Design of Real-Time Multiple Object Visual Detection and Tracking Systems

Abstract

Multiple object detection and tracking is one of the main topics in computer vision. The task is to identify all the objects of interest in a frame of a video and to assign an unique ID to those instances that correspond to the same object while it appears in the scene. This is a fundamental task of many video analytics applications like traffic monitoring or video surveillance, which usually requires real-time processing speed and its execution on different hardware devices. In this PhD Thesis we address the topic of Real-Time Multiple Object Detection and Tracking Systems, combining state-of-the-art detectors, trackers and data association techniques. Particularly, we focus on the design of Real- Time Multiple Object Detection and Tracking systems for both server architectures and embedded devices, that are able to work with dozens of objects in real-time.