摘 要:人流密度监控对于场所秩序以及在场公民的安全保障有十分重要的意义。基于 TinyYOLOv2 搭建人脸检测平台,通过 K210 系统级芯片实现了轻量化的人员统计与人流密度监控装置。静态数据库模拟和动态实景测试的结果表明,该装置可实现多目标检测,并在实景条件下达到96.5%的人员数量统计准确率,可在保证人流监控精度的前提下有效降低疫情交叉感染的风险。
关键词:人工智能;深度学习;神经网络;目标检测;人流监控
DOI:10.19850/j.cnki.2096-4706.2022.15.022
基金项目:天津市教委科研计划项目(JWK1605)
中图分类号:TP18 文献标识码:A 文章编号:2096-4706(2022)15-0081-03
Design and Implementation of Lightweight Pedestrian Flow Monitoring Device Based on TinyYOLOv2
LU Yuchen, MEI Jianqiang
(School of Electronic Engineering, Tianjin University of Technology and Education, Tianjin 300222, China)
Abstract: The monitoring of crowd density is of great significance to the order of the place and the safety of the citizens present. A face detection platform is built based on TinyYOLOv2, and a lightweight personnel statistics and crowd density monitoring device is realized through K210 system level chip. The results of static database simulation and dynamic real scene test show that the device can realize multi-target detection, and achieve a statistical accuracy of 96.5% in the number of people under real scene conditions, which can effectively reduce the risk of cross infection of the epidemic on the premise of ensuring the accuracy of pedestrian flow monitoring.
Keywords: artificial intelligence; deep learning; neural network; object detection; pedestrian flow monitoring
参考文献:
[1] 王毅勇,王珂 . 铁路车站人数统计系统的设计与实现 [J].铁路通信信号工程技术,2014,11(3):34-38.
[2] 程蕊,肖文涛 . 现代城市大型群体活动密集人群安全管理研究——基于福州市“欢乐闹元宵”活动的经验启示 [J]. 中国应急管理科学,2020(5):70-80.
[3] LI L,OUYANG W,WANG X G,et al. Deep learning for generic object detection: a survey [J].International Journal of Computer Vision,2020,128(2):261-318.
[4] DALAL N,TRIGGS B. Histograms of oriented gradients for human detection [C]//2005 IEEE computer society conference on computer vision and pattern recognition (CVPR'05).IEEE,2005:886-893.
[5] 许德刚,王露,李凡 . 深度学习的典型目标检测算法研究综述 [J]. 计算机工程与应用,2021,57(8):10-25.
[6] REDMON J,DIVVALA S,GIRSHICK R. You Only Look Once: Unified,Real-Time Object Detection [J/OL].(2015-06-08). https://arxiv.org/abs/1506.02640.
[7] REDMON J,FARHADI A. YOLO9000: Better,Faster, Stronger [C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Honolulu:IEEE,2017:6517-6525.
[8] CASIA.CASIA-FaceV5 [EB/OL].[2022-06-09].http:// biometrics.idealtest.org/.
作者简介:逯雨辰(2000.02—),男,汉族,福建人,学士,研究方向:嵌入式系统设计;梅健强(1981.12—),男,汉族,天津人,讲师,博士,研究方向:人工智能与数字图像处理。