Elderly Fall Detection using Deep Learning Enabled Internet of Healthcare Things
Ahona Ghosh, Azaharuddin Saikh, Sriparna Saha, Indranil Sarkar · 2024
The elderly population is increasing yearly across the world. The public health system is worried about health issues among the elderly, for whom falls are a common cause of post-traumatic problems, sometimes leading to death. The existing research has investigated several methods to detect falls, including wearable sensors, floor-mounted sensors, and depth imaging data. However, implementing these strategies into effect is challenging and costly. In this paper, after going through several existing literature and finding some limitations, an image-processing-based fall detection system has been presented where the significance of fall detection research utilizing a Convolution Neural Network (CNN) is emphasized. For comparative analysis, apart from the traditional model of CNN, different variants have been used to classify the image, where the real-time fall detection using the proposed CNN has shown promising results with 97.78% accuracy. When a fall has been detected, an alarm will be generated to alert the surroundings.