Facial Expression Algorithms Using Deep Learning Models

Konferenz: CAIBDA 2022 - 2nd International Conference on Artificial Intelligence, Big Data and Algorithms
17.06.2022 - 19.06.2022 in Nanjing, China

Tagungsband: CAIBDA 2022

Seiten: 5Sprache: EnglischTyp: PDF

Autoren:
Xu, Yijie (SWJTU-Leeds Joint School, Southwest Jiaotong University, SWJTU, Chengdu, China)

Inhalt:
The facial expression detection problem is worth to research on solving many related problems. such as human-machine interface to include emotion processing. The efficiency and accuracy of the detection is decisive. After analysis of the detailed structure of three methods including AlexNet, VGGNet and ResNet, the article compares three methods on the facial expression detection about their speed of convergence, test accuracy and time. The result shows that ResNet has lower training loss with more accurate test results while it costs more time. The comparison between different methods can give us more reference on improvement in efficiency in facial expression detection and on more applications. We will try to combine the advantages for different methods.