Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{188123,
        author = {Mrs.Shilpa Mathur},
        title = {A Machine Learning Approach to Real-Time Yoga Pose Detection and Validation Using MediaPipe and PoseNet},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {12},
        number = {7},
        pages = {1015-1022},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=188123},
        abstract = {Human activity recognition (HAR) plays a vital role in understanding and analyzing human movements for various applications, including fitness tracking, healthcare, and gesture control. In this study, we propose a novel approach for HAR focused on validating the correctness of yoga positions and classifying specific yoga poses using MediaPipe, PoseNet, and Python.The proposed system leverages the capabilities of MediaPipe, an open-source framework for building cross- platform applications to process multimedia data, and PoseNet, a deep learning model for real-time human pose estimation. Through a combination of these tools and Python programming language, our model aims to accurately recognize and classify yoga poses performed by individuals.The workflow begins with capturing input video frames of individuals performing yoga poses. These frames are then processed using MediaPipe to extract pose landmarks, representing key body joints and their spatial relationships. PoseNet is employed to analyze these landmarks and estimate the pose of the individual in real- time.Next, a classification model is trained using machine learning techniques to recognize and classify the specific yoga poses based on the extracted pose features. This model is trained on a labeled dataset comprising various yoga poses, with annotations indicating correct and incorrect executions of each pose.During inference, the trained model predicts whether the observed yoga pose is correct or incorrect, providing valuable feedback to the user. Additionally, the model identifies the name of the specific yoga pose being performed, enabling users to track their progress and adherence to proper form.Experimental results demonstrate the effectiveness of the proposed approach in accurately recognizing yoga poses and validating their correctness in real-world scenarios. The system offers potential applications in yoga training platforms, fitness monitoring apps, and personalized wellness programs, aiding individuals in achieving proper form and maximizing the benefits of their yoga practice.},
        keywords = {Human Activity Recognition,Machine learning,Healthcare,Sports, SecurityPosenet},
        month = {December},
        }