Neural Networks based solution for Door Automation

Abstract

Face Recognition is one of the most common biometric strategies which has gained popularity because of the accuracy and security. This paper presents the implementation of a Convolution Neural Network architecture for door automation. This model is devised to overcome the disadvantages of a traditional door system and other methods such as door automation using Bluetooth, figure prints, passwords, or retinal scans. It allows the authorized people to gain access to the house by face recognition. The proposed system makes use of convolution neural network architectures and RaspberryPi. The ResNet architecture [6] is used to implement face recognition and runs on RaspberryPi. The images of the residents of the house will be used to train the model. If the person is a resident of the house, the face will be recognized and the lock will open, else it will be recognized as a human and an alarm will ring and an email alert consisting of the image of the person in front of the door will be sent to the owner. It has numerous advantages as it is user-friendly especially for senior citizens, lesser maintenance, does not require the residents to carry the keys and reduces the threat of robbery.