Smart Surveillance System: An AI-Powered Real-Time Security Monitoring Platform

Large camera deployments generate a continuous stream of video that is difficult for human operators to analyse exhaustively, especially when multiple sites and threat types must be monitored simultaneously. This paper presents an integrated smart surveillance platform that uses deep learning and computer vision modules to transform raw video into structured security events in real time. The system ingests video from heteroge-neous cameras, performs single-stage object detection based on YOLOv8, associates detections across frames using a Deep SORT-style tracker, and augments tracks with face embeddings, pose landmarks, and contextual metadata for downstream analytics. A behaviour analysis engine leverages pose and motion cues to characterise activities, while specialised modules detect violence, weapons, fire and smoke, and license plates. Cross-camera person re-identification supports facility-wide tracking in multi-camera installations. The software architecture separates ingestion, per-ception, analytics, and presentation layers, enabling deployment on commodity GPU hardware with a web dashboard for con-figuration and alert management. Experiments on representative indoor and outdoor scenarios demonstrate that the prototype sustains practical frame rates for multiple streams and can highlight events such as loitering, restricted-zone violations, and visible threats with low end-to-end latency.

© 2026 — Authors retain the copyright of this article. This is an open access article distributed under the Creative Commons Attribution License (CC BY 4.0) , which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.