Simple GAN to be used as a base for other projects, based on Udacity GAN lectures

Purpose

For getting into creating new images with GAN, I found it easiest to abstract out a simple GAN and fiddle with the code

1. Installation

Download the latest version of miniconda that matches your system.

	Linux	Mac	Windows
64-bit	64-bit (bash installer)	64-bit (bash installer)	64-bit (exe installer)
32-bit	32-bit (bash installer)		32-bit (exe installer)

Install miniconda on your machine. Detailed instructions:

• Linux: http://conda.pydata.org/docs/install/quick.html#linux-miniconda-install
• Mac: http://conda.pydata.org/docs/install/quick.html#os-x-miniconda-install
• Windows: http://conda.pydata.org/docs/install/quick.html#windows-miniconda-install

2. Create and Activate the Environment

For Windows users, these following commands need to be executed from the Anaconda prompt as opposed to a Windows terminal window. For Mac/Linux, a normal terminal window will work.

1. Clone the repository, and navigate to the downloaded folder. This may take a minute or two to clone due to the included image data.

git clone https://github.com/ShayCichocki/GAN
cd GAN

2. Create (and activate) a new environment, named gan-playground with Python 3.6. If prompted to proceed with the install (Proceed [y]/n) type y.

   • Linux or Mac:

   conda create -f environment.yml
   source activate gan-playground
   

   • Windows:

   conda create -f environment.yml python=3.6
   activate gan-playground
   

3. Add your data and run the program

1. For this project to work correctly, the data needs to be structured with subfolders for example:

   -data-set
       |-train 	
           |-your images here
   

   You can configure the dataset location to be anywhere in the .env
2. After adding your relevant images to the train folder, copy .env.example into .env and fiddle with training params
3. Run python runner.py from your activated conda environment
4. This will print out each loss from the Discriminator and Generator
5. After the training is over your results will be displayed in a libplot window

TODOS

1. Add saving/loading of Discriminator/Generator
2. Cleanup code