University of Minnesota
From Body Parts to Body Structure: Visual Information and Depth Perception from Sparse Human Body Images
Abstract
dc:description.abstractHuman observers can infer three-dimensional body structure from remarkably sparse visual information. However, it remains unclear how body-part recognition, body structure, and geometric information contribute to depth perception when visual information is limited. This dissertation investigates depth perception from sparse body stimuli using a series of behavioral experiments and computational observer analyses. Chapter 2 showed that depth judgments remained above chance even when body-part recognition approached chance, demonstrating a dissociation between recognition and depth perception. Chapter 3 found that relations among body parts influenced depth judgments but were not uniformly beneficial; preserving natural spatial relationships could impair performance. Chapter 4 examined whether explicit structural inference contributes to depth perception. Although participants recovered body structure above chance, successful structural inference did not reliably predict depth performance. Instead, depth accuracy was more strongly related to geometric properties of the stimulus. Computational observer analyses further demonstrated that models based on geometric information substantially outperformed models based on visual appearance alone. Together, these findings suggest that depth perception from sparse body information depends more strongly on geometric cues than on successful recognition or explicit recovery of body structure. More broadly, the results indicate that recognition, structural inference, and depth perception are partially dissociable processes that make distinct contributions to perceptual inference under visual uncertainty.
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Liu, Jiaqi
Subjects
dc:subject × 2Rights
- Language dc:language.iso
- en
Identifiers
dc:identifier.*- Dc Identifier Citation
- Liu, Jiaqi. (2026). From Body Parts to Body Structure: Visual Information and Depth Perception from Sparse Human Body Images. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/283113.