Facial Expression Recognition with Multi-perspective Feature Fusion Under Deep Residual Convolution

Aiming at the problems of inaccurate facial expression recognition and large amount of calculation under multi-perspective in real life, a facial expression recognition model MVResNet-FER is proposed, which is based on multi-perspective feature fusion under deep residual convolution.The residual...

Full description

Saved in:
Bibliographic Details
Main Authors: GUAN Xiaorui, GAO Lu, SONG Wenbo, LIN Kezheng
Format: Article
Language:zho
Published: Harbin University of Science and Technology Publications 2023-04-01
Series:Journal of Harbin University of Science and Technology
Subjects:
Online Access:https://hlgxb.hrbust.edu.cn/#/digest?ArticleID=2203
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849716275454935040
author GUAN Xiaorui
GAO Lu
SONG Wenbo
LIN Kezheng
author_facet GUAN Xiaorui
GAO Lu
SONG Wenbo
LIN Kezheng
author_sort GUAN Xiaorui
collection DOAJ
description Aiming at the problems of inaccurate facial expression recognition and large amount of calculation under multi-perspective in real life, a facial expression recognition model MVResNet-FER is proposed, which is based on multi-perspective feature fusion under deep residual convolution.The residual block in ResNet is first improved and the conventional convolutional network is replaced with a depthwise separable network.Second, a CBAM module is added to enhance the extraction of effective features under multi-perspective and the supplementation of shallow feature information. Then use the RReLu activation function to replace the original ReLu to avoid deactivation of some nodes when the gradient is large.Finally, the global average pooling layer is used instead of the fully connected layer to achieve dimensionality reduction, and the generated feature vector is sent to Softmax for classification Experiments show that the proposed method produces excellent results on the CK+ and RaFD datasets, which can effectively improve the accuracy of facial expression recognition.
format Article
id doaj-art-074b43a478624a22a61b2a3528defd8a
institution DOAJ
issn 1007-2683
language zho
publishDate 2023-04-01
publisher Harbin University of Science and Technology Publications
record_format Article
series Journal of Harbin University of Science and Technology
spelling doaj-art-074b43a478624a22a61b2a3528defd8a2025-08-20T03:13:04ZzhoHarbin University of Science and Technology PublicationsJournal of Harbin University of Science and Technology1007-26832023-04-01280211712710.15938/j.jhust.2023.02.014Facial Expression Recognition with Multi-perspective Feature Fusion Under Deep Residual ConvolutionGUAN Xiaorui0GAO Lu1SONG Wenbo2LIN Kezheng3School of Computer Science and Technology, Harbin University of Science and Technology, Harbin 150080, ChinaHarbin Institute of Information Technology, 150431, ChinaSchool of Computer Science and Technology, Harbin University of Science and Technology, Harbin 150080, ChinaSchool of Computer Science and Technology, Harbin University of Science and Technology, Harbin 150080, China Aiming at the problems of inaccurate facial expression recognition and large amount of calculation under multi-perspective in real life, a facial expression recognition model MVResNet-FER is proposed, which is based on multi-perspective feature fusion under deep residual convolution.The residual block in ResNet is first improved and the conventional convolutional network is replaced with a depthwise separable network.Second, a CBAM module is added to enhance the extraction of effective features under multi-perspective and the supplementation of shallow feature information. Then use the RReLu activation function to replace the original ReLu to avoid deactivation of some nodes when the gradient is large.Finally, the global average pooling layer is used instead of the fully connected layer to achieve dimensionality reduction, and the generated feature vector is sent to Softmax for classification Experiments show that the proposed method produces excellent results on the CK+ and RaFD datasets, which can effectively improve the accuracy of facial expression recognition.https://hlgxb.hrbust.edu.cn/#/digest?ArticleID=2203multi-perspectivefeature fusiondepth separable convolutionresidual modelexpression recognition
spellingShingle GUAN Xiaorui
GAO Lu
SONG Wenbo
LIN Kezheng
Facial Expression Recognition with Multi-perspective Feature Fusion Under Deep Residual Convolution
Journal of Harbin University of Science and Technology
multi-perspective
feature fusion
depth separable convolution
residual model
expression recognition
title Facial Expression Recognition with Multi-perspective Feature Fusion Under Deep Residual Convolution
title_full Facial Expression Recognition with Multi-perspective Feature Fusion Under Deep Residual Convolution
title_fullStr Facial Expression Recognition with Multi-perspective Feature Fusion Under Deep Residual Convolution
title_full_unstemmed Facial Expression Recognition with Multi-perspective Feature Fusion Under Deep Residual Convolution
title_short Facial Expression Recognition with Multi-perspective Feature Fusion Under Deep Residual Convolution
title_sort facial expression recognition with multi perspective feature fusion under deep residual convolution
topic multi-perspective
feature fusion
depth separable convolution
residual model
expression recognition
url https://hlgxb.hrbust.edu.cn/#/digest?ArticleID=2203
work_keys_str_mv AT guanxiaorui facialexpressionrecognitionwithmultiperspectivefeaturefusionunderdeepresidualconvolution
AT gaolu facialexpressionrecognitionwithmultiperspectivefeaturefusionunderdeepresidualconvolution
AT songwenbo facialexpressionrecognitionwithmultiperspectivefeaturefusionunderdeepresidualconvolution
AT linkezheng facialexpressionrecognitionwithmultiperspectivefeaturefusionunderdeepresidualconvolution