Computer Vision-Based Human Activity Classification and Prediction

Abstract

The detection and classification of human activities in computer vision systems are one of the most confounding tasks having Human-Computer Interaction (HCI) features, tracking, security purpose, and health monitoring nonetheless IoT assists in healthcare. Human activities are indwelling for the objective to recognize recurrently occurring actions. Sitting, standing, sleeping, and eating are a few indications of such human actions. It is tenacious to propose a new advanced model keeping in mind recently published work for the classification and prediction of human activities for better comprehension of the activities associated with humans. The model should perform fast and gives highly accurate results in comparison with existing models. Human activity classification also requires sufficient and easy step-by-step solutions for the day-to-day activities of humans. In this respect, this paper attempts to apply an advanced supervised machine-learning model of human activity classification and prediction. In the classification phase, the report demonstrates a precision rate of 97% and a recall rate of 97% accuracy. The overall accuracy of the classification model is 97%, which is reflected in the F1 score. In comparison with the existing research work score of 90%, this proposed model significantly improved the defect determination accuracy.