预览加载中,请您耐心等待几秒...
1/2
2/2

在线预览结束,喜欢就下载吧,查找使用更方便

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

基于稀疏主题模型的交通监控视频异常检测(英文) Title:AnomalyDetectioninTrafficSurveillanceVideosBasedonSparseTopicModels Abstract: Withtheincreasingnumberofvehiclesonroads,ensuringthesmoothandsafeflowoftraffichasbecomeacriticalconcernforurbanmanagement.Trafficsurveillancesystems,equippedwithvideocameras,haveproventobeeffectivetoolsformonitoringandmanagingtraffic.However,manuallymonitoringthevastamountofvideodatageneratedbythesesystemsislabor-intensiveandpronetoerrors.Thispaperproposesananomalydetectionapproachintrafficsurveillancevideosbasedonsparsetopicmodels,aimedatautomaticallyidentifyingandanalyzingabnormaleventsinreal-time. 1.Introduction Trafficsurveillancevideoscaptureawealthofinformationaboutthemovementpatternsofvehicles,pedestrians,andobjectsontheroad.Thesevideosprovidevaluabledataforunderstandingtrafficflowandidentifyingpotentialissues,suchasaccidents,congestion,andillegalactivities.However,analyzingthisdatamanuallyistime-consumingandunreliable.Toaddressthischallenge,automatedanomalydetectionmethodsusingcomputervisiontechniqueshavegainedsignificantinterestinrecentyears. 2.RelatedWork Thissectionprovidesanoverviewofexistingapproachesforanomalydetectioninsurveillancevideos.Varioustechniques,includingmotion-basedmethods,appearance-basedmethods,anddeeplearning-basedmethods,havebeenproposed.However,thesemethodsoftensufferfromhighcomputationalcosts,limitedscalability,anddifficultiesinmodelingcomplexscenes. 3.SparseTopicModels Sparsetopicmodelshaveproventobeeffectiveindiscoveringlatentsemanticstructuresintextandimages.Inthispaper,weproposetheapplicationofsparsetopicmodelstotrafficsurveillancevideos.Bymodelingthevideoasacollectionofvisualwords,thetopicmodelscancapturetheunderlyingsemanticallymeaningfulpatternsandanomaliesinthevideodata. 4.ProposedApproach Ourproposedapproachconsistsofthefollowingsteps: 4.1Preprocessing:Thetrafficsurveillancevideosaresegmentedintoindividualframes,andeachframeisconvertedintoasetofvisualwordsusingapre-trainedconvolutionalneuralnetwork. 4.2SparseTopicModeli