Multi-camera Human Tracking via Clustering

Abstract

We describe an efficient method for tracking humans in a multi-camera network. Based on online clustering, the proposed method groups single-camera tracks into mutually exclusive subsets corresponding to individuals. Experiments indicate that the algorithm is capable of achieving strong tracking accuracy on simulated data.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Nov 01, 2011
Accession Number
ADA558220

Entities

People

  • Dmitri Kamenetsky

Organizations

  • Defence Science and Technology Group

Tags

Communities of Interest

  • Energy and Power Technologies
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Abstracts
  • Accuracy
  • Algorithms
  • Applied Computer Science
  • Artificial Intelligence
  • Australia
  • Clustering
  • Computations
  • Computer Science
  • Detection
  • Errors
  • Machine Learning
  • Probability
  • Reconnaissance
  • Simulations
  • Surveillance

Fields of Study

  • Computer science

Readers

  • Computer Vision.
  • Systems Analysis and Design