Daily activities and Fall detection
Daily or unexpected activities, such as walking, sitting, running, cycling, standing, falling, fighting, crowd assembling, etc., can be detected using non pervasive sensors that are either remotely positioned, e.g. a camera, or carried by humans, e.g. smart phones, smart watches, smart wristbands. The benefits of automatically recognize these activities can be exploited to great societal support, especially in real-life human centric applications such as eldercare and healthcare. In this project we have attempted to detect falls and distinguish between daily life activities using a smart phone carried inside the pocket of people and a single accelerometer sensor worn at the chest. The overall aim of this study is to evaluate the performance of various accelerometer-based time domain and frequency domain features along with various classifiers for the task of fall and daily activity detection.