B-HAR an open-source baseline framework for in-depth study of human activity datasets and workflows

Abstract

This paper discusses the technology of Human Activity Recognition (HAR), which uses machine and deep learning algorithms to monitor activities of various groups of people (e.g. athletes, elderly, children, employers) in order to provide services related to well-being, performance enhancement, safety, and education. However, there is currently no standard way to measure the effectiveness and efficiency of different HAR methodologies, making it difficult to compare them. To address this issue, we propose B-HAR (Baseline-HAR), an open-source framework for creating a standard workflow to evaluate and compare HAR approaches. B-HAR includes popular data processing methods and machine and deep learning models, and allows users to integrate their own models while keeping data pre-processing steps consistent. B-HAR is a program that has been made in Python. We have chosen to use this code and improve it in order to make B-HAR more useful. Therefore in this thesis we will propose a version two of B-HAR.

This paper discusses the technology of Human Activity Recognition (HAR), which uses machine and deep learning algorithms to monitor activities of various groups of people (e.g. athletes, elderly, children, employers) in order to provide services related to well-being, performance enhancement, safety, and education. However, there is currently no standard way to measure the effectiveness and efficiency of different HAR methodologies, making it difficult to compare them. To address this issue, we propose B-HAR (Baseline-HAR), an open-source framework for creating a standard workflow to evaluate and compare HAR approaches. B-HAR includes popular data processing methods and machine and deep learning models, and allows users to integrate their own models while keeping data pre-processing steps consistent. B-HAR is a program that has been made in Python. We have chosen to use this code and improve it in order to make B-HAR more useful. Therefore in this thesis we will propose a version two of B-HAR.