Age Progression and Regression with Spatial Attention Modules

This repository contains the inference code for paper Age Progression and Regression with Spatial Attention Modules presented in AAAI2020.

Environment requirements

• Python 2.7.5
• Pytorch 0.3.1
• CUDA 8.0

Usage

1. Put your test images under the folder ./datasets/<name_of_dataset>. Then write a text file with each line containing <name_of_image> and <index_of_age_group>, seperated by a space character. Example input images and list files are provided.
2. Please download pre-trained models (two generator networks) and place them under ./checkpoint.
3. Modify test.sh as it fits your environment.
4. run test.sh, and the generation results will be save into ./results. You could modify the corresponding code in age_cycle_gan_model.py to also generate attention maps.

Citation

@inproceedings{Li2020AgePA,
title={Age Progression and Regression with Spatial Attention Modules},
author={Qi Li and Yunfan Liu and Zhenan Sun},
booktitle={AAAI},
year={2020}
}

Acknowledgement

The framework of this project has borrowed from CycleGAN and GANimation. We appreaciate their great work!

Note

This repo is still under construction, please use with care.