Convolutional Neural Network Based Smart Door Lock System

Abstract

The worth of having liberty from any type of hazard, threat, risk, and any other vulnerabilities is known as security. It is the bane of modern-day society. Without security, all will act in a disadvantageous position both psychologically and mentally. A wide variety of locks are used but the most secure one in recent times is the biometric lock system. Moreover, by witnessing the recent situation it is quite evident that the biometric system can become a place of transmission of dangerous viruses as it is seen in the recent COVID-19 pandemic. Due to this reason, the government has banned all biometric systems and in some places, contactless security systems like a lock system using unique identification features like face, retina, and many more are encouraged. To overcome all these difficulties a contactless remote sensing locking system has a far-reaching advantage. In this context, a locking system is designed by collaborating IoT and machine learning techniques. So in this project, a system has been developed which can grant access by simply capturing your face snap. Every human has his/ her facial identification which is unique. Here the extracted feature is used as a passkey and is matched with the database. This project is a part of IoT and Machine Learning. This is developed on a Raspberry Pi and a Pi camera. The system is trained using the Convolutional Neural Network (CNN) approach to recognize the face of the authorized personnel only and to report the unauthorized in case of trespassing.