File size: 1,591 Bytes
82b75c8 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 |
# FExGAN-Meta: Facial Expression Generation with Meta-Humans
![FExGAN-Meta GIF Demo](https://github.com/azadlab/FExGAN-Meta/blob/master/FExGAN-Meta.gif?raw=true)
This is the demo of the FExGAN-Meta proposed in the following article:
[FExGAN-Meta: Facial Expression Generation with Meta-Humans](https://www.arxiv.com)
FExGAN-Meta is the extension of [FExGAN](http://arxiv.org/abs/2201.09061). It takes input an image of Meta-Human and a vector of desired affect (e.g. angry,disgust,sad,surprise,joy,neutral and fear) and converts the input image to the desired emotion while keeping the identity of the original image.
![FExGAN-Meta GIF Demo](https://github.com/azadlab/FExGAN-Meta/blob/master/results.png?raw=true)
# Requirements
In order to run this you need following:
* Python >= 3.7
* Tensorflow >= 2.6
* CUDA enabled GPU with memory >=8GB (e.g. GTX1070/GTX1080)
# Usage
You can either run this on google colab or run it on your local system
* Install the pre-requisites
* Download the models (if any link fails in the notebook due to google drive restriction, try downloading them manually)
* Execute the rest of the notebook
# Citation
If you use any part of this code or use ideas mentioned in the paper, please cite the following article.
```
@article{Siddiqui_FExGAN-Meta_2022,
author = {{Siddiqui}, J. Rafid},
title = {{FExGAN-Meta: Facial Expression Generation with Meta-Humans}},
journal = {ArXiv e-prints},
archivePrefix = "arXiv",
keywords = {Deep Learning, GAN, Facial Expressions},
year = {2022}
url = {http://arxiv.org/abs/2201.09061},
}
```
|