Desenvolvimento de um Sistema para o Monitoramento de Acidentes Domésticos de Idosos utilizando Visão Computacional

Abstract

Actually, the biggest problem encountered by the elderly people who live alone or in nursing homes is the risk of falls. This risk is associated with aging and its consequences, such as decreased functional capability. The research consists in the development of an autonomous system to monitoring the presence of domestic accidents in a residence, through computer vision techniques, sensors and a microcontroller. Temperature and gas sensors are used to detect gas leaks and/or fire and an accelerometer detect falls. The microcontroller is responsible for receive data from sensors and informing another device, via a Bluetooth module, of the moment of the fall and/or fire. This second equipment has a camera that confirms the accident and sends the images to the person responsible or family members, through a web server. Fire tests were carried out with the gas and temperature sensors and 30 tests for fall and false fall detection using the accelerometer and computer vision techniques. The results obtained had an accuracy of 93.34% for the detection of true falls and 100% for false falls. In this way, it is intended to reduce the time between the exact moment of the accident and medical care services, in order to reduce the consequences generated by the incident and ensure more safety for family members.