(2025-01-07) Chung, Minjae Lins; Santos, Alyson de Jesus dos; http://lattes.cnpq.br/5998752909180697; Santos, Alyson de Jesus dos; http://lattes.cnpq.br/5998752909180697; Rodrigues, Marlos André Silva; http://lattes.cnpq.br/0650250324042531; Ribeiro, Ewerton Andrey Godinho; http://lattes.cnpq.br/0187470129136467
The aging of the global population brings numerous challenges, both for society as a whole and on a personal/family level. Families face the dilemma of protecting elderly family members without compromising their autonomy and independence during this vulnerable phase of their lives. Along with the physical and psychological changes involved in the natural aging process, domestic accidents, particularly falls, become common events for most of the elderly population. In this context, a fall detection system was presented, employing pose recognition (poseNet) and action recognition (actionNet) technologies, capable of automatically sending alerts to relatives or caregivers. The results demonstrated an accuracy of 82.5% in the proposed model, with near-instantaneous time between alert delivery and receivement. Thus, the system proved to be efficient, meeting the outlined objectives and serving as a foundation for future projects.
