• 首页
  • 期刊简介
  • 编委会
  • 投稿指南
  • 收录情况
  • 杂志订阅
  • 联系我们
引用本文:刘思雨,薛劲松,景栋盛.基于分阶段深度神经网络的施工违章识别[J].软件工程,2020,23(9):32-35.【点击复制】
【打印本页】   【下载PDF全文】   【查看/发表评论】  【下载PDF阅读器】  
←前一篇|后一篇→ 过刊浏览
分享到: 微信 更多
基于分阶段深度神经网络的施工违章识别
刘思雨,薛劲松,景栋盛
(国网江苏省电力有限公司苏州供电分公司,江苏 苏州 215004)
sy2801@163.com; 6802569@qq.com; jds19810119@163.com
摘 要: 目前很多施工场地仍然使用人工方式检测施工人员是否佩戴安全帽。针对此,设计并实现了一个基于分 阶段深度神经网的施工违章识别系统,用以检测施工人员是否佩戴安全帽。系统利用深度神经网,通过在视频中采样获 得图片,然后将其分割成若干子区域,接着利用预处理后的数据训练模型,不断优化提升识别精度,然后将训练好的模 型应用到系统中。在室内、室外和红外线三个场景中进行测试。实验结果表明,系统具有良好的实时检测能力,总体平 均正确检出率达86.79%。
关键词: 违章识别;深度学习;神经网络;物体识别;视频监控
中图分类号: TP18    文献标识码: A
Construction Safety Violation Recognition based on Staged Deep Neural Network
LIU Siyu, XUE Jingsong, JING Dongsheng
(Suzhou Power Supply Branch, State Grid Jiangsu Electric Power Limited Company, Suzhou 215004, China )
sy2801@163.com; 6802569@qq.com; jds19810119@163.com
Abstract: In many building sites, people still manually monitor whether the construction crew wear safety helmets or not. To solve this problem, this paper designs a safety-violation identi cation system based on phased deep neural network in order to locate the construction crews who appear without safety helmets. By using deep neural network, this proposed system samples pictures from videos, divides them into sub-sections, and then uses processed data to train the model. When the identi cation accuracy is high enough, the model is then applied to practical use. The model is tested respectively in indoor scene, outdoor scene and infrared scene. The experiment results show that this system can achieve good real-time monitoring, with an average correct detection rate of 86.79%.
Keywords: violation recognition; deep learning; neural network; object recognition; video surveillance


版权所有:软件工程杂志社
地址:辽宁省沈阳市浑南区新秀街2号 邮政编码:110179
电话:0411-84767887 传真:0411-84835089 Email:semagazine@neusoft.edu.cn
备案号:辽ICP备17007376号-1
技术支持:北京勤云科技发展有限公司

用微信扫一扫

用微信扫一扫