서브메뉴
검색
From Parts to Pose: Understanding Hierarchical and Structural Representation in Human Body Perception
From Parts to Pose: Understanding Hierarchical and Structural Representation in Human Body Perception
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202105259
- ISBN
- 9798263320416
- DDC
- 153
- 저자명
- Liu, Ziwei.
- 서명/저자
- From Parts to Pose: Understanding Hierarchical and Structural Representation in Human Body Perception
- 발행사항
- [Sl] : University of Minnesota, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 112 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-05, Section: B.
- 주기사항
- Advisor: Kersten, Daniel.
- 학위논문주기
- Thesis (Ph.D.)--University of Minnesota, 2025.
- 초록/해제
- 요약The ability to perceive and interpret human body structure from incomplete or ambiguous visual input is central to visual cognition and social understanding. Yet the mechanisms that enable humans to infer coherent body configurations from sparse visual evidence remain an open question in visual cognition. This dissertation investigates the cognitive and computational mechanisms underlying human body perception, focusing on how individuals recognize and reconstruct body poses from partial, occluded, or low-resolution visual input. The research is structured to embody various levels of perceptual complexity, from recognizing isolated body parts to integrating parts into full body poses. Chapter 1 provides an overview of the theoretical foundations of object and body recognition, highlighting the debate between structural description and image-based models, and introducing body perception as a specialized domain of object recognition. Chapter 2, "From Pixels to Parts," compares human performance with deep convolutional neural networks (DCNNs) in recognizing isolated body parts under spatial constraints, revealing shared reliance on part-level structural features and highlighting how both systems scale recognition with increasing spatial context. Chapter 3, "From Parts to Pairs," extends this investigation by examining how recognition of body parts is influenced by their spatial relationships, emphasizing the role of pairwise spatial context in body part identification. Chapter 4, "From Pairs to Pose," delves into the hierarchical nature of body part perception, demonstrating how the visual system integrates local and global information to efficiently reconstruct body poses, and providing evidence for the use of relational knowledge to enhance body perception under varying visual complexity. The findings throughout this dissertation highlight a dynamic, relational framework for body perception, in which both part-level features and structural relationships are crucial for efficient body pose recognition. Together, this research offers novel theoretical insights into how the brain processes body information, advancing our understanding of visual inference and providing a benchmark for evaluating computational models of human body recognition.
- 일반주제명
- Cognitive psychology
- 일반주제명
- Experimental psychology
- 키워드
- Visual cognition
- 기타저자
- University of Minnesota Psychology
- 기본자료저록
- Dissertations Abstracts International. 87-05B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017360071
■00520260202105259
■006m o d
■007cr#unu||||||||
■020 ▼a9798263320416
■035 ▼a(MiAaPQ)AAI32280445
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a153
■1001 ▼aLiu, Ziwei.
■24510▼aFrom Parts to Pose: Understanding Hierarchical and Structural Representation in Human Body Perception
■260 ▼a[Sl]▼bUniversity of Minnesota▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a112 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-05, Section: B.
■500 ▼aAdvisor: Kersten, Daniel.
■5021 ▼aThesis (Ph.D.)--University of Minnesota, 2025.
■520 ▼aThe ability to perceive and interpret human body structure from incomplete or ambiguous visual input is central to visual cognition and social understanding. Yet the mechanisms that enable humans to infer coherent body configurations from sparse visual evidence remain an open question in visual cognition. This dissertation investigates the cognitive and computational mechanisms underlying human body perception, focusing on how individuals recognize and reconstruct body poses from partial, occluded, or low-resolution visual input. The research is structured to embody various levels of perceptual complexity, from recognizing isolated body parts to integrating parts into full body poses. Chapter 1 provides an overview of the theoretical foundations of object and body recognition, highlighting the debate between structural description and image-based models, and introducing body perception as a specialized domain of object recognition. Chapter 2, "From Pixels to Parts," compares human performance with deep convolutional neural networks (DCNNs) in recognizing isolated body parts under spatial constraints, revealing shared reliance on part-level structural features and highlighting how both systems scale recognition with increasing spatial context. Chapter 3, "From Parts to Pairs," extends this investigation by examining how recognition of body parts is influenced by their spatial relationships, emphasizing the role of pairwise spatial context in body part identification. Chapter 4, "From Pairs to Pose," delves into the hierarchical nature of body part perception, demonstrating how the visual system integrates local and global information to efficiently reconstruct body poses, and providing evidence for the use of relational knowledge to enhance body perception under varying visual complexity. The findings throughout this dissertation highlight a dynamic, relational framework for body perception, in which both part-level features and structural relationships are crucial for efficient body pose recognition. Together, this research offers novel theoretical insights into how the brain processes body information, advancing our understanding of visual inference and providing a benchmark for evaluating computational models of human body recognition.
■590 ▼aSchool code: 0130.
■650 4▼aCognitive psychology
■650 4▼aExperimental psychology
■653 ▼aBody-part recognition
■653 ▼aComputational modeling
■653 ▼aDeep convolutional neural networks
■653 ▼aHuman pose perception
■653 ▼aVisual cognition
■653 ▼aVisual perception
■690 ▼a0633
■690 ▼a0623
■690 ▼a0800
■71020▼aUniversity of Minnesota▼bPsychology.
■7730 ▼tDissertations Abstracts International▼g87-05B.
■790 ▼a0130
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17360071▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


