pix2pix-pytorch

PyTorch implementation of Image-to-Image Translation Using Conditional Adversarial Networks.

Based on pix2pix by Isola et al.

The examples from the paper:

Prerequisites

• Linux
• Python with numpy
• NVIDIA GPU + CUDA 8.0 + CuDNNv5.1
• pytorch
• torchvision

Getting Started

• Clone this repo:

  git clone [email protected]:mrzhu-cool/pix2pix-pytorch.git cd pix2pix-pytorch

• Get dataset

  unzip dataset/facades.zip

• Train the model:

  python train.py --dataset facades --nEpochs 200 --cuda

• Test the model:

  python test.py --dataset facades --model checkpoint/facades/netG_model_epoch_200.pth --cuda

Acknowledgments

This code is a concise implementation of pytorch-CycleGAN-and-pix2pix. Much easier to understand.

Highly recommend the more completed and organized code pytorch-CycleGAN-and-pix2pix by original author junyanz.