Pytorch implementation for Paper MirrorGAN: Learning Text-to-image Generation by Redescription by Tingting Qiao, Jing Zhang, Duanqing Xu, Dacheng Tao. (The work was performed when Tingting Qiao was a visiting student at UBTECH Sydney AI Centre in the School of Computer Science, FEIT, the University of Sydney).
-
Install PyTorch and dependencies from http://pytorch.org
-
Install Torch vision from the source.
-
Clone this repo:
git clone https://github.com/qiaott/MirrorGAN.git
cd MirrorGAN
-
Download our preprocessed data from here.
-
The STEM was pretrained using the code provided here
-
The STREAM was pretrained using the code provided here.
After obtaining the pretrained STEM and STREAM modules, we can train the text2image model.
- Train a model:
./do_train.sh
- Test a model:
./do_test.sh
If you use this code for your research, please cite our paper.
@article{qiao2019mirrorgan,
title={MirrorGAN: Learning Text-to-image Generation by Redescription},
author={Qiao, Tingting and Zhang, Jing and Xu, Duanqing and Tao, Dacheng},
journal={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
year={2019}
}