Rede neural convolucional aplicada à visão computacional para detecção de incêndio

Abstract

In this project, we developed a fire alarm system consisting in: a camera, a Raspberry Pi, a computer vision system and an android application. The Raspberry Pi is considered the leading piece. It is connected directly to a digital camera that processes the received image with an intelligent algorithm. The system algorithm consists of three parts: color filtering, motion detection and self-learning. Evaluating then, if there is a fire principle occurring. The Android application was developed with a user-friendly interface, with the duty to send warnings to the user and provide access to the system camera in real time.