Automated crowd detection and tracking system using AI-powered drones
Abhinav Raghav, Sarvesh Kumar Swarnakar, Gunjan Mittal Roy, Puneeta Singh, Shantanu Bindewari · 2025
Security in public places is of the utmost importance. Sometimes it becomes humanly impossible to monitor public places throughout the entire day. Any unwanted situation should be timely monitored and reported so that proper action can be taken. One of many unwanted situations may be the formation of dense crowds in public places. We are aiming to build a robust crowd detection system using the images captured by a surveillance drone. The proposed system has three stages. At stage one, the drone will capture the image and detect whether there are any humans in the image or not. If so, it will count the total number of people. For this purpose, we are using neural networks for human detection. If the number of humans exceeds a particular threshold value, then we send the captured image to the cloud for further investigation. On the cloud, we use a self-developed crowd detection algorithm to check whether any crowd is formed or not. If so, we alert a third station by ringing an alarm. The expected outcome of this work is a cutting-edge system that seamlessly integrates drone technology and cloud computing for crowd monitoring. Through the utilization of the YOLOv8 machine learning model, the system will rapidly and accurately detect humans in captured images, providing real-time insights into crowd size and dynamics. When number of humans reaches to certain limits, it communicates automatically with clouds. With rapid, data-driven decision-making capabilities, these insights will enable event planners, security guards, and emergency responders to improve public safety and optimize resource allocation in a variety of situations, from major public gatherings to security operations and disaster response.