An Accurate People Counting System Based on Object Detection and Tracking

The explosion of technology and data in recent years has completely changed life, tracking and managing data has always been a top concern. People counting is an important challenge in the fields of image processing and video analysis. Furthermore, it has many practical applications, such as traffic management, security control, and determining the number of people in locations. Therefore, this study introduces an accurate people-counting system based on object detection and tracking. Firstly, the system takes individual frames at a time from the incoming video and processes them separately. Subsequently, a CNN model is utilized for person detection on each image, and the YOLOv8 model is selected for developing this system. Result of this model training has very high accuracy with an mAP50 value of 0.878. Besides, the Deep SORT model is used to track objects to help track the movement of people through frames. Based on the information from the above steps, the number of people entering and leaving the area can be counted. Actual test results show that this method is not only effective but also easy to deploy to monitor people entering and leaving the hospital.