deep lambertian networks

deep lambertian networks

ID:34654643

大小:4.65 MB

页数:8页

时间:2019-03-08

deep lambertian networks_第1页
deep lambertian networks_第2页
deep lambertian networks_第3页
deep lambertian networks_第4页
deep lambertian networks_第5页
资源描述:

《deep lambertian networks》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库

1、DeepLambertianNetworksYichuanTangtang@cs.toronto.eduRuslanSalakhutdinovrsalakhu@cs.toronto.eduGeoffreyHintonhinton@cs.toronto.eduDepartmentofComputerScience,UniversityofToronto,Toronto,Ontario,CANADAAbstractdirectionandintensitychangesinascene,dramaticchangesinimageintensityoccur

2、.Thisisdetrimen-Visualperceptionisachallengingproblemintaltorecognitionperformanceasmostalgorithmsusepartduetoilluminationvariations.Apos-imageintensitiesasinputs.Anaturalwayofattack-siblesolutionistofirstestimateanillumi-ingthisproblemistolearnamodelwherethealbedo,nationinvarian

3、trepresentationbeforeusingsurfacenormals,andthelightingareexplicitlyrepre-itforrecognition.Theobjectalbedoandsentedasthelatentvariables.Sincethealbedoandsurfacenormalsareexamplesofsuchrep-surfacenormalsarephysicalpropertiesofanobject,resentations.Inthispaper,weintroduceatheyaref

4、eatureswhichareinvariantw.r.t.illumina-multilayergenerativemodelwherethelatenttion.variablesincludethealbedo,surfacenormals,andthelightsource.CombiningDeepBe-SeparatingthesurfacenormalsandthealbedoofliefNetswiththeLambertianreflectanceas-objectsusingmultipleimagesobtainedunderdif

5、-sumption,ourmodelcanlearngoodpriorsferentlightingconditionsisknownasphotometricoverthealbedofrom2Dimages.Illumina-stereo(Woodham,1980).Hayakawa(1994)describedtionvariationscanbeexplainedbychangingamethodforphotometricstereousingSVD,whichonlythelightinglatentvariableinourmodel.e

6、stimatedtheshapeandalbedouptoalineartrans-Bytransferringlearnedknowledgefromsim-formation.Usingintegrabilityconstraints,Yuilleetal.ilarobjects,albedoandsurfacenormalses-(1999)proposedasimilarmethodtoreducetheambi-timationfromasingleimageispossibleinguitiestoageneralizedbasrelief

7、ambiguity.Arelatedourmodel.Experimentsdemonstratethatproblemistheestimationofintrinsicimages(Barrowourmodelisabletogeneralizeaswellasim-&Tenenbaum,1978;Gehleretal.,2011).However,inproveoverstandardbaselinesinone-shotfacethoseworks,theshading(innerproductofthelight-recognition.in

8、gvectorandthesurfacenormalvector)insteadofthesu

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。