Development of a Hybrid Defensive Embedded System with Face Recognition
Maria Tsourma, Minas Dasygenis · 2016
The rapid development of Internet of Things and the current capabilities of high performance embedded systems have made them more attractive for the replacement of human personnel in hazardous or tedious duties. One such application is for guarding certain areas, such as the entrance of a warehouse for example. The aim of this work is to design an autonomous embedded security system for surveillance reasons. This system consists of a sentry gun and a camera. The software directing the sentry gun's actions is based on face and motion detection and face recognition. This gun locates, tracks and aims those who enter guarded areas and after verbal warning it fires, when facial recognition fails. This process is automated and works with high accuracy in real time circumstances. It also has the ability to inform its user when an intruder enters the area. The system meets the characteristics of ubiquitous computing as it is autonomous, can be used in any location and can be operated remotely via a web interface. It is hybrid because it supports two different versions of software with different hardware on each one, in order to allow the user to select the proper system version for him. The asymmetric embedded system's first version consists of a single board computer and the second version consists of a mini desktop for the computationally intensive tasks. A microcontroller is being used for high accuracy on the actuators and the system has the capability to use both USB and stereoscopic camera.