Computer vision based intelligent monitoring system for independent living elderly people

Abstract

In recent years, video surveillance cameras act an important role in newlinesociety. The advancements and availability of technologies can be employed newlineto improvise day-to-day life. Human Activity Recognition (HAR) research newlinehave been mainly explored using imagery but is currently evolving to the use newlineof sensors and has the ability to have a positive impact, including individual newlinehealth monitoring and removing the barrier of healthcare. Human activity newlinerecognition has gained importance in recent years due to its applications in newlinevarious fields such as health, security and surveillance, entertainment, and newlineintelligent environments. A significant amount of work has been done on newlinehuman activity recognition and researchers have leveraged different newlineapproaches, such as wearable, object-tagged, and device-free, to recognize newlinehuman activities. Elderly care at home is a matter of great concern if the newlineelderly live alone since unforeseen circumstances might occur that affect their newlinewell-being. Technologies that assist the elderly in independent living are newlineessential for enhancing care in a cost-effective and reliable manner. Elderly newlinecare applications often demand real-time observation of the environment and newline-driven system. As an emerging area ofresearch and development, it is necessary to explore the approaches of the elderly care system in the literature to identify current practices for future research directions. So, a monitoring system is needed to monitor the behavior and give alerts to the care givers. The recently developed Deep newlineLearning (DL) approaches can be employed to design accurate and timely newlineactivity recognition and monitoring systems. With this motivation, this newlineresearch work focuses on finding three different activities of elderly people newlinesuch as non-fall activities, fall events and daily living activities. State of the newlineart method Dynamic Bayesian Network is proposed to monitor most newlineemergency situations and a synthetic dataset is created. Fall events are newlineclassified using the CNN-GRU model. Automatic feature extraction newlinetechniqu