Machine Learning-Based Real-Time CCTV Surveillance for Violence and Crowd Mishap Detection

The machine learning-based real-time surveillance framework for detecting violence and crowd mishaps using CCTV infrastructure. With the increasing frequency of public events and large gatherings, conventional manual monitoring approaches often fail to provide timely alerts for potential threats. The proposed system leverages computer vision, artificial intelligence, and natural language processing to analyze live video streams, estimate crowd density, and detect abnormal behaviors such as panic movements, fights, or sudden gatherings. A web-based dashboard provides real-time visualization and historical data for analytics, enabling authorities to make data-driven decisions for public safety. This paper reviews related literature, explores the methodology and technology stack used, and highlights the potential of Al/ML-driven surveillance in transforming crowd management and safety assurance.