A Tensorflow implementation of a Conditional GAN (Generative Adversarial Network) for generating human faces from a text description. This project was created for my CSC 340 (Artificial Intelligence) final project.

• Does not produce samples accurate to condition yet

The label is a vector with 5 indices each corresponding to a different facial feature

The values for each feature can either be 1 or -1

• Black hair
• Blonde hair
• Brown hair
• Male
• Beard

Git clone the repository and cd into the directory

git clone https://github.com/greerviau/Sketch-Artist.git && cd Sketch-Artist

Download the CelebA dataset here and extract

In CGAN.py add data directory to CelebA object

• Make sure that directory contains list_attr_celeba.csv and img_align_celeba

celebA = CelebA(output_size, channel, sample_size, batch_size, crop, data_dir=<path-to-data>)

python CGAN.py train
python CGAN.py test