Convolutional Neural Network Based Monocular Depth Estimation

Suleman, Iqra

DSpace Home
→
E-Theses
→
CEME
→
Computer Engineering
→
MS
→
View Item

dc.contributor.author	Suleman, Iqra
dc.date.accessioned	2023-08-09T07:49:40Z
dc.date.available	2023-08-09T07:49:40Z
dc.date.issued	2019
dc.identifier.other	00000203984
dc.identifier.uri	http://10.250.8.41:8080/xmlui/handle/123456789/35972
dc.description	Supervisor: Dr. Muhammad Usman Akram	en_US
dc.description.abstract	Foriailongitime,istereoicamerasihaveibeenideployediinivisualisimultaneousi locationiandimappingi(SLAM)isystemsitoiobtaini3Diinformation.iAlthoughistereoi camerasishowigoodiperformance,itheimainidrawbackiisitheicomplexiandi expensiveihardwareisetupiitirequires,iwhichirestrictsitheiuseiofitheisystem.i Monocularicamerasiareia simpleriandicheaperialternative.iRecentiworkihasi shown ithatiaccessitoidepthimapsiinitheimonocularisystemiisibeneficialiasitheyicanibeiuse ditoiimprovei3Direconstruction.iThisiworkiproposesiaideepineuralinetworkithatipr edictsidenseihigh-resolutionidepthimapsifromimonoculariRGBiimages.i Networkiarchitectureifollowsianiencoder-decoderistructureiiniwhichimultiscaleiinformationiisicapturediandiskip-connectionsiareiuseditoiretrieveidetails.i TheinetworkiisitrainediandievaluatedioniaiNYUiv2idatasetiwithiresultsi comparableitoistate-ofi-ithe-artimethods.iTheiproblemiofidepthiestimationiisiani importanticomponentiforiunderstandingitheigeometryiofiaisceneiandifori navigatingithroughispace.iMoreiunderstandingiofitheienvironment,isuchi asirecognitioniactivities,Icontributesitoichangesiiniotherifields. Inimanyiapplications,iaccurateimeasurementiofidepthifromiimagesiisiaicrucialitask involvingiinterpretationiandirestorationiofitheiscene.iExistingimethodsifori calculatingidepthiofteniyieldifuzzyiapproximationsiwithilowiresolution.iThisi thesisidescribesiaiconvolutionineuralinetworkitoiuseitransferilearningitoi computeiaihighiresolutionimapiofidepthiwithioneisingleiRGBiimage.iUsingiai typicaliencoder-decoderimodel,iwheniinitializingiouriencoder,iweiexploitifeaturesi derivediusingihighiperformanceipre-trainedinetworksialongiwithiextensioniandi testing techniques ithatiresultiinimoreiaccurateiresults. xiii Weidemonstrateihowiouriapproachicaniachieveiaccurate,ihigh-resolutioni depthimaps,ieveniforiaiveryisimpleidecoder.iWeitrainidatasetionithreeimodelsii.e.i denseneti169,i201iandiResnet50.iOurinetworkiconductsiresultsiwithistateiofitheiart ionitwoidatasetsiwithilessiparametersianditrainingiiterations,iandialsoioffersi qualitativelyibetteritestsithatireflectihumaniboundariesimoreiaccurately.iOuri algorithmigivesistateiofitheiartiresultsiandiitigivesirmsivalueiofi0.4611.	en_US
dc.language.iso	en	en_US
dc.publisher	College of Electrical & Mechanical Engineering (CEME), NUST	en_US
dc.subject	Keywords:idepthiestimation,istereo,imonocular,icameras,iconvolutionalineuralinetworks,i NYU Depth v2.	en_US
dc.title	Convolutional Neural Network Based Monocular Depth Estimation	en_US
dc.type	Thesis	en_US