Groups and Crowds: Behaviour Analysis of People Aggregations

Automatic analysis of human behavior in social environment is a key topic for the computer vision community, with applications in security and video surveillance. While human behavior at an individual (single person) level has been widely studied in the past years, analysis of groups and crowd behavior, is still at a preliminary stage, with room for new approaches to emerge. Recently, there has been significant research effort dedicated to the development of automated computer vision techniques, intended to enhance safety of our societies by monitoring human behaviors and their actions in groups and crowd level. In particular, groups are usually formed by number of people who gathered for private meeting, birthday party, or wedding, while we consider crowd as huge number of people are gathered together to participate for a national or religious event, or protest due to some dissatisfaction. In this chapter, we will provide a broad overview on proposed approaches on human behavior analysis in group and crowd level, as well as, a detailed of some most recent state-of-the-art methods along with extensive experiments and comparison