A Robust Approach for Smile Recognition via Deep Convolutional Neural Networks

Smile recognition is a difficult research issue in the fields of computer vision and pattern recognition. Most of existing algorithms are only suitable for western people's smile recognition in simple backgrounds, and cannot well recognize Chinese people's smile in complex backgrounds. In...

Full description

Saved in:
Bibliographic Details
Published in2022 7th International Conference on Image, Vision and Computing (ICIVC) pp. 60 - 64
Main Authors Liu, Yuanzhu, Liu, Zuoli, Zhao, Yong, Xu, Junli
Format Conference Proceeding
LanguageEnglish
Published IEEE 26.07.2022
Subjects
complex background
Computer vision
convolution neural network
Convolutional neural networks
dataset construction
Face recognition
Image recognition
lightweight network
Neural networks
residual network
Residual neural networks
smile recognition
Online AccessGet full text

Cover

More Information
Summary:Smile recognition is a difficult research issue in the fields of computer vision and pattern recognition. Most of existing algorithms are only suitable for western people's smile recognition in simple backgrounds, and cannot well recognize Chinese people's smile in complex backgrounds. In order to solve this problem, we first construct a dataset composed of 4,000 western face images and 4,000 Chinese face images. Especially, 5,000 images in this dataset have complex backgrounds. Then, we use this dataset to train a convolutional neural network, a residual neural network, and a lightweight neural network for smile recognition, respectively. Various experiments show that our algorithm has a good generalization ability to recognize the smile of both western people and Chinese people robustly even in complex backgrounds.
DOI:10.1109/ICIVC55077.2022.9886093