End-to-end facial expression recognition using Convnets

This is a coursework for Msc Project

Model

VGG16 accuracy 81%
link: https://drive.google.com/file/d/1eGXCBxhfKwqRTH0VYUfyAaJ-i7qnIcqk/view?usp=drive_link

VGG19 accuracy 79%
link: https://drive.google.com/file/d/17nR72s_Xg0fk3BDhnU6X1yw20pyZRU8T/view?usp=drive_link

VGG16 accuracy 77%
https://drive.google.com/file/d/15DHiQ_85_cdN9KT-aQc3uqnZuwiarwrA/view?usp=drive_link

VGG16 accuracy 74%
https://drive.google.com/file/d/1cl8LrTUj7iBGDmRN6pudWZSq6bEbipVm/view?usp=drive_link

Installation

1.)PyQt5

PyQt5 >= 5.15.9

2.)pytorch

Please install throught official website of pytorch
torch               1.13.1+cu116
torchaudio          0.13.1+cu116
torchvision         0.14.1+cu116

3.)dilib

dlib == 19.22.99

4.)Opencv

pip install opencv-python == 4.8.0.74

Training data

1. FER2013+

Please download FER2013+ via https://drive.google.com/file/d/1coaX4AfvFKmgZu26yMnX0Gh1KmQPXC8k/view?usp=drive_link

or https://www.kaggle.com/datasets/moqixiang/fer2013

2. RAF-DB

Please download RAF-DB via https://drive.google.com/file/d/1cKCqrj7jR0G39InvYLVjs5YhxD2sKZGf/view?usp=drive_link

Demo video

https://youtu.be/XM60aIuO0gY

Authors

Qixiang Mo

e-mail: moqixiang99927@gmail.com
        moqixiang99927@163.com