An Efficient System for Human activity recognition and monitoring for elderly people using Machine Learning

Abstract

"Human activity recognition" is essential to the success of numerous real-world applications, such as the detection of abnormal behavior and elderly health surveillance (HAR). As part of our current research, we developed and tested a machine learning (ML) model that can accurately identify an activity based on raw data collected by wearable technology. The "WISDM Smartphone and Smartwatch Activity and Biometrics Database" was used in this particular study. Feature selection can aid in reducing a dataset's overall dimension and developing AI models. A confusion matrix for the model was created while the values for precision and recall were determined. We have used a variety of machine classification techniques, including KNN, SVM, DT, and NB, to evaluate human actions to evaluate the multidimensional data we have chosen to collect. The experiments' results showed that the decision tree algorithm was able to produce such remarkable results with a high degree of precision.