Intelligent Door Lock System using Ensemble Machine Learning Algorithms

Abstract

The Internet of Things (IoT) is now a critical component of home automation and security, particularly for door lock systems. Face recognition powered by IoT is becoming more popular because it is dependable and simple to use. To prevent algorithmic errors and produce enhanced accuracy than existing systems, our system employs a facial recognition door lock with the integration of three algorithms on a single camera. The system detects unauthorized people near the entrance and notifies the user via a mobile app. For hardware communication, we used a camera module, a Raspberry PI, passive infrared motion sensors, and wireless internet access. The Python and OpenCV libraries are used in the algorithm's development. When the motion sensor detects activity in front of the door, the camera is activated, and the system gathers camera feeds and sends the images to three algorithms for analysis. If any two or all of the algorithm's results match the image in the database, the door unlocks; otherwise, it remains locked. The development of this system aims to enhance accuracy by combining three algorithms and to provide consumers with data about what is happening in front of their home via a smart door lock system.