Designing Human Pose Recognition for Yoga Posture in Web-based Augmented Reality

This study presents the design of a system to recognize human yoga postures in a web-based augmented reality (AR). The system provides real-time feedback and personalized guidance through web browsers, enhancing the traditional practice of yoga with advanced technological support. The system employs computer vision techniques to accurately identify yoga poses, ensuring proper form. Leveraging modern web technologies, the solution is accessible and scalable, allowing users to practice yoga from anywhere with an internet connection. The design phase focuses on creating a posture identifier system that calculates angles between body landmarks for pose detection. The implementation incorporates MediaPipe and TensorFlow for efficient tracking and analysis. This study highlights the technical challenges and solutions in implementing pose recognition within a web-based AR framework, demonstrating the potential of this approach to revolutionize digital yoga training. By bridging tradition and innovation, this research empowers practitioners to deepen their practice, improve alignment, and promote holistic wellness.