Crowd behavior anomaly detection using correlation of optical flow magnitude

Abstract

Nowadays, crowd monitoring is a contentious issue. Because of the increasing population and diversity of human activities, crowd scenarios in the real world are becoming more common, demanding the need for an automotive anomaly detection system. Crowd behavior is influenced by the thoughts and attitudes of others around them. An unexpected event can turn a peaceful crowd into a riot. A mechanism based on optical flow must be implemented to compensate for all of these factors. The amount of motion present in two successive frames is estimated using optical flow. It includes information on velocity in the x & y plane, along with magnitude and line of action. By means of “anomalous event” in this paper is quick and sudden dispersal of the crowd. For detecting an event the magnitude of two successive frames should be taken into account followed by estimating a correlation. We expect a high correlation, slight motion, and low rate of change in velocities at non-anomalous events, but as soon as an anomalous event occurs, the correlation begins to decrease with a significant change in velocity and large motion vectors. The methodology was tested on a dataset from the University of Minnesota that included 11 movies from three different circumstances. Almost all anomalous occurrences in videos were successfully detected using this method.

Anomaly Detection in Crowd Surveillance using Edge Computing: A Comprehensive Survey

“…Metrics are reported but exhibit lower accuracy in certain scenarios. With a focus on crowd dispersal, this paper [22] analyzes optical flow and correlation coefficients but primarily addresses one type of anomaly. It lacks metrics and comparisons with other object detection solutions.…”

“…Challenges persist, including model limitations, dataset diversity, and edge computing complexities. Computer science publications, 20 million+ articles, conferences, books UCSD [12], [13], [18], [19], [21], [24], [28], [34], [36] Anomaly detection, video, 1617 videos, normal/abnormal activities CUHK Avenue [12], [13], [17], [18], [19], [21], [24], [34], [36] Pedestrian re-identification, high-resolution, 31k images, 1.5k identities ShanghaiTech [12], [17], [21], [24], [34] Person re-identification, large-scale, 486 cameras, 306k images, 111k identities UMN [13], [22], [24], [28], [29], [ [25] Traffic flow forecasting, anomaly detection Vishnu Society Data [26] Gated community, India, vehicles, pedestrians, events, images, videos Hockey Fights [29] Video, hockey fights, 11k videos, fight detection, outcome Violent Flows [29] Video, crowd anomaly detection, 126 videos, violent/non-violent events How-ever, through collaborative efforts and diverse methodologies, the surveyed papers are expanding the horizons of what's achievable, enhancing public safety and security. As the field continues to evolve, it is the intersection of these insights that will pave the way for the future of anomaly detection in video surveillance, ensuring a safer world.…”

Methodological Insights Towards Leveraging Performance in Video Object Tracking and Detection

“…There are various unique studies being carried out towards object detection in the perspective of event detection and behavioral analysis. The work of Chakole et al [71] has addressed the issues pertaining to anomaly detection for the use-case of crowd behavior using correlation-based optical flow. Object detection has also been studied with respect to recognition mechanism of human activity.…”