A Comparison of Human Activity Recognition [HAR] Based on Machine Learning Classifiers

Abstract

Today’s most challenging issue faced by the world is the COVID-19 pandemic, it is essential to track or observe the daily activities of people who living alone or taking individual isolation. So ADL (Activity Daily Living) is an essential thing, especially activity recognition plays a significant role in the medical field. In this paper, human activity recognition (HAR) can be detected with the help of web applications based on the various machine learning classifiers. Along with the activity detection a detailed study is carried to learn about the accurate prediction of each classifier during the training process. Activity detection is done by these processes mainly collecting data, feature extraction, matrix creation, testing, and training process. After that a comparison in the accuracy, precision can be done. The machine learning classifier used to compare is MLP, random forest, SVM, logistic regression, naïve Bayes, and KNN classifier. By using this web app easily detect human activity and monitoring the daily living of a person.