INTELLIGENT CROWD ANALYSIS BY DETECTING AND COUNTING PEOPLE IN DENSELY CROWDED AERIAL IMAGES
Author(s):
Saeeda Kanwal1, Muhammad Khurram2, Muhammad Asad Arfeen3, Jie Li4
1 Postgraduate student, Department of Computer and Information Systems Engineering, NED University of Engineering and Technology, Pakistan, Ph. +92-336-8458203, Email: mkhurrum@neduet.edu.pk.
2 Associate Professor, Department of Computer and Information Systems Engineering, NED University of Engineering and Technology, Pakistan, Ph. +92-21-99261261, Fax: +92-21-99261255, Email: mkhurrum@neduet.edu.pk.
3 Assistant Professor, Department of Computer and Information Systems Engineering, NED University of Engineering and Technology, Pakistan, Ph. Ph. +92-21-99261261, Fax: +92-21-99261255, Email: arfeen@neduet.edu.pk.
4 Chief Technology Officer, Graham Innovations, Melbourne, Australia, Ph. +61-3-98744954, Email: Jack@grahaminnovations.com.
Volume:
Thematic Issue on Advances in Image and Video Processing
Pages:
57 - 64
Date:
May 2018
Abstract:
Crowd monitoring and analysis is an extremely focused area for visual surveillance and management of crowded events. A crowd counting method in densely crowded situations has been presented in this paper. The developed technique detects people and counts them in a massive dense crowd based on head region detection using a single aerial snap. Gaussian mixture model (an unsupervised machine learning technique) was employed to train the intelligent system for foreground detection (head) to obtain authentic segmentation outcomes. Further, by using blob analysis the numbers of people were estimated in the crowd. The approach is tested on different types of images and analysed in variety of situations with the head counts up to four thousand five hundred sixty three. The results showed accuracy up to ninety percent under different environmental conditions.