摘要
随着科技的发展人脸识别技术得到了巨大的应用,实现人脸识别的方法也越来越多,本文先简单对比了MLP、RNN、CNN这三个神经网络,然后再对CNN的基础结构进行了一个较为详细的介绍,主要通过对LeNet-5卷积神经网络模型结构的分析来了解卷积神经网络,然后设计了一款针对Olivetti Faces人脸数据库的卷积神经网络模型,通过更改卷积层中卷积核个数以及学习速率来进行一系列实验,最终确定在本次实验当中,当学习速率为0.05时,第一层卷积层卷积核数目为20,第二层卷积层数目为40的时候,能够得到一个针对Olivetti Faces人脸数据库有着较高识别率的一个新的卷积神经网络模型。
Face recognition technology has been widely used along with the development of technology,there are more and more methods to realize face recognition.First,this paper simply compares MLP,RNN and CNN,and then gives a detailed introduction to CNN’s basic structure,after analyzing the structure of the convolutional neural network model,convolutional neural network designed a convolutional neural network model for Olivetti Faces database,by changing the number of convolution cores and the learning rate,a series of experiments were carried out.In the experiment,when the learning rate was 0.05,the number of convolution cores in the first layer was 20,the second level,with 40 convolution layers,yields a new convolutional neural network model with a high recognition rate for Olivetti Faces.
作者
杨玉涟
官钰翔
沈毅
陈豪
朱霞
Yang Yulian;Guan Yuxiang;Shen Yi;Chen Hao;Zhu Xia(Jinling College of Science and technology,College of Network and Communication Engineering,Nanjing Jiangsu,211169)
出处
《电子测试》
2020年第21期60-61,99,共3页
Electronic Test
关键词
卷积神经网络
人脸识别
深度学习
Convolutional Neural Networks
face recognition
Deep learning
