CycleGAN is one of the most interesting works I have read. Although the idea behind cycleGAN looks quite intuitive after you read the paper: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks, the official PyTorch implementation of cycleGAN is difficult to understand(as the code has multiple things implemented together). I had to write cycleGAN to use it for some other work. So, I thought of making my version of cycleGAN public for those who are looking for an easier implementation of the paper.
- The code has been written in Python (3.5.2) and PyTorch (0.4.1)
- To download datasets (eg. horse2zebra)
$ sh ./download_dataset.sh horse2zebra
- To run training
$ python main.py --training True
- To run testing
$ python main.py --testing True
Coming Soon