A smart door prototype with a face recognition capability

Abstract

This research aimed to integrate a face recognition capability in a smart door  prototype. By using a camera - based face recognition, the house owner does  not need to make physical contact to open the door. Avoid physical contact  is  important  due  to  the  coro navirus  disease  2019  (COVID19)  pandemic.  Raspberry Pi 3B  was  used  as  the  main  controller,  while  a  servo motor  was  utilized as a locking door actuator. The program was developed using  Node - RED ,  Blynk ,  and  message  queue  telemetry  transport  ( MQTT ) platforms  which are very powerful for developing  internet of  things (IoT) devices. All  of  the  programs  were  coded  using  Python.  Haar  cascade  and  local  binary  pattern histogram methods were implemented on the face recognition stage. Google Assistant int egration was done by using  D ialogflow  and Firebase as  Google  Cloud  services.  Integration  of  face  recognition  and  the  smart  door  was  successful.  The  smart  door  was  unlocked  if  faces  were  recognized  (average threshold=60%). If a face was not recognized, an e mail notification  containing  a  face  image  is  sent  to  the  house  owner.  The  Google  Assistant  could  handle  user  requests  successfully  with  a  success  rate  of  92.8%  from  147 trials .