A study on Human Activity Recognition Using Gyroscope, Accelerometer, Temperature and Humidity data

An adequate variety of sensors employed in smartphones, being accessible, a great prospect concerning the research sphere towards data mining and machine learning has been discerned. Regarding the discerned arena of data mining and machine learning, a total of ten human activities encompassing sitting, walking, jogging, lying, walking upstairs and downstairs, cycling, standing, squatting in a toilet and fallen down have been subjected to be recognized in this paper. Employing a total of four sensors comprising Accelerometer sensor, Gyroscope sensor, Humidity sensor and Temperature sensor, we collected labeled data in connection with daily activities of three subjects and epitomized in 1Hz frequency. Afterwards, a model respecting the forecasting of activity recognition was brought out incorporating our training dataset. We consider the recognition of pristine activities as well as higher accuracy as the quirky attribute of our research. Through this research, heterogeneous applications regarding analysis of disease associated with physical action, overseeing of physical action and elderly care can be accomplished.

I. Preface and Associated Researchs

Smartphone sensors lead to a divergent genre of data respecting the disparity of the motion of the smartphone. This sort of disparity in the sensors data can be employed in terms of the identification of physical motions. The physical motions being discerned can be employed in terms of the generation of health reports and analysis of the diseases associated with human activity.