No Description

Paul Cuciureanu c81906c61e init dvc 1 month ago
.dvc c81906c61e init dvc 1 month ago
datasets d7ee5b8611 added experimental results 2 years ago
0.png 9a74a3a0fa add some images 2 years ago
1.PNG 9a74a3a0fa add some images 2 years ago
5.PNG 9a74a3a0fa add some images 2 years ago
6.PNG 9a74a3a0fa add some images 2 years ago
LICENSE 835faa1559 Initial commit 2 years ago
README.md 4ae5faf981 modified maps link 1 year ago
ch2oi.png d7ee5b8611 added experimental results 2 years ago
da2ni.png d7ee5b8611 added experimental results 2 years ago
la2ph.png d7ee5b8611 added experimental results 2 years ago
main.py 3139ecf4c8 add more results 2 years ago
model.py d7ee5b8611 added experimental results 2 years ago
oi2ch.png d7ee5b8611 added experimental results 2 years ago
ops.py 3139ecf4c8 add more results 2 years ago
ph2la.png d7ee5b8611 added experimental results 2 years ago
ph2sk.png d7ee5b8611 added experimental results 2 years ago
sk2ph.png d7ee5b8611 added experimental results 2 years ago
train.sh 3139ecf4c8 add more results 2 years ago
utils.py 3139ecf4c8 add more results 2 years ago

Data Pipeline

Legend
DVC Managed File
Git Managed File
Metric
Stage File
External File

README.md

# ICCV paper of DualGAN DualGAN: unsupervised dual learning for image-to-image translation

please cite the paper, if the codes has been used for your research.

architecture of DualGAN

architecture

How to setup

Prerequisites

  • Python (2.7 or later)

  • numpy

  • scipy

  • NVIDIA GPU + CUDA 8.0 + CuDNN v5.1

  • TensorFlow 1.0 or later

  • unzip

Getting Started

steps

  • clone this repo:
git clone https://github.com/duxingren14/DualGAN.git

cd DualGAN
  • download datasets (e.g., sketch-photo), run:
bash ./datasets/download_dataset.sh sketch-photo
  • train the model:
python main.py --phase train --dataset_name sketch-photo --image_size 256 --epoch 45 --lambda_A 20.0 --lambda_B 20.0 --A_channels 1 --B_channels 1
  • test the model:
python main.py --phase test --dataset_name sketch-photo --image_size 256 --epoch 45 --lambda_A 20.0 --lambda_B 20.0 --A_channels 1 --B_channels 1

optional

Similarly, run experiments on facades dataset with the following commands:

bash ./datasets/download_dataset.sh facades

python main.py --phase train --dataset_name facades --image_size 256 --epoch 45 --lambda_A 20.0 --lambda_B 20.0 --A_channels 3 --B_channels 3

python main.py --phase test --dataset_name facades --image_size 256 --epoch 45 --lambda_A 20.0 --lambda_B 20.0 --A_channels 3 --B_channels 3

datasets

some datasets can also be downloaded manually from the website. Please cite their papers if you use the data.

facades: http://cmp.felk.cvut.cz/~tylecr1/facade/

sketch: http://mmlab.ie.cuhk.edu.hk/archive/cufsf/

maps: http://www.cs.mun.ca/~yz7241/dataset/maps.zip

oil-chinese: http://www.cs.mun.ca/~yz7241/, jump to http://www.cs.mun.ca/~yz7241/dataset/

day-night: http://www.cs.mun.ca/~yz7241/dataset/

Experimental results:

day2night da2ni la2ph ph2la sk2ph ph2sk ch2oi oi2ch

Acknowledgments

Codes are built on the top of pix2pix-tensorflow and DCGAN-tensorflow. Thanks for their precedent contributions!