Sistema de monitoramento seguro por reconhecimento facial baseado em Internet das Coisas

Abstract

The main idea of this project was to develop a security system based on facial recognition with real time video streaming, where it could be accessed from anywhere in the world through the platforms (both web and mobile). The user having an immersive experience while using the platforms and interacting with it’s own home security system was the backbone of this project. To build that experience a set of upcoming Internet Of Things based technologies were put together, giving the user full control of the camera position, which can rotate 180 degrees on the X and Y axis. When logged in, it’s possible to register new people to be recognized by the facial recognition algorithm, which runs on a local Raspberry Pi assembled in a custom 3D printed case. To conserve the user Internet bandwidth and reduce the amount of data traffic between the local systems and the server (which was built upon Firebase), the Edge computing paradigm was used, by processing all the images gathered by the Raspberry Pi camera locally, using the Cloud only to store user data and do its authentication. User data on the system consists in pictures and videos synchronously obtained by commands sent by the user through the platforms, or asynchronously by the facial recognition algorithm, notifying the user in real time about a detected person identity.