micro-expression-recognition-with-deep-learning's Introduction

MICRO-Expression-Recognition-with-deep-learning

Main stream of the model:

Related paper: Enriched Long-term Recurrent Convolutional Network for Facial Micro-Expression Recognition

Environment：

matlab 2015b

python 3.6

Keras (tensorflow as backend)

Files:

TIM-temporal-interpolation-model-master contains interpolation programs to preprocess data.

tvl1flow_3 is the package for computing optical flow and optical strain.

Recognition-Experiment contains model and training files.

How to run:

Prepare your data (e.g. CASMEII dataset http://fu.psych.ac.cn/CASME/casme2-en.php)
Detect and crop the faces with opencv or some other packages by yourself.
Run interpolation files to interpolate your video into 16 frames(can set by yourself).
Compute optical flow image and optical strain image.
When your data is ready, run train.py

Result:

I have trained the model for 1000 epochs and the accuracy is arround 70%. If you need the model, contact me. And I will upload it to cloud disk later.

Updates:

Trained on CASMEII: link：https://pan.baidu.com/s/1tRWcHmPRVvgpFz95kybheA key：1234

micro-expression-recognition-with-deep-learning's People

Contributors

Stargazers

Watchers

micro-expression-recognition-with-deep-learning's Issues

Hi, could you send me the model ? ( [email protected] ) PLEASE!!!

AS TITLE

前辈您好，请问能否获得一份训练后的模型供学习使用，我已发邮件给您的邮箱（[email protected]）。万分感谢！

万分感谢！

How to use the pre-trained model?

Dear author, it is my great honor to star your project. May I ask how to use your pre-trained model for micro-expression recognition?

python小白，想问问作者大大这个程序应该怎么修改之后才能运行呢？

我下了作者大大提供的训练之后的模型，并且在Recognition_Experiments文件夹下面新建了一个models文件夹，将两个训练之后的模型文件放了进去，但是运行之后没有任何反应。请问应该如何继续修改才能让程序跑起来呢？我是否需要修改raw_path，flowimg_path 和strainimg_path呢？
另外这个程序是否属于无监督学习呢？