Analysis and detection of crowd behaviour using cognitive models in smart surveillance

Abstract

Surveillance is the practice of monitoring the activities and behavior of individuals and newlineobjects in public places to ensure the safety and security of people and their assets. newlineNowadays, traditional closed-circuit television (CCTV) cameras have been replaced newlineby smart surveillance systems, which are equipped with smart visual sensors, and use newlineartificial intelligence (AI) -based programs to analyze captured video and act accordingly. newlineThe analysis of video data by applying computer vision methods and creating newlinereal-time intelligence appropriate for the observed environment is coined with the term newlinevideo analytics. In a public surveillance system, video analytics helps to detect unusual newlinemovements, breaking of traffic rules, parking in unauthorized areas, etc. One of the newlinemajor application areas of a smart surveillance system is monitoring individuals and newlinetheir activities, especially in crowded areas where there are chances for disasters and newlinecrime-related incidents. Monitoring and managing the crowd is a tedious task due to newlinethe complex and unpredictable behavior exhibited by the crowd. The crowded scenarios newlinehave a high tendency to turn into abnormal situations due to sudden external pressures newlinesuch as gunshots/fire or internal stress such as overcrowding, where things may often newlineget out of control, and the consequences are devastating. Over the years, crowd-related newlineincidents and their casualties have been increasing, accompanied by post-disaster suffering. newlineThe main reason behind such disasters is the non-adaptive behavior of the crowd newlineinstead of the actual cause. Therefore, a smart surveillance system should detect and newlinepredict mishaps by analyzing the psychological factors of these unpredictable behaviors. newlineAccording to Edward Bernays, an Austrian-American pioneer, and father of public newlinerelations, if a monitoring system understands the non-adaptive mindset and psychological newlineaspects of the crowd, the behaviors can be identified easily, and thus the crowd newlinemanagement is effortless.