This is an unofficial chainer re-implementation of a paper, Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks. This implementation is based on this
- Python 3.5+
- Chainer 2.0+
- ChainerCV 0.7+
- Numpy
- Matplotlib
Downloadable datasets are listed in ./datasets/download_cyclegan_dataset.sh
./datasets/download_cyclegan_dataset.sh <dataset>
python train.py --load_dataset <dataset> --gpu <gpu>
python single_image_test.py <input_image> --gpu <gpu> --load_gen_model <trained_generator> --output <output_image>
left: input (horse), right: output (zebra)
- [1]: JY. Zhu, et al. "Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks", in ICCV, 2017.
- [2]: Original implementation in pytorch
- [3]: Chainer-v1 implementation