본문

서브메뉴

Shape-Biased Representations for Object Category Recognition
Shape-Biased Representations for Object Category Recognition
Shape-Biased Representations for Object Category Recognition

상세정보

자료유형  
 학위논문 서양
최종처리일시  
20260209102903
ISBN  
9798263394059
DDC  
006
저자명  
Stojanov, Stefan.
서명/저자  
Shape-Biased Representations for Object Category Recognition
발행사항  
[Sl] : Georgia Institute of Technology, 2023
발행사항  
Ann Arbor : ProQuest Dissertations & Theses, 2023
형태사항  
121 p
주기사항  
Source: Dissertations Abstracts International, Volume: 87-05, Section: A.
주기사항  
Advisor: Rehg, James M.;Hoffman, Judy.
학위논문주기  
Thesis (Ph.D.)--Georgia Institute of Technology, 2023.
초록/해제  
요약While object recognition is a classical problem in computer vision that has witnessed incredible progress as a result of contemporary deep learning research, the key challenges of developing systems that can learn object categories from continually arriving data, from a few samples, and with limited supervision still remain. In this dissertation, we aim to borrow the learning strategy of shape bias and environmental bias of repetition, both of which are observed in young children, and apply them to continual, low-shot, and self-supervised learning of objects and object parts. In the continual learning domain, we demonstrate that repetition of learned concepts significantly ameliorates catastrophic forgetting. For low-shot learning we develop two methods for learning shape-biased object representations with decreasing supervision requirements: based on learning a joint image and 3D shape metric space from point clouds, and by self-supervised learning of object parts from multi-view pixel correspondences. We demonstrate that these methods of introducing a shape bias improve low-shot category recognition. Last, we find that contrastive learning from multi-view images allows for category-level part matching with performance competitive with baselines that have over 10 times more parameters, while being trained only on synthetic data. To support our investigations, we present two synthetic 3D object datasets, Toys200 and Toys4K, and develop a series of highly realistic synthetic data rendering systems that enable real-world generalization.
일반주제명  
Deep learning
일반주제명  
Failure analysis
일반주제명  
Computer vision
일반주제명  
School environment
일반주제명  
Toys
일반주제명  
Visualization
일반주제명  
Geometry
일반주제명  
Semantics
일반주제명  
Computer science
일반주제명  
Educational leadership
일반주제명  
Educational administration
기타저자  
Georgia Institute of Technology.
기본자료저록  
Dissertations Abstracts International. 87-05A.
전자적 위치 및 접속  
로그인 후 원문을 볼 수 있습니다.

MARC

 008260203s2023        us                              c    eng  d
■001000017365959
■00520260209102903
■006m          o    d                
■007cr#unu||||||||
■020    ▼a9798263394059
■035    ▼a(MiAaPQ)AAI32315563
■035    ▼a(MiAaPQ)GeorgiaTech73115
■040    ▼aMiAaPQ▼cMiAaPQ
■0820  ▼a006
■1001  ▼aStojanov,  Stefan.
■24510▼aShape-Biased  Representations  for  Object  Category  Recognition
■260    ▼a[Sl]▼bGeorgia  Institute  of  Technology▼c2023
■260  1▼aAnn  Arbor▼bProQuest  Dissertations  &  Theses▼c2023
■300    ▼a121  p
■500    ▼aSource:  Dissertations  Abstracts  International,  Volume:  87-05,  Section:  A.
■500    ▼aAdvisor:  Rehg,  James  M.;Hoffman,  Judy.
■5021  ▼aThesis  (Ph.D.)--Georgia  Institute  of  Technology,  2023.
■520    ▼aWhile  object  recognition  is  a  classical  problem  in  computer  vision  that  has  witnessed  incredible  progress  as  a  result  of  contemporary  deep  learning  research,  the  key  challenges  of  developing  systems  that  can  learn  object  categories  from  continually  arriving  data,  from  a  few  samples,  and  with  limited  supervision  still  remain.  In  this  dissertation,  we  aim  to  borrow  the  learning  strategy  of  shape  bias  and  environmental  bias  of  repetition,  both  of  which  are  observed  in  young  children,  and  apply  them  to  continual,  low-shot,  and  self-supervised  learning  of  objects  and  object  parts.  In  the  continual  learning  domain,  we  demonstrate  that  repetition  of  learned  concepts  significantly  ameliorates  catastrophic  forgetting.  For  low-shot  learning  we  develop  two  methods  for  learning  shape-biased  object  representations  with  decreasing  supervision  requirements:  based  on  learning  a  joint  image  and  3D  shape  metric  space  from  point  clouds,  and  by  self-supervised  learning  of  object  parts  from  multi-view  pixel  correspondences.  We  demonstrate  that  these  methods  of  introducing  a  shape  bias  improve  low-shot  category  recognition.  Last,  we  find  that  contrastive  learning  from  multi-view  images  allows  for  category-level  part  matching  with  performance  competitive  with  baselines  that  have  over  10  times  more  parameters,  while  being  trained  only  on  synthetic  data.  To  support  our  investigations,  we  present  two  synthetic  3D  object  datasets,  Toys200  and  Toys4K,  and  develop  a  series  of  highly  realistic  synthetic  data  rendering  systems  that  enable  real-world  generalization.
■590    ▼aSchool  code:  0078.
■650  4▼aDeep  learning
■650  4▼aFailure  analysis
■650  4▼aComputer  vision
■650  4▼aSchool  environment
■650  4▼aToys
■650  4▼aVisualization
■650  4▼aGeometry
■650  4▼aSemantics
■650  4▼aComputer  science
■650  4▼aEducational  leadership
■650  4▼aEducational  administration
■690    ▼a0800
■690    ▼a0984
■690    ▼a0449
■690    ▼a0514
■71020▼aGeorgia  Institute  of  Technology.
■7730  ▼tDissertations  Abstracts  International▼g87-05A.
■790    ▼a0078
■791    ▼aPh.D.
■792    ▼a2023
■793    ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17365959▼nKERIS▼z이  자료의  원문은  한국교육학술정보원에서  제공합니다.

미리보기

내보내기

chatGPT토론

Ai 추천 관련 도서


    신착도서 더보기
    최근 3년간 통계입니다.

    소장정보

    • 예약
    • 소재불명신고
    • 나의폴더
    • 우선정리요청
    • 비도서대출신청
    • 야간 도서대출신청
    소장자료
    등록번호 청구기호 소장처 대출가능여부 대출정보
    TF15739 전자도서 대출가능 마이폴더 부재도서신고 비도서대출신청 야간 도서대출신청

    * 대출중인 자료에 한하여 예약이 가능합니다. 예약을 원하시면 예약버튼을 클릭하십시오.

    해당 도서를 다른 이용자가 함께 대출한 도서

    관련 인기도서

    로그인 후 이용 가능합니다.