A Deep Learning and IoT-Based Smart Access Control System Using Facial Biometrics andTelegram-Assisted Remote Authentication

Abstract

The security systems in smart homes are changing quickly and incorporating Internet of Things (IoT) technologies and biometric authentication capabilities. Smart home security systems are quickly evolving and becoming more sophisticated, including with the introduction of Internet of Things (IoT) technologies and biometric authentication capabilities. In this paper, a smart door unlocking system is presented which unlocks the door automatically by using face recognition through raspberry Pi 4 and the ESP32 NodeMCU module integrated with Telegram chatbot for the of remote door unlock and notifications of the visitor. Using OpenCV and CNN(Resnet) algorithm, the system detects and recognizes face. In the case of an unknown person appearing on the camera, the photo of that person is sent to the owner via Telegram for identification. While using the chatbot, the user can enter unlocking commands that unlock the door remotely. The experiment conducted revealed an accuracy of around 96-99% for the recognition process, attained with low latency. The envisioned system provides a dependable and inexpensive smart security for residences and workplaces.