Facial Expression Recognition System Based on Improved Residual Network

摘要：

　　In order to achieve fast and accurate expression recognition of visitors,an expression recognition system based on an improved Residual Network(ResNet)model is designed.The access control system plays a vital role in maintaining the security of a closed place and creating a good environment.This paper designs an expression recognition access control system to avoid security risks more effectively.The system uses the improved ResNet,which can extract the important discriminable local facial expression features.This network model also uses a very efficient channel attention module,which can assign different weights to the extracted features map and locate the significant regions in the facial expression images that can discriminate the expression classification based on the size of the weights.In the Fer2013 dataset,experiments show that the model can effectively improve the expression recognition accuracy.

关键词： DeepLearning Facial Expression Recognition Resnet Attention Mechanism Fer2013 Dataset

作者: Gao zhiyu Mu jing

作者单位: School of Computer Science and Engineering,Xi'an Technological University,Shaanxi,Xi'an,710021

会议类型: 国内会议

会议名称: 第22届中国系统仿真技术及其应用学术年会(CCSSTA2021)

会议地点: 合肥

会议语种:英文

页码: 385-391

在线出版日期: 2021-07-01（万方平台首次上网日期，不代表论文的发表时间）