Human Action Recognition On Hybrid Features Extraction From Silhouettes And Neural Networks

Abstract

Computer vision has several key application domains, one of which is human action recognition. The major objective of this system is to derive an accurate description of human actions and the interconnections between those actions from a data series that had not been seen before and was collected by sensors. The capacity to detect, comprehend, and anticipate complex human actions paves the way for the development of a wide variety of crucial applications, including intelligent surveillance systems, human-computer interfaces, health care, security, and military applications. In this article, we propose a brand-new algorithm for Human Action Recognition that is based on the extraction of hybrid features from silhouettes and the use of neural networks for classification. The recognition rate was 98.9% when we evaluated this algorithm using a worldwide human action dataset. This recognition rate is the only criterion for system evaluation that has been employed up until this point.