Monitoramento de ambientes e controle de frequência acadêmica através de visão computacional

Abstract

The evolution of computer vision and the increase of available processing power enabled a paradigm shift that can change the use of computers from an intrusive to a non-intrusive approach. The idea of a non-intrusive system is to allow users to interact with computers without the use of peripherals or direct contact with a machine. Computer vision can be used to achieve this paradigm shift. This work presents the implementation and validation of a software able to monitor an environment, acting as an attendance checker for use in the classroom. The application uses a camera attached to a computer capturing a sequence of digital images. By analyzing those images, the system is able to detect and recognize human faces captured by the camera, registering the date and the time for each person found. A set of tests were designed and performed to measure the recognition time and how it was affected by different factors as the environment illumination or the amount of training images used to add a new person to the application database.