Módulo auxiliar de monitoramento veicular através de câmeras e deep learning

Abstract

The number of traffic fatalities in Brazil is increasing gradually each year, a part of this increase is due to the disrespect and the distraction of the driver in traffic. Currently, urban centers use radars to monitor vehicle speed, and statistical data prove that the use of such equipment has dramatically reduced the number of accidents and consequently the number of deaths, today with the current technology it is already possible to employ radars directly in vehicles, but little if there were of them, the use of the current technology for the development of these radars can serve as an aid to the driver, not an equipment to penalize, with the use of cameras in the vehicle can create equipment that helps the driver , transforming visual information captured by the cameras into sound information, sent to the driver, with machine learning techniques, more specifically in the area of deep learning, it is already possible to develop algorithms that can identify a certain action of the driver by his gestures and thus execute another action to help him, the study to be presented aim is to develop an algorithm that through cameras can detect the direction to which the driver is looking and thus with cameras distributed by the vehicle can automatically monitor regions in which the driver is not observing and in case of a certain risk situation, this algorithm shall issue a sound notification to the driver.