Anomaly Detection in Crowded Places: Review

In today's world, security is based on the manually monitored surveillance cameras used at various places. Since, human based monitoring is very tiring and difficult task, this raises the demand for automatic system to detect anomaly at crowded places. It involves understanding crowd behavior, their interaction/collision with each other, individual speed, crowd density, different classification methods to identify anomaly etc. This paper studies the different techniques used to study crowd behavior, individual and crowd characteristics. It also compares the results of various methods used in survey papers for common datasets.