dennis199441 / gans

Pokemon generation using Generative Adversarial Network

This is the assignment 4 for CMSC5707 - Advance Topics in Artificial Intelligence.

Dataset

1. MNIST from Tensorflow
2. Pokedex from https://github.com/PokeAPI/sprites/tree/master/sprites/pokemon

Model

Deep Convolutional Generative Adversarial Network

How to run

1. Install dependencies

pip3 install -r requirements.txt

2. Train GAN for MNIST

# 100 epochs, generate 16 images at once, batch size = 256 
python3 mnistGAN.py 100 16 256

3. Train GAN for Pokedex

git clone https://github.com/PokeAPI/sprites.git

cp ./sprites/pokemon .

python3 preprocess.py

# 3000 epochs, generate 16 images at once, batch size = 8, noise dimension = 128 
python3 pokeGAN.py 3000 16 8 128