Engineering HDR Seminar Series 19, 2011: Research & Development of an Intelligent Surveillance System with Highly Reliable Human Detection and Tracking Capabilities

Mr Mohammed Shahnewaz Chowdhury, PSCT postgraduate student

Date: 2011-12-15
Time: 10:00 to 11:00
Venue: Meeting Room 5-4-22, School of Engineering


Intelligent Visual Surveillance has a wide spectrum of promising applications and has been researched extensively for the past three decades. Numerous approaches have been proposed but due to lack of reliability, very few are actually commercialized into real-world applications to provide public security and safety. The reliability of an intelligent surveillance system depends on multiple factors such as real-time performance, accuracy and robustness in a wide array of real-life problems. This research focuses on narrowing the gap between existing theoretical algorithms and practical implementation by developing an adaptive, completely autonomous, unsupervised learning based multi-camera intelligent visual surveillance system with highly reliable human detection and tracking capabilities in unconstrained scenarios with minimal assumptions. In the preliminary work, the framework of the system is developed to perform unsupervised learning-based human detection. Results show that, equipped with scene contextual information and an environment specific classifier, the unsupervised learning-based detection can perform better than the state-of-the art. Proposed research work includes integrating multiple feature fusion and coherent tracking framework with a boosted version of the unsupervised learning based detection method into a multiple-camera system to follow the trend of wide-area surveillance.

About the Speaker

Mr Mohammed Shahnewaz Chowdhury graduated from Monash University Sunway Campus in 2009 with a 1st class honors in Bachelor of Engineering (Electrical and Computer Systems). He is currently enrolled in the Master of Engineering Science (Research). His research interests are image processing, machine learning and human motion analysis and their integration in the field of intelligent surveillance systems for human detection, human tracking and behavior recognition.