Detecting Irregular Contexts Using Image Recognition and Data Analysis for Elderly Monitoring

In developed countries including Japan, the population is rapidly aging, and dealing with the increasing number of people requiring care is an urgent issue. Assistive technology (AT) can solve problems in medical care, welfare, and home care issues with engineering technology. Previous studies proposed sensors and physical devices for the elderly, but individualized detection and analysis have not yet been realized. Thus, we developed proposed a method for detecting and analyzing situations in the daily life of the elderly. The time-series data was analyzed using an image recognition pre-trained model with lightweight machine learning. The model helps to improve quality of life (QoL) and promote a stable daily life for the elderly at home.