基于改进型FaceBoxes的人头检测算法  

Head detection algorithm based on improved FaceBoxes

作　　者：张鹏举 严佩敏[1] 朱秋煜[1] ZHANG Peng-ju;YAN Pei-min;ZHU Qiu-yu(School of communication and information,Shanghai University,Shanghai 200444,China)

机构地区：[1]上海大学通信与信息学院,上海200444

出　　处：《微电子学与计算机》2021年第1期33-37,44,共6页Microelectronics & Computer

摘　　要：对于监控视频下的垂直视角,常常会出现人体遮挡,背景模糊等导致检测人体精度低的问题,本文提出以人头为目标代替人体进行检测的监控视频人数统计的方法.该方法对一种轻量型的网络FaceBoxes进行改进,并加入多尺度特征融合的方式,提升密集人头检测的精度.由于监控视频中,人头目标比较小,一般不会出现特大目标,因此使用k-means方法对候选框的尺度进行聚类,再进行训练,并使用交叠区域检测方法提高预测框的精度.实验结果表明,改进型的FaceBoxes方法相比之前的网络精度提高了约4%,其速度稍有降低.For the vertical perspective of surveillance video,human body occlusion and background blur often occur,which lead to low accuracy of human body detection.This paper proposes a surveillance video number statistics method that takes human head as the target to detect human body.This method improves a lightweight network of FaceBoxesand adds multi-scale feature fusion to improve the accuracy of dense head detection.In the surveillance video,the head target is relatively small,and generally there is no large target.Therefore,k-means method is used to cluster the scale of candidate frame,and then to train,and overlapping region detection method is used to improve the accuracy of prediction frame.The experimental results show that the improved FaceBoxesmethod is about 4%more accurate and slightly slower than the previous one.

关 键 词：FaceBoxes 人头检测 神经网络 特征融合 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]