摘要
城市流动人口激增,给城市社区的治安和犯罪的治理带来一个新的挑战.为了更好在监控视频中识别目标嫌疑人员,利用深度学习的方法对目标进行判别.基于卷积神经网络的深度学习是主流方法,包含人脸检测和人脸辨识两个步骤.对图片采用卷积神经网络进行训练,得到模型参数,然后根据区域卷积神经网络在监控中判别出目标.具体采用卷积层、激活层、池化层、全连接层联合的深度学习方法,结果显示,该方法具有不错的效果,识别精度可以达到90%以上.
The surge of urban floating population has brought a new challenge to the public security of urban communities and the governance of crime.In order to identify the target suspects in the surveillance video,the deep learning method is used to discriminate the target.Deep learning based on convolutional neural network is the mainstream technology,including face detection and face recognition.The image is trained by convolution neural network,and the model parameters are obtained.Then the target is identified in the monitoring based on the regional convolution neural network.The convolution layer,activation layer,pooling layer and full connection layer are combined in depth learning method.The results show that the method has good effect and the recognition accuracy can reach more than 90%.
作者
王保平
Wang Baoping(College of Software,Nanyang Normal University,Nanyang 473061,China)
出处
《信息通信》
2019年第8期72-73,共2页
Information & Communications
基金
河南省科技攻关基于移动终端大规模图像采集的目标识别与预警关键技术研究及应用(182102310752)
关键词
人脸识别
卷积层
神经网络
激活函数
池化
Face recognition
Convolutional layer
Neural network
Activation function
pooling