Understanding the Motion Adaption of Machine Using Long Short – Term Memory Networks for voiceless Virtual Assistant

Abstract

In recent years, the study of computer vision and pattern recognition has seen a significant increase in the popularity of video-based human action recognition as a research topic. Many different fields, including surveillance, robotics, healthcare, video searching, and human-computer interaction, are among its many potential uses. Human action identification in videos faces several difficulties, including crowded backdrops, occlusions, viewpoint fluctuation, execution rate, and camera motion. Over the years, numerous strategies have been put up to deal with the difficulties. For research, three different dataset types—single perspective, multiple viewpoints, and RGB-depth videos—are used. This paper provides an overview of several cutting-edge deep learning-based methods for the recognition of human actions on three different kinds of datasets. Given the increasing.[6]. Here we are using Long short-term Memory networks, as it is a part of Neural Networks. It is more efficient and accurate to create the structures so that the model can understand the method easily.

Keywords: Holistic, OpenCV, LSTM, RNN, Numpy, mediapipe, os.