摘要
针对现阶段卷积神经网络参数量较大,检测速度较慢,无法嵌入至移动端电子设备,且在复杂环境下检测精度较低的问题,设计了两层前后分离轻量级的卷积神经网络的人脸检测方法。第一层网络采用全卷积神经网络,用于快速提取人脸特征,并生成大量的人脸边界候选框。第二层网络采用深层全连接卷积神经网络,将第一层网络推断的人脸候选区域进行筛选,并输出人脸大小、坐标和置信度。实验表明,本文设计的人脸检测方法在人脸基准数据集FDDB上具备较高的检测精度和检测速度,轻量级的网络设计使得算法移植到前端电子设备成为了可能。
The current convolutional neural network has some disadvantages such as a large amount of parameters,a slow detection speed,low detection accuracy in a complex environment,and it cannot be embedded in mobile e lectronic devices.For improvement,this paper designs a face detection method with two layers of front-to-back separation and lightweight convolutional neural network.The first layer of the network uses a full convolutional neural network to quickly extract facial features and generate a large number of face boundary candidate frames.The second layer of the network uses a deep fully connected convolutional neural network to screen the candidate regons of the face inferred by the first layer of the network and output the face size,coordinates and confidence.The experiments show that the face detection method designed in this paper has higher detection accuracy and de tection speed on the Face Detection Data Set and Benchmark(FDDB),and the light"weight net"work design makes it possible to transplant the algorithm to front-end electronic devices.
作者
姚高华
于健海
卢振坤
Gao-hua YAO;Jian-hai YU;Zhen-kun LU(College of Electronic and Information Engineering,Wuzhou University,Wuzhou 543002,China;Guangxi Colleges and Universities Key Laboratory of Image Processing and Intelligent Information System,Wuzhoo University,Wuzhoo 543002,China;Colleg e of Information Science and Engineerinx,Guanxxi University for Nationalities,Nanninx 530000,China)
出处
《机床与液压》
北大核心
2020年第18期202-208,共7页
Machine Tool & Hydraulics
基金
国家自然科学基金项目(61562074)
广西自然科学基金项目(2018GXNSFAA294019)
广西高校中青年教师基础能力提升项目(2017KY0632)。
关键词
参数量
电子设备
复杂环境
全卷积网络
人脸边界候选框
轻量级
Parameter quantity
Electronic equipment
Complex environment
Full convolution network
Face boundary candidate box
Lightweight
