A Cost-Effective IoT-Driven Intruder Detection System Using Machine Learning Based Face Recognition

Abstract

The present study explores about Cost-Effective IoT-Driven Intruder Detection System using machine learning-based face recognition. An intruder may enter the premises without the owner's knowledge. A motion detection system using a PIR sensor is employed to detect any movement near the entrance. IoT security solutions to avoid theft require an intelligent security system that is convenient and requires minimum human effort. A USB camera is triggered to capture the intruder's image when motion is detected. This image is then processed using Machine Learning algorithms and OpenCV for face detection and recognition. The Raspberry Pi compares the detected face with a database of approved images. It processes 28 images per second and sends an email notification to the owner, indicating whether the person is authorized or unauthorized. The owner can verify the authentication via the Internet of Things.  The developed low-cost system is fast, highly accurate, gives efficient alerts, and serves as a monitoring system. It is convenient to solve security problems, which will help reduce or stop break-ins.