.

ISSN 2063-5346
For urgent queries please contact : +918130348310

A Mediapipe Blaze pose model to evaluate Yoga posture with immediate feedback

Main Article Content

Dr. L. Jaba Sheela, M.Arpana , P Ramesh Reddy, Dr. G. Sudhagar
» doi: 10.31838/ECB/2023.12.SI3.411

Abstract

The computerized Self-training Artificial intelligence structures for sports and fitness can advance participant performance and thwart damages. Concocting an interactive web application that uses the webcam to recognize the user's yoga and exercise poses, and estimates each pose to assist the user practicing those postures and tracks successful shots at various stances. Our approach seeks to recognize the yoga asanas based on the data attained by Data Collection from an open-source dataset. The sensed critical points are conceded to our prototype where neural networks find patterns and Sequential model-CNN analyze their evolution over time. Mediapipe framework is used in detecting the landmarks of the human body to retrieve the stick figure of the user for estimating their yoga poses and predict its accuracy by passing it to the CNN model. Finally, the system contains a meditation coach which is programmed to give commands at standard intervals and subsequently, the user is ushered by a voice that instructs them to maintain their breathing pace thereby creating a soothing environment.

Article Details