3D Human Pose Estimation for Yoga Training in Deep Learning

Abstract

Through diverse asana, meditation practices, and breathing techniques, yoga asana has
traditionally been utilized to bring the body and mind together. As a result of the massive
increase in stress in our lives as a result of our modern lifestyle, yoga has gained in popularity
around the world. For learning asana, self-study and going to a professional instructor are
both viable options. The majority of people prefer to learn it through videos and tutorials on
YouTube, but they are confused if they are doing it right. As a result, reading the lessons in
this report that we developed for our project will assist students in overcoming the difficulties
they encounter when performing yoga positions.
Our study report establishes the framework for such a system by employing a range of
algorithms to effectively classify yoga poses on yoga pose images and from users' webcams,
as well as to signal the accuracy of their stance.