Nighttime in-home action monitoring for eldercare

In this work, we develop a system to automatically monitoring actions of elderly people at home for safety enhancement and health monitoring. We use an Infrared camera embedded in a living environment to capture images. We study the characteristics of different clothing in Infrared images and develop an efficient silhouette extraction method for Infrared (IR) images using spatio-temporal filtering. We recognize human action using supervised learning methods. Our experimental results demonstrate that our proposed method is efficient.