Human Counting And Tracking In Dense Population

Abstract

With the rapid growth of urban populations and increasing use of surveillance systems, monitoring crowded environments has become an important requirement for ensuring public safety and efficient crowd management. Traditional surveillance systems rely on manual monitoring of CCTV cameras, which is time-consuming and prone to human error. Therefore, there is a need for intelligent systems that can automatically detect and analyze crowd behavior in real time. This project presents an AI-based system for human counting and tracking in dense populations using computer vision and deep learning techniques. The system utilizes the YOLOv8 object detection model to detect people present in video frames captured from surveillance cameras. The model processes each frame, identifies individuals by drawing bounding boxes around them, and calculates the total number of people present in the monitored area. The proposed system extracts frames from the video stream, applies deep learning-based object detection, and displays the detection results visually along with the crowd count. The system is capable of processing video data in real time, making it suitable for applications such as public surveillance, crowd management at events, transportation hubs, and smart city monitoring. The experimental results demonstrate that the system can effectively detect and count people in crowded environments with good accuracy and performance. The proposed approach reduces the need for continuous manual monitoring and provides an efficient automated solution for crowd analysis and surveillance. Future improvements may include integration with smart city infrastructure, advanced crowd behavior analysis, and deployment on cloud-based monitoring platforms for large-scale applications.