Yoga Pose Detection using MobileNetV3+pose comparison and instruction

In the contemporary era, the ancient practice of yoga has witnessed a remarkable resurgence, transcending cultural and geographical boundaries to become a global phenomenon. The ascent of yoga as a global wellness practice underscores the need for accurate yoga pose execution, particularly among novices and those practicing without direct supervision. Leveraging recent advancements in pose detection algorithms, this study proposes a novel framework utilizing MobileNetV3 and Convolutional Neural Networks (CNNs) to detect and classify yoga poses in real-time, coupled with immediate instructional feedback<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">[4]</sup>. This integration aims to enhance practitioners’ experience by ensuring correct pose alignment, thereby making yoga more accessible and maximizing its health benefits. A comprehensive dataset, inclusive of diverse yoga styles and practitioner demographics, underpins the system’s development, ensuring broad applicability and effectiveness.