摘要
行人检测分类是一种利用计算机视觉算法对图像或视频序列中的人物数量、年龄、性别以及衣着进行统计分类的技术。为了准确获取行人属性信息,降低行人漏检问题,经过分析现有数据集及行人所处的实际场景,提出基于深度学习的行人检测分类方案。首先使用行人检测算法检测行人身体属性,人脸识别算法识别人脸属性,然后结合身体属性与人脸属性对行人进行分类,最后总结了基于深度学习的行人检测分类以及发展前景。
Pedestrian detection classification is a technology that uses computer vision algorithms to statistically classify the number,age,gender,and clothing of people in an image or video sequence.In order to accurately obtain pedestrian attribute information and reduce the problem of pedestrian detection missing,a pedestrian detection classification scheme based on deep learning is proposed after analyzing the existing data set and the actual scenes of the pedestrian,First,the pedestrian detection algorithm is used to detect the body attributes of pedestrians and the face recognition algorithm is used to recognize the face attributes.Then the pedestrians are classified by according to the body attributes and face attributes.Finally,the classification and development prospects of pedestrian detection based on deep learning are summarized.
作者
李超
童林
刘永辉
彭登靖
LI Chao;TONG Lin;LIU Yong-hui;PENG Deng-jing(Department of mechanical and electrical engineering,GuiZhou Vocational Technology College of Electronics&Information,Kaili,Guizhou 556099;Liupanshui Normal University,Liupanshui,Guizhou 553004;Zhejiang University of Technology,Hangzhou,Zhengjiang 310014;Earthquake Administration of Yunnan Province,Kunming,Yunnan 650224,China)
出处
《集宁师范学院学报》
2021年第2期77-81,共5页
Journal of Jining Normal University
基金
贵州省智造技术省级协同创新中心(GZGC-2018-19)。
关键词
深度学习
目标检测
卷积网络
人脸识别
行人检测
deep learning
target detection
convolutional network
face recognition
pedestrian detection