k-eak / e-AttnGAN

e-AttnGAN (The original code (AttnGAN architecture) is taken from https://github.com/taoxugit/AttnGAN)

Pytorch implementation for e-AttnGAN

Dependencies

python 3.7

Pytorch

In addition, please add the project folder to PYTHONPATH and pip install the following packages:

• python-dateutil
• easydict
• pandas
• torchfile
• nltk
• scikit-image

Data

Instructions for Fashion Synthesis and FashionGen datasets will be added soon.

Training

• Pre-train DAMSM models:

  • For Fashion Synthesis dataset: python pretrain_DAMSM.py --cfg cfg/DAMSM/fashion_gen.yml --gpu 0

• Train AttnGAN models:

  • For Fashion Synthesis dataset: python main.py --cfg cfg/fashion_gen.yml --gpu 0

Citing e-AttnGAN

If you find e-AttnGAN useful in your research, please consider citing:

to-do

Citing AttnGAN

If you find AttnGAN useful in your research, please consider citing:

@article{Tao18attngan,
  author    = {Tao Xu, Pengchuan Zhang, Qiuyuan Huang, Han Zhang, Zhe Gan, Xiaolei Huang, Xiaodong He},
  title     = {AttnGAN: Fine-Grained Text to Image Generation with Attentional Generative Adversarial Networks},
  Year = {2018},
  booktitle = {{CVPR}}
}

Reference

• StackGAN++: Realistic Image Synthesis with Stacked Generative Adversarial Networks [code]
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks [code]