APRENDIZADO DE MÁQUINA APLICADO PARA AUXÍLIO AO MOTORISTA UTILIZANDO RASPBERRY PI

Brazil has the fifth highest death toll in the planet. Generally accidents are caused by human failure, involving inattention and disrespect to the law. In order to help the driver to act in a preventive and responsible manner, computer systems can establish ways to issue alerts when recognizing situations of risk to the safety in the traffic. The challenge of this work was to perform the detection and recognition of some traffic signals considered necessary for road safety. This work aimed at the development of an embedded system of assistance to the driver based on computer vision and machine learning. The function of the system is to recognize dangerous situations and alert the driver to the signals found on the tracks (maximum permissible speed, stop, preference and bearing tracks). We used a Raspberry Pi 3 and a camera of 5 megapixels to be the embedded hardware. The work aimed the development of algorithms that perform the task of assisting human perception in guiding vehicles, with execution in low-processing hardware in real time.