A Review of Services-Based Architectures for Video Surveillance

Visual Video Surveillance: Detection, Monitoring, And Analysis Of The Behaviour Of People And Objects

In the field of video surveillance, investigators are confronted with the problem of automatic image processing, which makes it necessary to provide computer components with intelligence so that they can automatically analyze image sequences.

Several works analyze behavior in group activities and their interaction with individual behavior. Similarly, around abnormal behavior in complex scenarios, methods based on the representation of movement are investigated. This presents the novelty of utilizing context pattern information. It consists of modeling crowd activity using movement and context information, taking into account that for all objects or regions with detected movement, context information is provided between them (Girshick, Donahue, Darrell, & Malik, 2014). The authors conclude that this method improves on its predecessors since it does not only consider the movement of the crowd but also uses context information provided by the surrounding objects. Also, false alarms are detected when there is significant perspective information or the size of the objects is very varied, which has a solution using different scales of objects in the training (Sudhakaran & Lanz, 2017).