Monitoramento de idosos em ambientes internos por meio da análise sintática de dados visuais

Abstract

Systems for monitoring human activities are used to detect risk situation that may compromise one’s physical integrity or health. These systems support more efficient actions from people responsible for guarding and monitoring the elderly, children, or ill through identification of dangerous activities. This work aims to perform fall recognition of elderly people indoors properly monitored by a camera system. The developed approach uses syntactic descriptions of primitives extracted and characterized from the movements of the elderly. The recognition is given by finite-state automata governed by pre-established grammatical rules. This methodology was applied in several daily situations of elderly people who eventually suffer a fall indoors. The results obtained prove to be satisfactory and point to the feasibility and integration of the methodology in the monitoring systems of elderly activities.