/FMEG

Facial Micro-Expression Generation based on Deep Motion Re-targeting and Transfer Learning

Primary LanguageJupyter NotebookOtherNOASSERTION

Facial Micro-Expression Generation based on Deep Motion Re-targeting and Transfer Learning

by Xinqi Fan, Ali Shahid, and Hong Yan

Introduction

This repository is for our work Facial Micro-Expression Generation based on Deep Motion Re-targeting and Transfer Learning.

Usage

Requirement

Python 3.6

PyTorch 1.6

Download

Clone the repository:

git clone https://github.com/xinqi-fan/FMEG
cd FMEG

Prepare data

  • Please download CASME, SMIC, and SAMM.
  • Modify the path to your dataset in the following config/camse-256.yaml, config/smic-256.yaml, config/samm-256.yaml.

Pretrained weights

  • Please find them here

Train the model

  • Please refer to train.sh and modify the related checkpoint path.

Test the model

  • Please refer to demo.sh and modify the related paths.
  • The pretrained weights can be found in the above mentioned links.

Citation

To be updated

Comment

We welcome any pull request for improving the code.

Acknowledge

More information can be found from the original first-order-motion repo.