pedestrian detection an evaluation of the state of the art

pedestrian detection an evaluation of the state of the art

ID:39759759

大小:4.25 MB

页数:20页

时间:2019-07-11

pedestrian detection an evaluation of the state of the art_第1页
pedestrian detection an evaluation of the state of the art_第2页
pedestrian detection an evaluation of the state of the art_第3页
pedestrian detection an evaluation of the state of the art_第4页
pedestrian detection an evaluation of the state of the art_第5页
资源描述:

《pedestrian detection an evaluation of the state of the art》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库

1、SUBMISSIONTOIEEETRANSACTIONSONPATTERNANALYSISANDMACHINEINTELLIGENCE1PedestrianDetection:AnEvaluationoftheStateoftheArtPiotrDollar,ChristianWojek,BerntSchiele,andPietroPerona´Abstract—Pedestriandetectionisakeyproblemincomputervision,withseveralapplicationsthathavethepotentialto

2、positivelyimpactqualityoflife.Inrecentyears,thenumberofapproachestodetectingpedestriansinmonocularimageshasgrownsteadily.However,multipledatasetsandwidelyvaryingevaluationprotocolsareused,makingdirectcomparisonsdifficult.Toaddresstheseshortcomings,weperformanextensiveevaluation

3、ofthestateoftheartinaunifiedframework.Wemakethreeprimarycontributions:(1)weputtogetheralarge,well-annotatedandrealisticmonocularpedestriandetectiondatasetandstudythestatisticsofthesize,positionandocclusionpatternsofpedestriansinurbanscenes,(2)weproposearefinedper-frameevaluation

4、methodologythatallowsustocarryoutprobingandinformativecomparisons,includingmeasuringperformanceinrelationtoscaleandocclusion,and(3)weevaluatetheperformanceofsixteenpre-trainedstate-of-the-artdetectorsacrosssixdatasets.Ourstudyallowsustoassessthestateoftheartandprovidesaframewo

5、rkforgaugingfutureefforts.Ourexperimentsshowthatdespitesignificantprogress,performancestillhasmuchroomforimprovement.Inparticular,detectionisdisappointingatlowresolutionsandforpartiallyoccludedpedestrians.IndexTerms—pedestriandetection,objectdetection,benchmark,evaluation,datas

6、et,CaltechPedestrianDatasetF1INTRODUCTIONPeopleareamongthemostimportantcomponentsofamachine’senvironment,andendowingmachineswiththeabilitytointeractwithpeopleisoneofthemostinterestingandpotentiallyusefulchallengesformodernengineering.Detectingandtrackingpeopleisthusan(a)Caltec

7、h[3](b)Caltech-Japan[3]importantareaofresearch,andmachinevisionisboundtoplayakeyrole.Applicationsincluderobotics,enter-tainment,surveillance,carefortheelderlyanddisabled,andcontent-basedindexing.JustintheUS,nearly5,000ofthe35,000annualtrafficcrashfatalitiesinvolvepedes-trians[1

8、],hencetheconsiderableinterestinbuildingautomatedvisionsystem

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。