Análise da fusão de sensores para identificação de atividades diárias

Abstract

The detection of daily activities and falls is vital for the prevention of diseases and the treatment of the health of older people. Information on these topics can be obtained by image-based technologies or sensors of various natures. The use of cameras allows for obtaining accurate data on human activities, but it is limited in coverage and, mainly, significantly interferes with the privacy of the elderly. The use of sensors of different natures allows for bringing relevant information from the elderly and maintaining privacy; additionally, wearable sensors allow constant monitoring of the elderly with a larger coverage area than the area provided by a camera. Wearable sensors can be a solution to monitor the daily activities of the elderly without the need for caregivers or changes in the environment. In addition, the recognition of daily activities is a research topic that can benefit different areas, including personal care and the prevention or detection of falls. This work seeks to analyze the fusion of sensors commonly found in wearable devices to understand how the union of specific sensors can help to more assertively identify daily activities, especially those that can help identify falls