预览加载中，请您耐心等待几秒...

基于轻量化神经网络的桥梁裂缝识别检测.docx

预览
1/3
2/3
3/3

在线预览结束，喜欢就下载吧，查找使用更方便

下载提示 文本预览

如果您无法下载资料，请参考说明：

1、部分资料下载需要金币，请确保您的账户上有足够的金币

2、已购买过的文档，再次下载不重复扣费

3、资料包下载后请先用软件解压，在使用对应软件打开

基于轻量化神经网络的桥梁裂缝识别检测 Abstract Bridgeisanimportanttransportationinfrastructuresthatconnectspeopleandresources.However,thedamageanddeteriorationofbridgeshappenfrequently,whichmayresultincatastrophicaccidents.Therefore,theearlydetectionofcracksinbridgescangreatlyimprovetheirsafety.Inthispaper,weproposealightweightneuralnetwork-basedbridgecrackdetectionmethodtoimprovetheefficiencyofbridgemaintenance. Introduction Bridgesplayavitalroleinmodernsocietybyconnectingpeopleandresources.However,duetotheharshnaturalenvironmentanddailyusage,bridgedamageanddeteriorationhappenfrequently.Amongthesedamages,bridgecracksareoneofthemostcommonanddangerousissuesthatmayleadtocatastrophicaccidents.Therefore,theearlydetectionofcracksinbridgescaneffectivelyimprovetheirsafetyandextendtheirservicelife. Traditionally,bridgeinspectionandmaintenanceweremainlybasedonmanualmethods,suchasvisualinspectionandhammertesting,whicharetime-consuming,labor-intensive,andmaymisssomesubtlecracks.Withthedevelopmentofartificialintelligenceandcomputervision,automaticbridgecrackdetectionmethodshavegraduallyattractedattentioninrecentyears. Inthispaper,weproposealightweightneuralnetwork-basedbridgecrackdetectionmethodtoimprovetheefficiencyofbridgemaintenance.Themaincontributionsofthispaperareasfollows: -Weproposealightweightneuralnetworkstructureforbridgecrackdetection,whichcanachievehighaccuracywithalownumberofparametersandfastcomputationalspeed. -Wecollectadatasetofbridgecrackimagesanduseittoevaluatetheeffectivenessoftheproposedmethod. -Weconductexperimentstocompareourmethodwithseveralstate-of-the-artmodelsanddemonstrateitssuperiorityinimage-basedbridgecrackdetection. RelatedWork Inrecentyears,manyresearchershaveproposedvariousmethodsforbridgecrackdetection,suchasimage-basedmethods,vibration-basedmethods,andacousticemission-basedmethods. Image-basedmethodshaveattractedthemostattentionduetotheirlowcostandhighaccessibility.Earlyimage-basedmethodsmainlyreliedontraditionalcomputervisionalgorithms,suchasedgedetection,thre