Crowd Behaviour Analysis System

This Project titled "Crowd Behaviour Analysis Using AI" introduces a real-time surveillance framework aimed at identifying weapons such as firearms and knives, as well as placards, sticks, and aggressive behaviours within densely populated settings. Utilizing the YOLOv8 algorithm and featuring a web interface developed on Streamlit, this system is capable of processing images, video footage, and live feeds from cameras to improve public safety measures. It provides a user-friendly solution by delegating computational processes to a server, which allows it to function effectively on lower-end devices that have internet access. The application marks detected items with bounding boxes and corresponding confidence scores, facilitates the download of results, and demonstrates robust performance across a variety of environmental conditions. Tailored for urban safety, event oversight, and security applications, this system confronts issues such as occlusion and fluctuating lighting conditions, with plans for future upgrades to enhance the detection of additional threats.