NVlabs/noise2noise
TensorFlow implementation of the Noise2Noise deep learning paper for learning image restoration from corrupted examples without clean data.
Not currently ranked — collecting fresh signals.
star history
This repository contains the official implementation of a research paper that demonstrates neural networks can learn to restore images by training only on corrupted examples, without needing clean target data. The approach handles diverse corruption types including photographic noise, synthetic Monte Carlo rendering noise, and undersampled MRI scans. It provides training code using the ImageNet validation dataset as well as MRI-specific denoising instructions.
Frequently asked
- What is NVlabs/noise2noise?
- TensorFlow implementation of the Noise2Noise deep learning paper for learning image restoration from corrupted examples without clean data.
- Is noise2noise open source?
- Yes — NVlabs/noise2noise is an open-source project tracked on heatdrop.
- What language is noise2noise written in?
- NVlabs/noise2noise is primarily written in Python.
- How popular is noise2noise?
- NVlabs/noise2noise has 1.6k stars on GitHub.
- Where can I find noise2noise?
- NVlabs/noise2noise is on GitHub at https://github.com/NVlabs/noise2noise.