Підхід до розпізнавання студентів за фото для системи моніторингу відвідування занять

Abstract

The main aim of this work is to create an optimal algorithm of person recognition in context of student attendance monitoring system. Recognition in real time of a large number of people requires quit high amount of hardware resources. However, there are number of basic optimization methods that can change it. We will discuss complex measures which involve more than one optimization method. In order to resolve this issue we will develop custom approach especially for our system. Basics of this approach are distribution of photo preprocessing work between IoT devices and keeping balance between possible losses of data and accuracy