/pgdgan

Code for the paper: Solving Linear Inverse Problems using GAN priors

Primary LanguagePythonMIT LicenseMIT

Solving Linear Inverse Problems using GANs

Code for the paper: Solving Linear Inverse Problems Using GAN Priors: An Algorithm with Provable Guarantees.

Requirements

To run this code, you require Python 2.7, Tensorflow 1.0.1 (preferably with GPU support), Scipy and PyPNG.

Instructions

  1. Clone the repository, and run all the commands from the parent directory, pgdgan/.

  2. Download the datasets with the script*:

    $ ./setup/download_data.sh

    [update] I noticed that the download links for the data and the model no longer works. Since I used the same dataset and model as Bora et al., kindly visit their github repo and follow the updated download instructions from there.

  3. To train the DCGAN on celebA from scratch, please visit https://github.com/carpedm20/DCGAN-tensorflow, and follow the instructions. Else, pretrained GAN model is available, courtesy Bora et al. To download it, please run the following script*:

    $ ./setup/download_models.sh

    Make sure the model is located at ./models/celebA_64_64/

  4. Run following to run the experiment:

    $ python ./src/pgdgan.py

    You can also use the script available in ./exp_scripts/

* replicated from https://github.com/AshishBora/csgm .