Dynamic Emerging Pathways in Entrance and Exit Detection: Integrating Deep Learning and Mathematical Modeling

Entrance and exit event detection in dynamic environments has a lot of real-world applications in security, crowd management, and retail analytics. Traditional methods used for this problem, namely Line Partition and Bounding Box Diameter methods often struggle in complex scenarios that contain less predictable movement patterns of individuals. This paper proposes a model that integrates deep learning-based object detection and tracking techniques with linear regression to enhance the overall performance of enter and exit detection in static and dynamic environments. This approach captures the movement patterns using advanced object detection and tracking algorithms, enabling the extraction of y-coordinate variations from bounding box centers which are used to calculate the tangent of the linear regression equation and determine if the event is entrance or exit. Experimentations were conducted on 132 video sequences and show the superiority of our approach over the traditional methods achieving an overall accuracy of 86.36% and an F1-score of 0.86. These results demonstrate the high efficiency of this approach to accurately detect entrance and exit events, making it highly reliable and applicable to this problem. This research contributes to computer vision by integrating object detection and tracking algorithms with linear regression offering a solution for enhancing entrance and exit events detection in dynamic environments.