Efficient Human Activity Recognition Using Machine Learning and Wearable Sensor Data

With the rapid advancement of global development, there is an increasing demand for health monitoring technologies. Human activity recognition and monitoring systems offer a powerful means of identifying daily movement patterns, which helps in understanding human behaviors and provides valuable insights for life management. This paper explores the issue of human motion state recognition using accelerometers and gyroscopes, proposing a human activity recognition system based on a majority decision model that integrates multiple machine learning algorithms. In this study, the majority decision model was compared with an integer programming model, and the accuracy was assessed through a confusion matrix and cross-validation based on a dataset generated from 10 volunteers performing 12 different human activities. The average activity recognition accuracy of the majority decision model can be as high as 91.92%. The results underscore the superior accuracy and efficiency of the majority decision model in human activity state recognition, highlighting its potential for practical applications in health monitoring systems.