blavad / CycleGAN

Tensorflow implementation of CycleGAN to learn transfering images from domain A to domain B.

CycleGAN

A quick implementation of CycleGAN using tensorflow 1. The implementation is similar to the original paper.

CycleGAN train/test

• Download repository and install dependancies

git clone https://github.com/blavad/CycleGAN.git
pip intall -e CycleGAN
cd CycleGAN

• Train a model from scratch (with standard settings)

python main.py -dA /path/to/datasetA -dB /path/to/datasetB

• Train a model from pretrain model (with standard settings)

python main.py -dA /path/to/datasetA -dB /path/to/datasetB --restore path/to/model

• Test a model

python main.py -dA /path/to/datasetA -dB /path/to/datasetB --restore path/to/model --testing