jolibrain / pytorch-CycleGAN-and-pix2pix

Image-to-Image Translation in PyTorch

Prerequisites

• Linux or macOS
• Python 3
• CPU or NVIDIA GPU + CUDA CuDNN

Getting Started

Installation

• Clone this repo:

git clone https://github.com/jolibrain/pytorch-CycleGAN-and-pix2pix
cd pytorch-CycleGAN-and-pix2pix

• Install PyTorch and other dependencies (torchvision, visdom and dominate, FID).
  • For pip users, please type the command pip install -r requirements.txt.

JoliGAN train

• Options :

Model	Network	Decoder
CycleGAN, CycleGAN_semantic, CycleGAN_semantic_mask	resnet, Unet, mobile_resnet	Vanilla, Sty2

• Train a cycleGAN :

You can tune the hyperparameters in ./scripts/train_cyclegan.sh and then use the following line command.

bash ./scripts/train_cyclegan.sh

• Train a cycleGAN with labels :

You can tune the hyperparameters in ./scripts/train_cyclegan_semantic.sh and then use the following line command.

bash ./scripts/train_cyclegan_semantic.sh

• Train a cycleGAN with mask labels :

You can tune the hyperparameters in ./scripts/train_cyclegan_semantic_mask.sh and then use the following line command.

bash ./scripts/train_cyclegan_semantic_mask.sh

Datasets

• Unaligned : apple2orange, horse2zebra
• Unaligned with labels : svhn2mnist
• Unaligned with mask labels : glasses2noglasses,

Dataloader

To choose a dataloader please use the flag --dataset_mode dataloader_name. There are three dataloaders for different dataset architectures :

• Unaligned (unaligned)
• Unaligned with labels (unaligned_labeled)
• Unaligned with mask labels (unaligned_labeled_mask)

Acknowledgments

Our code is inspired by pytorch-CycleGAN-and-pix2pix.