Improving Image Segmentation with Adaptive, Recurrent, Spiking Neural Network Models of the Primary Visual Cortex

Abstract

. Automatic object recognition from still imagery, insensitive to clutter and partial occlusion, is an unsolved computer vision problem with countless applications to military readiness. Ambiguity of segmentation of complex images into objects is the major stumbling block. Incorporation of certain structural features of the primate early visual system into computational models has been suggested as a potential solution. However, little is known about effects of these features on segmentation performance of either humans or computational models

Open PDF

Document Details

Document Type
Technical Report
Publication Date
May 19, 2017
Accession Number
AD1050673

Entities

People

  • Ilya Nemenman

Organizations

  • Emory University

Tags

Communities of Interest

  • Human Systems

DTIC Thesaurus Topics

  • Abstracts
  • Agreements
  • Ambiguity
  • Artificial Intelligence Computing
  • Artificial Intelligence Software
  • Artificial Neural Networks
  • Automatic
  • Computer Vision
  • Computers
  • Department Of Defense
  • Education
  • Engineering
  • Image Segmentation
  • Information Operations
  • Mathematics
  • Military Research
  • Neural Networks
  • Object Recognition
  • Recognition
  • Simulations
  • Students
  • Vascular System Injuries
  • Visual Cortex

Fields of Study

  • Computer science

Readers

  • Educational Psychology
  • Neural Network Machine Learning.
  • Radar Systems Engineering.

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks