Vision-Enhanced Intruder Detection: An Edge AI Security Framework Utilizing Deep Learning and Alert Systems

Vision-Enhanced Intruder Detection: An Edge AI Security Framework Utilizing Deep Learning and Alert Systems

Authors- Santhosh Pv, Assistant Professor Dr.A.Poongodi

Abstract-Intrusion detection is a critical aspect of modern security systems, especially for homes and restricted-access areas. This paper proposes an intelligent, real-time intruder detection framework that integrates facial recognition with automated alert mechanisms using Edge AI. The system utilizes Local Binary Pattern Histogram (LBPH) for efficient face recognition and OpenCV for image processing, ensuring high accuracy in identifying unauthorized individuals from webcam input. Upon detecting an intruder, the system triggers an audible alarm and dispatches an SMS alert via the Twilio API to notify users immediately. An added password verification module enables secure deactivation of the alarm. The proposed system operates with minimal hardware requirements and leverages open-source tools, making it cost-effective and scalable for deployment in small to medium security infrastructures. Experimental evaluations demonstrate high recognition accuracy and low false-positive rates in various lighting conditions, validating the system’s effectiveness for real-time security applications.

DOI: /10.61463/ijset.vol.13.issue3.106