Embedded Vision-Based Assistive Care for the Elderly

As the global population of senior citizens continues to grow, the demand for assistive care for the elderly is also increasing. Since a caregiver at a senior home cannot accompany each patient throughout the day, there is a need for a non-intrusive monitoring system to alert the caretaker in case of a fall. Considering that wearables are not preferred by the elderly, a better way to tackle this problem is to use embedded vision for monitoring. This paper proposes an embedded vision based assistive care system that detects falls and alerts caregivers. It also delivers notifications to carry out the daily tasks like brushing, medications to the elderly while providing reminders to the caretaker who is responsible for multiple seniors living in a home. STM32H747I Discovery board has been used for this purpose along with BCAMS-OMV camera to monitor the seniors. Mobilenetv2 is used to detect falls and it achieves an accuracy of 82.6 % when implemented on the given board.