Evaluation of Human Activity Recognition and Fall Detection Using Android Phone

Human Activity Recognition (AR) using kinematic sensors is one of the widely used researched area based on Smartphone. Development in sensor networks technology provided birth to the applications that can give intelligent and amicable services based on the AR of people. Although, this technology supports analyzing different activities pattern, empowering applications to identify the activities performed user independently is still a fundamental concern. For improvement quality of life and personal safety, care giving process can be enhanced by introducing the AR, automatic fall detection, and prevention systems. Modern smartphones have different built in sensors like accelerometer, magnetometer, proximity, and gyroscope which can be used for AR as well as fall detection. In this paper, we present an AR and fall detection system which used built in sensors with alarm notification service. We use Signal Magnitude Vector (SMV) algorithm to analyze the fall like events. To overcome the false alarm activation problem, system uses different threshold values to determine the daily life activities like walking, standing, and sitting, that could be wrongly detected as a fall. For assessment, a trial setup is done to acquire sensor´s information of diverse positions.