and powerful conclusion: it is possible to learn to restore images by only looking at corrupted examples, at performance at and sometimes exceeding training using clean data, without explicit image priors or likelihood models of the corruption. In practice, we show that a single model learns photographic noise removal, denoising synthetic Monte Carlo images, and reconstruction of undersampled MRI scans - all corrupted by different processes - based on noisy data only.
APA
Lehtinen, J., Munkberg, J., Hasselgren, J., Laine, S., Karras, T., Aittala, M. & Aila, T.. (2018). Noise2Noise: Learning Image Restoration without Clean Data. Proceedings of the 35th International Conference on Machine Learning, in Proceedings of Machine Learning Research 80:2965-2974 Available from https://proceedings.mlr.press/v80/lehtinen18a.html.

Related Material