Human-Activity-Recognition

Human Activity Recognition (HAR) is a broad spectrum that has the potential to benefit today’s fast pacing world by developing assistive technologies in order to aid chronically ill people, provide insights into real-time human activities, monitoring elderly with special healthcare needs and for overall improved wellbeing of the individuals. It is quite a tedious and challenging task to attain high accuracy in this domain because human activity is highly diverse and very complex. Hence, such a large-scale data is extracted using sensors embedded in smartphones and smartwatches (e.g., accelerometer, gyroscope, pulsometer etc.). A comparative study on Human Activity Recognition (HAR) dataset based on four machine learning techniques, namely, K-Nearest Neighbors (KNN), Random Forest (RF), Artificial Neural Network (ANN) and Support Vector Machines (SVM), has beencarry out in this paper. Out of these four models, SVM showed the highest accuracy of 96.5%.