johncai117 / generative-adversarial-network

Having fun with some GANs (:

Generative Adversarial Network

Having fun with some GANs (:

Also, practicing some TensorFlow - much more used to using PyTorch for my daily deep learning programming, so want to get refreshed with a different framework.

Set up

Download the kaggle dataset using

kaggle competitions download -c gan-getting-started

Attempt 1

Simply run main.py - this is based on a popular kaggle notebook from this link: https://www.kaggle.com/amyjang/monet-cyclegan-tutorial

python3 main.py

This notebook uses the CycleGAN - read more here: https://arxiv.org/pdf/1703.10593.pdf

The images produced are pretty cool! (They are taken from real natural images and made to look like Monet paintings)

For e.g.

Stay Tuned for more improvements later on!