Development of a Smart Wearable Device for Real-Time Activity Detection

This paper presents a wearable real-time activity detection device which is based on a 9-axis IMU and a microcontroller. By applying SVM algorithm and with high-speed sampled data, the system is expected to give an accurate human activity detection result, which can be used in the healthcare of the elder people or activity monitoring. An idea of introducing a kind of output filter into the classification algorithm has been proved to improve the performance of detection accuracy, by an experiment on hundreds of action samples collected from 8 different people.