A Comparative Analysis of Techniques for Crowd Behaviour Detection in Dense Scenes

Abstract

Behaviour analysis is considered a critical area of research in the computer vision research community. Recently, visual monitoring systems of human gathering have found application in diferent areas such as safety, security, entertainment, and personal archives. Although many approaches have been proposed, certain limitations exist and many unresolved issues remain open. The objective of the paper is to present recent advances on abnormal human behaviour analysis and hierarchical crowd behaviour classifcation based on the level of complexity. The paper also provides a clear perspective with a broad and in-depth review of the research conducted in this area. In addition, it points out unresolved problems that demand improvement. In general, researchers can use this paper as a starting point for further advancement of behavioural analytic methods to propose novel approaches as well as an exploration of approaches that have received meager attention. This study also investigates the performance comparison of state-of-the-art techniques on anomaly detection: hand-crafted feature approach, and deep learning approach. Finally, limitations of the current methods and promising future research directions are presented.