Human action detection and identification has a wide range of applications, such as video storage and retrieval, intelligent video surveillance and environmental home monitoring, intelligent human–machine interfaces and identity recognition which targets many research topics in computer perception, including human detection in video, human pose estimation, human tracking, and analysis and understanding of time series data. The Human Activity Recognition System (HARS) tries to classify activities based on a series of observations of many subjects' actions and a diversity of environmental variables. The purpose of this research work is to first investigate and compare the accuracy of various HARS for different human actions shown in videos, and then to offer a superior solution. In this paper 6 categories of human activities (jogging, hand waving, walking, running, and handclapping, and boxing) have been recognized with a mean average precision (mAP) of 79.33% at the frame-based and 84.4% at the image-based measurement on the HAR datasets. Extensive experiments on dataset shows that the suggested approach outperforms the current state-of-the-art in action recognition.