DCGAN Face Generation

A DCGAN is trained on a dataset of faces. A generator network generates new images of faces that look very realistic. The DCGAN utilizes 64 convolutions for both the discriminator and generator starting point.

Installation

1. Clone this repository into directory under running Jupyter notebook instance:

   git clone git@github.com:rigganni/DCGAN-Face-Generation.git

2. Ensure the necessary Python environment is set up. See the Anaconda environment file conda.yml in this repository. The environment can be created by the following:

   conda env create -f conda.yml

Files

• dcgan_face_generation.ipynb: Jupyter notebook containing all code to create DCGAN
• problem_unittests.py: Unit tests to check if DCGAN set up correctly
• train_samples.pkl: Training samples to visualize once training completes
• conda.yml: Anaconda enivronment file to reproduce development environment

Usage

Run the Jupyter notebook dcgan_face_generation.ipynb.

Future Improvements

• Run additional epochs to reduce blurriness
• Stop and/orstore the model then the Generator is at a lower loss than the best previous epoch loss
• Obtain a more diverse dataset that is more representative of all human faces