Abstract
Being able to provide a low cost but highly accurate fall detection mechanism is decidedly beneficial not only to senior people but also to people of all ages. Most existing approaches are expensive and all subject to the shortfalls of being sensitive to user physique and personal factor. Additionally, most approaches are developed using limited, simulated fall data and often perform poorly in field tests. To resolve these issues, we propose, in this paper, an accurate, crowdsourcing-based, adaptive, fall detection approach using smart devices with built in wireless connection and sensors. We adaptively refine the fall detection algorithm and user grouping for improved accuracy based on the crowdsourced real data. The field tests show that the fall detection accuracy rate can be improved from 68% to 97% with our proposed approach.