Development of a Real-Time Campus Security Surveillance System for UNN Using CCTV and AI

Campus security incidents, including theft, vandalism, and unauthorized access, have continued to challenge the University of Nigeria, Nsukka administration, necessitating a more proactive surveillance approach. This study develops a real-time campus security surveillance system integrating closed-circuit television (CCTV) cameras with an AI-powered motion detection and anomaly alerting module. The system employs OpenCV and a YOLOv8 object detection model to analyze live video feeds and flag suspicious events, sending real-time alerts to a security operations dashboard and a mobile application. The backend infrastructure runs on a Python Flask server with a PostgreSQL event log database. Six IP cameras were configured in a simulated campus corridor environment within UNN for system evaluation. Development followed a prototype-based research design. Evaluation was conducted over a three-week period using scripted security scenario tests covering loitering, trespassing, and object abandonment. Results indicate that the AI detection module achieved a 91 percent true positive rate for anomaly detection with a false positive rate of 8 percent. Alert notification latency averaged 2.3 seconds from event detection to dashboard display. Video storage compression reduced footage archive size by 44 percent compared to unprocessed recordings. The study concludes that AI-integrated CCTV systems significantly enhance campus security responsiveness at UNN. Recommendations include expanding camera coverage and integrating facial recognition with appropriate privacy safeguards.