Development of AI-Based Surveillance Software

Abstract

This paper presents the development of an intelligent video surveillance software designed to perform real- time processing of security camera footage. The system combines motion detection, object detection, and tracking to separate relevant activities from background variations. A web- based interface was designed to support camera management and recording access. Results from a 30-day test with four cameras showed that the software reduced storage usage by approximately 98.4% compared to continuous recording. The system was also able to process video in real-time at 10!FPS per camera. This makes multi-camera monitoring both efficient and user-friendly by recording only when activity (e.g., a person) is detected within user-defined zones and allowing easy event filtering.