ABSTRACT
Pedestrian detection and tracking is a key aspect of many modern computer vision applications concerning human activity recognition. Digital video analysis is exploited to obtain contextual information in several application areas; contextual information may concern, e.g., those related to recognition of human behavior (e.g., detection of suspicious activities or events of threats), localization of people, as well as detection of crowd flows and traffic (e.g., for monitoring the number of people entering a building or crossing a street). These video systems are composed of a combination of digital camera sensors which acquire images depicting the scene of interest, and computer processing units which run dedicated software for the analysis of the scene itself. According to Borges et al.,[1] a tool that applies automatic analysis techniques on video footages for inferring contextual information and performing scene understanding is called an intelligent (or smart) video system.
