摘要
虽然将深度学习与人脸识别相结合,可以提高系统响应速度和人脸识别精度,但该算法的计算量、参数量极大,对硬件性能所提出要求十分严格,因此,始终未能得到推广。近几年,为控制计算成本,研究人员先后开发多种轻量化网络,文章以此为背景,首先说明了人脸识别的原理,其次介绍了基于LCNN的识别技术,包括MTCNN、Mobile Net等,最后围绕LCNN在人脸识别过程中的具体应用展开了讨论,内容涉及构建数据集、选择检测方法、处理数据等方面,供有关人员参考。
although the combination of deep learning and face recognition can improve the response speed and the accuracy of face recognition,the computational complexity and parameters of the algorithm are very large,and the requirements for hardware performance are very strict,as a result,it has not been popularized.In recent years,in order to control the cost ofcomputing.researchers have developed a variety of lightweight networks,the article as a background,first explained the principle of face recognition,and then introduced the recognition technology based on CNN,it includes MTCNN,Mobile Net and so on,At last,it discusses the concrete application of CNN in face recognition,which involves constructing data set,selecting detection method,processing data and so on.
作者
黎润潼
LI Runtong(Jiangxi University of Science and Technology,Nanchang 341099,Jiangxi)
出处
《长江信息通信》
2023年第11期38-40,共3页
Changjiang Information & Communications
关键词
人脸识别
卷积神经网络
轻量化网络
深度学习
face recognition
convolutional neural network
lightweight networks
deep learning