Dynamics-Based Video Analytics
R4-A1

Download Project Report

Project Description

This research aims to substantially enhance the ability to conduct autonomous, video based, persistent intelligent surveillance, and reconnaissance and threat assessment in highly uncertain, adversarial scenarios, such as urban environments. The main idea is the use of operator theoretic and convex analysis methods to recast several key sub-problems arising in this context (tracking, dynamic appearance and activity recognition) into a finite dimensional convex optimization that can be efficiently solved.

Project Leader
  • Octavia Camps
    Professor
    Northeastern University
    Email

  • Mario Sznaier
    Professor
    Northeastern University
    Email