Adaptive Detection and Tracking of Multiple Persons in Real-Time Video Surveillance

Video surveillance networks are currently being deployed in a growing number of security checkpoints and retails stores. Genetec Inc. provides IP-based solutions for video surveillance, access control and license plate recognition, all integrated into their Security Center platform. Although video analytic techniques to search for people in a scene have become a key business priority, real-time detection and tracking of several individuals remains a challenging problems in real-world environments. The objective of this project consists in developing adaptive detection and tracking systems that can robustly locate multiple individuals in complex and real-time video surveillance applications. These systems will autonomously create, associate and remove individual tracks based on information extracted from the appearance of each person’s head. Tracking-by-detection approaches will be developed for robust head tracking, where each head is modeled using an ensemble of classifiers that is continuously adapted using input samples. The facial ROIs captured along each head track will be regrouped and ranked using quality measures for visualisation, analysis and recognition of each person’s face. The successful outcome of this research project will lead to the commercialization by Genetec Inc. of cost-effective software tools based on the proposed adaptive systems. TO BE CONT’D

Faculty Supervisor:

Éric Granger

Student:

Partner:

Genetec Inc

Discipline:

Engineering

Sector:

Professional, scientific and technical services

University:

École de technologie supérieure

Program: