Smart doorbell for home security using Raspberry Pi

Abstract

In recent years, the crimes targeting household properties has grown considerably high. Therefore, it’s necessary to come up with a proper solution to secure a household and to let house owners aware about the surrounding activities. The proposed system is a hardware device that combined with technologies like OpenCV deep learning based facial recognition and WebRTC. Further, PhoneGap is used to develop the application for hand-held devices which will interact with the client (house owner). To implement one-way video conferencing between the user (house owner) and visitor, WebRTC with Socket.io will be used. Apart from that, WebRTC with Socket.io combined technologies are responsible for streaming live feed of household environment. Motion detection and video capturing are done using PIR sensor, camera module, speaker and microphone. The system will initiate by turning on the camera when either a user presses on the doorbell ‘bell’ push button or PIR sensor detects motion. When the visitor presses the doorbell ‘call’ push button, visitor is connected to an one-way voice conference call with the home owner (House owner can see and hear the voice of the visitor through a hand-held device and visitor can only hear the voice of the owner through doorbell device.). Home owners can record a pre-defined message for known users as well. The camera module will perform facial detection by resolving a classifier in OpenCV. Afterwards, an image will be captured with a brief video. Facial information is saved in facial repository and video files are stored in the cloud. During the facial recognition if a match is found, the house owner will be notified with visitor details. For unknown users, house owner will be prompted to identify. The system can be improved and customized as a security system that can be installed in server rooms and also it can be enhanced to use as a student attendance recorder module.