Yoga Posture Monitoring Using CNN and Machine Learning

This paper introduces "YoJa," a state-of-the-art posture monitoring system for yoga using Convolutional Neural Networks (CNN) and machine learning techniques to achieve the accuracy of practicing correct yoga postures. With yoga becoming increasingly acceptable and popular across the globe due to the multifaceted benefits of both physical and mental well-being, great care must be taken to avoid injury while at the same time availing oneself of such practices. This system is training the model using a curated dataset of diverse images and videos to give real-time feedback on a user's pose. Results were an accuracy of 97% on the test dataset with YoJa, and thus it has shown the ability to classify yoga postures effectively. User experience is improved further through integration of immediate visual and auditory feedback, making learning faster and performance enhanced. The importance of this study lies in the evidence that technology can actually propagate good yoga culture and safe practices, translated into well-being.