Intelligent Video Surveillance System

Abstract

Intelligent Video Surveillance Systems (IVSS) have emerged as critical tools for ensuring safety and security in various environments. This paper presents an IVSS capable of real-time detection and analysis of critical events, including falls and vehicle crashes. The proposed system employs state-of-the-art deep learning techniques, specifically the YOLO (You Only Look Once) object detection algorithm, to analyze video streams obtained from diverse sources such as surveil- lance cameras and recorded videos. For fall detection, the system utilizes the aspect ratio of detected persons’ bounding boxes to distinguish between falls and normal activities. Vehicle crashes are identified by detecting cars in each frame and assessing the proximity and intersection of their bounding boxes. The system operates in a continuous loop, processing video frames in real-time and generating alerts when critical events are detected. Experimental results demon- strate the effectiveness and reliability of the proposed IVSS in enhancing safety and security across different environments, including public spaces, workplaces, and transportation systems.