基于深度学习的高空作业安全监测方法

打开文本图片集

中图分类号：TP391.4 文献标志码：A 文章编号：2095-2945（2025）34-0134-04

DOI：10.19981/j.CN23-1581/G3.2025.34.032

Abstract：Withtherapiddevelopmentoftheconstructionindustryacidentsof workingatheight haveshownanincreasing trend.Toadressteisufcientdyaicsandintellgenceoftraditionalsafetymonitringmethds，amonitoringmodelfororkingatheightsafetybasedondeeplearningwasconstructed.ThemethodimprovedtheYOLOv4algorithmnetwork structureand combinedwithtransferlearningprinciplestotestandanalyzedetectionacuracyindiferentworkingscenarios.ExperimentalesultsshowthatintheCPUoperatingenvironment，thedetectionspeedincreasedby2.7timescomparedtotraditionalmethods;in theGPUoperatingenvironment，thesingle-framevideodetectionspeedsforsingletarget，multipletargets，andsmalltargetswere 25.34，25.49，and 26.59 ms respectively，achieving average accuracies of 91.57% ， 89.69% ，and 86.63% .Theresearch findings provide new insights for preventing accidents in working at height.

Keywords： deep learning; working at height; safety monitoring;intelligent recognition; YOLOv4

高空作业是建筑施工领域的高风险作业类型之一，据统计数据显示，高处坠落事故占建筑业安全事故总数的 35% 以上。（剩余5452字）