Wearable camera for fall detection embedded system

Abstract

The fall is one of the major problems that threaten the health of the elderly. According to world statistics, between 28% and 35% of seniors aged over 65 suffer from at least one fall per year. Continuous monitoring and early detection of critical events such as falls, allow for rapid response and immediate medical management. For this, several fall detection devices have been designed by the researchers. In this paper, we propose a fall detection system based on a portable camera. The camera is worn by the user, which allows him to follow his activity wherever he moves, whether indoors or outdoors. In addition, since it is the environment and not the individual that is being observed, the privacy issue is mitigated. The proposed fall detection method is a new approach based on the original Histogram of Oriented Gradient version (HOG) that we combine with the Optical Flow to improve system performance in detecting true falls. Our method has been implemented on a Raspberry embedded on the developed device, which allows us to parallelize the calculations and thus to answer the real time constraint. The results of the 20 tests performed on 09 subjects; show that the falls can be detected with a sensitivity of 95% from the standing position. The combination of optical flow with HOG has improved the specificity, for the rotation scenario, by increasing it by 50% compared to the use of HOG alone. As a result, we have achieved a specificity of 90%. Experimental results show the success of the proposed method.