Human activity recognition using wearable sensors

Abstract

The article describes systems of human activity recognition (HAR) that uses wearable sensors by the systematization of types of sensors for human activity recognition and methods of data collection. The model for implementation process of HAR is described and each component of the recognition process is thoroughly analyzed. Methods for identifying human activities using different sensors are proposed and their strengths and weaknesses are identified. The process of finding temporary matches between frames is presented in the diagram with a detailed explanation of each transition. Based on the analysis, a combination of both algorithms and methods is proposed, which will increase the HAR system's efficiency as a whole.