PoseRight: An AI Empowered Automatic Posture Estimation and Correction Model

Helmy, Michel Magdy; Yousef, Mary Saad; Shokry, Mina Anis; Gendy, Mina Khalifa; Favez, Mina Lotfy; Galal, Kirollos Aziz; Kawashti, Yomna A.; Hanan Hindy;

Abstract


This paper introduces PoseRight, a novel framework that leverages real-time pose estimation and recognition technology to guide users in performing yoga and gym exercises correctly, thereby reducing the risk of injuries. PoseRight empowers users to achieve effective workouts without the need for gym memberships or personal trainers, addressing the growing demand for accessible fitness solutions. Through extensive experimentation with various deep learning models (CNNs, LSTMs, Transformers) and pose estimation techniques (MediaPipe BlazePose, MoveNet), the proposed framework achieves state-of-the-art performance, surpassing the accuracy of current research. PoseRight demonstrates exceptional test accuracies (99.01% for gym exercises, 96.8% for yoga exercises) while maintaining real-time processing speeds in both domains. This research signifies an advancement towards promoting safe and effective exercise guidance, ultimately fostering healthier lifestyles for users.


Other data

Title PoseRight: An AI Empowered Automatic Posture Estimation and Correction Model
Authors Helmy, Michel Magdy; Yousef, Mary Saad; Shokry, Mina Anis; Gendy, Mina Khalifa; Favez, Mina Lotfy; Galal, Kirollos Aziz; Kawashti, Yomna A.; Hanan Hindy 
Keywords CNN;Fitness;LSTM;Mediapipe Blazepose;MoveNet;Pose Estimation;Transformer
Issue Date 1-Jan-2024
Conference 4th International Mobile Intelligent and Ubiquitous Computing Conference Miucc 2024
ISBN [9798350367775]
DOI 10.1109/MIUCC62295.2024.10783566
Scopus ID 2-s2.0-85216109667

Recommend this item

Similar Items from Core Recommender Database

Google ScholarTM

Check



Items in Ain Shams Scholar are protected by copyright, with all rights reserved, unless otherwise indicated.