基于OpenCV的人脸图像识别系统设计与实现  

Design and Implementation of Face Image Recognition System Based on OpenCV

作　　者：胡北辰[1] HU Beichen(Department of Information and Intelligent Engineering,Anhui Vocational College of Electronics and Information Technology,Bengbu Anhui 233000,China)

机构地区：[1]安徽电子信息职业技术学院信息与智能工程系,安徽蚌埠233000

出　　处：《佳木斯大学学报（自然科学版）》2022年第2期123-126,共4页Journal of Jiamusi University：Natural Science Edition

基　　金：安徽省高校优秀青年人才支持计划项目(gxyq2020141)。

摘　　要：基于轻量级的OpenCV软件库,提出了一种人脸表情图像识别系统。该系统结合了迁移学习策略和具有中心损失的联合监督方法,以优化人脸识别的过程。为了提高识别的准确性和速度,该系统采用MobileNet这一个轻量级的卷积神经网络模型,以实现脱机和实时框架中部署。为了验证该系统的有效性,分别使用两个常用的数据集对系统进行评估。实验结果显示,该系统能够提高人脸识别的准确度。Based on the lightweight OpenCV software library,this paper proposes a facial expression image recognition system.The system combines a transfer learning strategy and a joint supervision method with center loss to optimize the process of face recognition.In order to improve the accuracy and speed of recognition,the system uses MobileNet,a lightweight convolutional neural network model,to implement offline and real-time deployment in the framework.In order to verify the effectiveness of the system,two commonly used data sets were used to evaluate the system.Experimental results show that the system can improve the accuracy of face recognition.

关 键 词：人脸表情识别 卷积神经网络 OPENCV 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]