This is the implementation of Qingxing Cao et al.'s CVPR-17 work Attention-Aware Face Hallucination via Deep Reinforcement Learning.

Citation

Please cite Attention-FH in your publications if it helps your research:

@article{Cao2017Attention,
  title={Attention-Aware Face Hallucination via Deep Reinforcement Learning},
  author={Cao, Qingxing and Lin, Liang and Shi, Yukai and Liang, Xiaodan and Li, Guanbin},
  pages={1656-1664},
  year={2017},
}

Prerequisites

• Computer with Linux or OSX
• Torch-7
• For training, an NVIDIA GPU is strongly recommended for speed. CPU is supported but training is very slow.
• Cuda 8.0

Installing dependencies

• Install Torch: http://torch.ch/docs/getting-started.html
• Install Element-Research/rnn: https://github.com/Element-Research/rnn

Train

• After all dependencies installed, you should put your own training and testing files as:

../lfw_funneled_dev_128/train/Aaron_Eckhart/Aaron_Eckhart_0001.jpg
../lfw_funneled_dev_128/test/Aaron_Guiel/Aaron_Guiel_0001.jpg

Our data has processed with 2 points aligned and crop the centric part. We suggest you to use CFSS or other face alignment methods to pre-process your data. Then you can train the model by using the following command:

python RNN_main.lua

Feedback

If you have any questions or suggestions of this work, please feel free to contact the authors by sending email to caoqxATmail2.sysu.edu.cn or shiyk3ATmail2.sysu.edu.cn.

Acknowledgement

This work was benifited a lot from DCGAN and Recurrent Models of Visual Attention, thank for their great job.