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

Top-down learning of low-level vision tasks

M J Jones1, P Sinha, T Vetter

  • 1E25-201, Center for Biological and Computational Learning, Massachusetts Institute of Technology, 45 Carleton Street, Cambridge, Massachusetts 02142, USA.

Current Biology : CB
|February 21, 1998
PubMed
Summary
This summary is machine-generated.

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

Mid-upper arm circumference predicts treatment outcomes in paediatric TB: a cohort study.

IJTLD open·2026
Same author

Observation of Charmonium Sequential Suppression in Heavy-Ion Collisions at the Relativistic Heavy Ion Collider.

Physical review letters·2026
Same author

Impact of indoor ventilation on TB transmission risk: implications of climate change.

The international journal of tuberculosis and lung disease : the official journal of the International Union against Tuberculosis and Lung Disease·2026
Same author

Nutritional interventions for TB: a critical component of integrated TB care amid funding challenges.

The international journal of tuberculosis and lung disease : the official journal of the International Union against Tuberculosis and Lung Disease·2026
Same author

Low TB1 and TB2 antigen-nil is associated with increased QuantiFERON Gold Plus reversions.

The international journal of tuberculosis and lung disease : the official journal of the International Union against Tuberculosis and Lung Disease·2026
Same author

Energy Independence of the Collins Asymmetry in p^{↑}p Collisions.

Physical review letters·2026
Same journal

Pitch selectivity in ferret auditory cortex.

Current biology : CB·2026
Same journal

A cell size-dependent competition between geometry and polarity governs nuclear and spindle positioning in early embryos.

Current biology : CB·2026
Same journal

Trophic cascades drive sustainability in the agricultural heritage rice-fish coculture system.

Current biology : CB·2026
Same journal

Tracking Satb2-positive retinal ganglion cells in zebrafish unveils developmental functional reorganization.

Current biology : CB·2026
Same journal

RhoGAP54D promotes cell size asymmetry and inhibits pulsatile myosin activity in Drosophila neural stem cells.

Current biology : CB·2026
Same journal

Increased rates of hybridization in swordtails are associated with water pollution.

Current biology : CB·2026
See all related articles

This study introduces a novel computational model for vision tasks. The model learns object-specific, top-down processing, showing high tolerance to noise and incomplete data, unlike traditional methods.

Area of Science:

  • Computer Vision
  • Cognitive Science
  • Computational Neuroscience

Background:

  • Low-level vision tasks like edge detection are typically solved using generic, bottom-up methods.
  • These traditional approaches are often sensitive to noise and incomplete visual information.

Purpose of the Study:

  • To present a simple computational model for perceptual tasks using a top-down, object-class-specific, and example-based approach.
  • To evaluate the model's performance on edge detection and 3D structure prediction and compare it with conventional methods.

Main Methods:

  • Developed a computational model employing a top-down, example-based strategy.
  • Tested the model on edge-detection and view-prediction tasks for 3D objects.
  • Compared the model's robustness against sensor noise and incomplete input with bottom-up strategies.

Related Experiment Videos

Main Results:

  • The model's performance on edge detection and view prediction aligns with human perceptual expectations.
  • The computational model demonstrated high tolerance to sensor noise and incomplete input data.
  • Conventional bottom-up strategies exhibited significantly less immunity to these visual challenges.

Conclusions:

  • The model's success supports the hypothesis that the human visual system might learn to perform seemingly low-level vision tasks via top-down processing.
  • Object-class-specific, example-based learning offers a robust alternative to generic, bottom-up approaches in computer vision.