Skip to content
No description, website, or topics provided.
Jupyter Notebook Python
Branch: master
Clone or download
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
include added jacobian of mult-layer dd over time simulation Oct 28, 2019
test_data added files for figures Oct 29, 2019
Jacobian_inner_product_noisevsimg.ipynb
Jacobian_multi_layer_deep_decoder.ipynb added jacobian of mult-layer dd over time simulation Oct 28, 2019
README.md added experiments for denoising performance Nov 12, 2019
denoising_MSE_curves.ipynb added files for figures Oct 29, 2019
denoising_bm3d_example.ipynb added denoising experiment Nov 9, 2019
denoising_bm3d_imagenet_selected100_paper.ipynb added experiments for denoising performance Nov 12, 2019
denoising_imagenet_selected100_paper.ipynb added experiments for denoising performance Nov 12, 2019
denoising_performance_example.ipynb added denoising experiment Nov 9, 2019
image_fitted_faster_than_noise_on_imgnet.ipynb added files for figures Oct 29, 2019
kernels_and_associated_dual_kernels.ipynb inital commit Oct 1, 2019
linear_least_squares_selective_fitting_warmup.ipynb
noise_vs_img_fitting_different_architectures.ipynb added files for figures Oct 29, 2019
visualization_linear_approximation.ipynb new visualization linear approximation Oct 25, 2019

README.md

Denoising and Regularization via Exploiting the Structural Bias of Convolutional Generators

This repository provides code for reproducing the figures in the paper:

``Denoising and Regularization via Exploiting the Structural Bias of Convolutional Generators'', by Reinhard Heckel and Mahdi Soltanolkotabi. Contact: reinhard.heckel@gmail.com

The paper is available online here.

Organization

  • Figure 1: denoising_MSE_curves.ipynb
  • Figure 2: denoising_performance_example.ipynb, denoising_bm3d_example.ipynb
  • Figure 4,8: noise_vs_img_fitting_different_architectures.ipynb
  • Figure 5: linear_least_squares_selective_fitting_warmup.ipynb
  • Figure 6: kernels_and_associated_dual_kernels.ipynb
  • Figure 7: Jacobian_multi_layer_deep_decoder.ipynb
  • Figure 10: image_fitted_faster_than_noise_on_imgnet.ipynb
  • Figure 12: Jacobian_inner_product_noisevsimg.ipynb
  • Table 1: denoising_imagenet_selected100_paper.ipynb, denoising_bm3d_imagenet_selected100_paper.ipynb

Installation

The code is written in python and relies on pytorch. The following libraries are required:

  • python 3
  • pytorch
  • numpy
  • skimage
  • matplotlib
  • scikit-image
  • jupyter

The libraries can be installed via:

conda install jupyter

A small part of the code compares performance to the deep image prior. This part requires downloading the models folder from https://github.com/DmitryUlyanov/deep-image-prior.

Citation

@article{heckel_denoising_2019,
    author    = {Reinhard Heckel and Mahdi Soltanolkotabi},
    title     = {Denoising and Regularization via Exploiting the Structural Bias of Convolutional Generators},
    journal   = {arXiv:1910.14634 [cs.LG]},
    year      = {2019}
}

Licence

All files are provided under the terms of the Apache License, Version 2.0.

You can’t perform that action at this time.