TailorGAN

TailorGAN: Making User-Defined Fashion Designs.

• paper link: https://arxiv.org/abs/2001.06427

Prerequisites

• Linux system or MacOS
• Python 3.6, PyTorch version 1.1.0
• NVIDIA GPU + CUDA

Installation

• Install PyTorch and dependencies from http://pytorch.org

Test

• Default, using test.py file to synthesize the garment from the images in the example folder. Pre-trained checkpoints download needed and as the instruction in the checkpoints folder to download.

$ python test.py

• Sample results are presented in the example folder.

Train

• Training process is divided into two parts, the first part is reconstruction, using following command to train first step, and putting the checkpoints to corresponding folder to train second step.

$ python collarRecon.py --step step1

• Second step, using following command:

$ python collarSyn.py --step step2

Dataset

• The dataset is available online, please download it from the link: https://drive.google.com/open?id=11WMP0uCgzF--Kt18Dwh75Ja-nK1llMcx, and put it under the /data foler.

Citation

• If you think our work is useful, please use the following:

@misc{chen2020tailorgan,
    title={TailorGAN: Making User-Defined Fashion Designs},
    author={Lele Chen and Justin Tian and Guo Li and Cheng-Haw Wu and Erh-Kan King and Kuan-Ting Chen and Shao-Hang Hsieh and Chenliang Xu},
    year={2020},
    eprint={2001.06427},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}