Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Manhattan world: orientation and outlier detection by Bayesian inference.

James M Coughlan1, A L Yuille

  • 1Smith-Kettlewell Eye Research Institute, San Francisco, CA 94115, U.S.A. coughlan@ski.org

Neural Computation
|June 14, 2003
PubMed
Summary

Many visual scenes utilize a "Manhattan" 3D grid structure, influencing image statistics. A Bayesian model accurately estimates viewer orientation and detects outliers within this Manhattan world framework.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

NavGraph: Enhancing Blind Travelers' Navigation Experience and Safety.

Proceedings of the ACM on interactive, mobile, wearable and ubiquitous technologies·2025
Same author

Enhancing Walk-Light Detector Usage for the Visually Impaired: A Comparison of VR Exploration and Verbal Instructions.

... International web for all conference. Web for All Conference·2025
Same author

MapIO: A Gestural and Conversational Interface for Tactile Maps.

IEEE access : practical innovations, open solutions·2025
Same author

Systematically Evaluating Digital Map Tools Based on the WCAG.

Journal on technology and persons with disabilities : ... Annual International Technology and Persons with Disabilities Conference·2025
Same author

Designing Born-Accessible Courses in Data Science and Visualization: Challenges and Opportunities of a Remote Curriculum Taught by Blind Instructors to Blind Students.

Eurographics/IEEE VGTC Symposium on Visualization : EUROVIS : [proceedings]. Eurographics/IEEE VGTC Symposium on Visualization·2025
Same author

Accessible Point-and-Tap Interaction for Acquiring Detailed Information about Tactile Graphics and 3D Models.

Computers helping people with special needs : ... International Conference, ICCHP ... : proceedings. International Conference on Computers Helping People with Special Needs·2024

Area of Science:

  • Computer Vision
  • Computational Imaging
  • 3D Scene Understanding

Background:

  • Visual scenes often exhibit regularities in their structure.
  • Understanding 3D scene geometry is crucial for image interpretation.
  • Previous models may not fully capture inherent scene regularities.

Purpose of the Study:

  • To propose and validate a Bayesian model based on the "Manhattan world" assumption.
  • To estimate viewer orientation relative to a 3D grid structure.
  • To identify outlier structures deviating from the assumed grid.

Main Methods:

  • Constructed a Bayesian model incorporating the Manhattan world assumption.
  • Estimated viewer orientation using the developed Bayesian model.
  • Implemented a null hypothesis model and log-likelihood ratio test to assess model applicability.

Related Experiment Videos

  • Compared model-based orientation estimates with manual estimations.

    • Main Results:

    • The Bayesian model accurately estimates viewer orientation for images conforming to the Manhattan world assumption.
    • The model effectively detects structures misaligned with the grid.
    • The Manhattan world assumption significantly improves orientation estimation accuracy.

    Conclusions:

    • The Manhattan world model provides a robust framework for analyzing visual scenes with grid-like structures.
    • Viewer orientation estimation is significantly enhanced by assuming a Manhattan world.
    • This approach offers a reliable method for identifying scene outliers and understanding 3D geometry.