A lightweight and efficient CNN based on VGG-16 variant for emotion recognition

Siqi Bai; Shuhao Hu; Xinyue Cui; Yongbo Wu

doi:10.1117/12.2656497

20 October 2022 A lightweight and efficient CNN based on VGG-16 variant for emotion recognition

Siqi Bai, Shuhao Hu, Xinyue Cui, Yongbo Wu

Proceedings Volume 12451, 5th International Conference on Computer Information Science and Application Technology (CISAT 2022); 124514Y (2022) https://doi.org/10.1117/12.2656497
Event: 5th International Conference on Computer Information Science and Application Technology (CISAT 2022), 2022, Chongqing, China

Abstract

Emotion recognition plays an important role in medicine, criminal investigation, human-computer interaction and other fields. Emotion recognition through machine learning has become a promising research direction. VGG network is a kind of classic convolutional neural networks (CNN) which can be used for emotion recognition, and VGG-16 is one of the classic architectures of VGG. However, VGG-16 has too many parameters and it seems that VGG-16 doesn't perform well on FER2013 because of overfitting. To solve these problems, this paper proposes a CNN network based on VGG-16 variant called VGG-LIGHT. This network inherits characters of small kernel size and similar architectures between different convolutional layers of VGG. Compared with traditional VGG architecture, VGG-LIGHT has fewer parameters. Compared with other lightweight network, VGG-LIGHT is more efficient and needs less training time. The paper also built a medical data set which is composed by faces of different patients and verified the proposed network by comparing its results with Deepface, a famous framework for face analysis. The experimental results show that the recognition results of VGG-LIGHT are in good agreement with Deepface.

Citation Download Citation

Siqi Bai, Shuhao Hu, Xinyue Cui, and Yongbo Wu "A lightweight and efficient CNN based on VGG-16 variant for emotion recognition", Proc. SPIE 12451, 5th International Conference on Computer Information Science and Application Technology (CISAT 2022), 124514Y (20 October 2022); https://doi.org/10.1117/12.2656497

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
5 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Facial recognition systems

Network architectures

Convolution

Data modeling

Detection and tracking algorithms

Information technology

Machine learning

Show All Keywords

Keywords/Phrases

Search In:

Publication Years