基于DenseNet的人脸图像情绪识别研究
Research on emotion recognition of face image based on densenet
投稿时间:2022-01-22  修订日期:2022-02-27
DOI:
中文关键词: 人脸情绪识别  稠密神经网络  空洞卷积  中心损失函数  深度学习优化器.
英文关键词: facial emotion recognition  densenet  atrous convolution  center loss function  optimizer.
基金项目:湖北省科技重大专项(2020AEA011);武汉市科技计划应用基础前沿项目(2020020601012267)
作者单位邮编
雷建云 中南民族大学 
马威 中南民族大学 
夏梦 中南民族大学 430000
郑禄 中南民族大学 
田望 中南民族大学 
摘要点击次数: 198
全文下载次数: 0
中文摘要:
      针对人脸情绪识别类内差异大,类间差异小的特点,结合学生人脸图像的线上课堂情绪识别的场景,提出多尺度空洞卷积模块提取不同空间尺度特征的稠密深度神经网络模型,实现自然场景下学生人脸图像识别. 该模型主要由多尺度空洞卷积和DenseNet神经网络两个子网络组成,其中多尺度空洞卷积由不同空洞率的四分支网络提取不同尺度特征,空洞卷积减小特征图尺寸,减少DenseNet内存资源占用;最后在DenseNet网络中结合Adam优化器和中心损失函数.使用稠密网络的旁路连接,加强情绪特征传递和复用.研究结果表明:基于稠密深度神经网络的情绪识别网络模型能够有效提高情绪分类的准确率,模型对预处理后的FER2013+数据集识别准确率达到93.99%,为线上教学反馈提供技术支持.
英文摘要:
      The characteristics of large intra-class differences and small inter-class differences in facial emotion recognition, combined with the scene of online classroom emotion recognition of student face images, a dense deep neural net-work model with multi-scale atrous convolution modules to extract features of different spatial scales is proposed. Realize student face image recognition in natural scenes. The model is mainly composed of two sub-networks: mul-ti-scale atrous convolution and DenseNet neural network. The multi-scale atrous convolution extracts features of different scales by four-branch networks with different atrous rates. Atrous convolution Reduce the size of the feature map and reduce the memory resource occupation of DenseNet; finally, the Adam optimizer and the central loss func-tion are combined in the DenseNet network. The bypass connection of the dense network is used to strengthen the transfer and reuse of emotional features. The research results show that: based on dense deep neural network The emotion recognition network model of the network can effectively improve the accuracy of emotion classification, and the recognition accuracy rate of the model for the preprocessed FER2013+ data set reaches 93.99%, which pro-vides technical support for online teaching feedback.
View Fulltext   查看/发表评论  下载PDF阅读器
关闭