[1]周 婕,马明栋.基于改进的 ResNet 网络的人脸表情识别[J].计算机技术与发展,2022,32(01):25-29.[doi:10. 3969 / j. issn. 1673-629X. 2022. 01. 005]
　ZHOU Jie,MA Ming-dong.Facial Expression Recognition System Based on Improved ResNet[J].,2022,32(01):25-29.[doi:10. 3969 / j. issn. 1673-629X. 2022. 01. 005]
点击复制

基于改进的 ResNet 网络的人脸表情识别()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
32
期数:
2022年01期
页码:
25-29
栏目:
人工智能
出版日期:
2022-01-10

文章信息/Info

Title:
Facial Expression Recognition System Based on Improved ResNet
文章编号:
1673-629X(2022)01-0025-05
作者:
周 婕1 ; 马明栋2
1. 南京邮电大学 通信与信息工程学院,江苏 南京 210003;
2. 南京邮电大学 地理与生物信息学院,江苏 南京 210003
Author(s):
ZHOU Jie1 ; MA Ming-dong2
1. School of Telecommunications & Information Engineering,Nanjing University ofPosts and Telecommunications,Nanjing 210003,China;
2. School of Geographical and Biological Information,Nanjing University of Posts andTelecommunications,Nanjing 210003,China
关键词:
表情识别; 深度残差网络; 深度学习; OpevCV; 人脸检测
Keywords:
expression recognition; ResNet; deep learning; OpenCV; face detection
分类号:
TP391
DOI:
10. 3969 / j. issn. 1673-629X. 2022. 01. 005
摘要:
近几年来,人工智能的热度一直居高不下,其中作为人机交互的一种重要方法冥人脸表情识别已经成为计算机视觉研究的热点。 从传统的机器学习算法到现在的深度学习,识别效率也在不断地提高,为了进一步提高人脸表情识别率,在传统的卷积神经网络的基础上,提出了一种基于改进的 ResNet 卷积神经网络的表情识别方法。 该方法基于 ResNet 网络的基本结构,采用的中间卷积部分是前后各一个卷积核为 1*1 的卷积层,中间是卷积核大小为 3*3 的卷积层,同时将下采样移到后面的 3*3 卷积层里面去做,减少信息的流失,并用 PReLU 替代 ReLU 激活函数。 与 ResNet 模型相比,改进的网络结构可以减少计算量,提高识别速度和识别率。 利用 Tensorflow 构建经过改进的 ResNet 卷积神经网络框架,并在增强的 Fer2013 数据集上进行了训练,得到了准确且高效的人脸表情识别模型,最后再结合 OpenCV 中的人脸检测分类器,从视频中抓取人脸进行识别,实现了实时识别人脸表情效果的输出。 实验结果表明,改进的 ResNet 卷积神经网络模型较其他的人脸表情识别方法在识别率上有了一定的提高。
Abstract:
In recent years,the popularity of artificial intelligence has remained high. Among them,as an important method of human -computer interaction,facial expression recognition,has become a hotspot in computer vision research. From traditional machine learningalgorithms to the current depth learning, recognition efficiency is constantly improving. In order to further improve the recognition rate offacial expressions, on the basis of convolutional neural network, an expression recognition method based on the improved ResNetconvolutional neural network is proposed. This method is based on the basic structure of the ResNet network. The middle convolutionpart is a 1*1 convolution layer before and after the convolution kernel,and the middle is a convolution layer with a convolution kernelsize of 3*3,and the downsampling is shifted do it in the following 3*3 convolution to reduce the loss of information and replace theReLU function with PReLU. Compared with the ResNet model,the improved network structure can reduce the amount of calculation andincrease the recognition speed and recognition rate. The Tensorflow is used to build an improved ResNet framework and trains it on theenhanced Fer2013 data set to obtain an accurate and efficient facial expression recognition model,and finally combines the face detectionclassifier in OpenCV to grab faces from the video for recognition,thus realizing the output of real-time recognition of facial expressions.The experimental results based on the data set show that the recognition rate of the improved ResNet convolutional neural network modelis indeed improved compared with other facial expression recognition methods.

相似文献/References:

[1]刘洋 季桂树 王刚.表情识别中基于轴对称的人眼定位方法[J].计算机技术与发展,2008,(12):47.
　LIU Yang,JI Gui-shu,WANG Gang.Eyes Location Method Based on Axial Symmetry Using for Expressions Recognition[J].,2008,(01):47.
[2]陈开云 高珏 孙柏林[] 许华虎[].基于混合特征提取的人脸情感识别研究[J].计算机技术与发展,2012,(02):96.
　CHEN Kai-yun,GAO Jue,SUN Bai-lin,et al.Studies on Facial Expression Recognition Based on Hybrid Features Extraction[J].,2012,(01):96.
[3]刘敦强,沈峘,夏瀚笙,等.一种基于深度残差网络的车型识别方法[J].计算机技术与发展,2018,28(05):42.[doi:10.3969/j.issn.1673－629X.2018.05.010]
　LIU Dun-qiang,SHEN Huan,XIA Han-sheng,et al.A Vehicle Model Ｒecognition Algorithm Based on Deep Residual Network[J].,2018,28(01):42.[doi:10.3969/j.issn.1673－629X.2018.05.010]
[4]艾玲梅,薛亚庆,李天东.深度残差网络在脉搏信号亚健康检测中的应用[J].计算机技术与发展,2020,30(07):109.[doi:10. 3969 / j. issn. 1673-629X. 2020. 07. 024]
　AI Ling-mei,XUE Ya-qing,LI Tian-dong.Application of Deep Residual Network in Pulse Signal Sub-health Detection[J].,2020,30(01):109.[doi:10. 3969 / j. issn. 1673-629X. 2020. 07. 024]
[5]姜丽莉,黄承宁.融合注意力机制改进残差网络的表情识别方法[J].计算机技术与发展,2022,32(05):42.[doi:10. 3969 / j. issn. 1673-629X. 2022. 05. 007]
　JIANG Li-li,HUANG Cheng-ning.An Expression Recognition Method Based on Fusion of Attention Mechanism and Improved Residual Network[J].,2022,32(01):42.[doi:10. 3969 / j. issn. 1673-629X. 2022. 05. 007]
[6]李大舟,张诗瑞,高 巍.基于深度残差网络的脊柱疾病分类预测[J].计算机技术与发展,2022,32(05):195.[doi:10. 3969 / j. issn. 1673-629X. 2022. 05. 033]
　LI Da-zhou,ZHANG Shi-rui,GAO Wei.Classification and Prediction of Spine Diseases Based onDeep Residual Network[J].,2022,32(01):195.[doi:10. 3969 / j. issn. 1673-629X. 2022. 05. 033]
[7]张文海,陈春玲.基于改进 GhostNet 模型的表情识别研究[J].计算机技术与发展,2022,32(08):60.[doi:10. 3969 / j. issn. 1673-629X. 2022. 08. 010]
　ZHANG Wen-hai,CHEN Chun-ling.Research on Expression Recognition Based on Modified GhostNet Model[J].,2022,32(01):60.[doi:10. 3969 / j. issn. 1673-629X. 2022. 08. 010]

更新日期/Last Update: 2022-01-10