Analysis of Yoga Position using Machine Learning based Approach

YogaMood Harmony is an innovative research work designed to enhance emotional well-being and physical health through personalized yoga recommendations and real-time pose correction feedback. The system utilizes deep learning and computer vision techniques to provide tailored yoga pose suggestions based on the user’s mood input. Upon selecting a pose, the system guides the user through performing it and offers feedback on posture alignment. The research work encompasses four main modules: mood-based yoga recommendation, yoga pose detection, yoga pose estimation, and pose correction. Mood-based yoga recommendation employs Singular Value Decomposition (SVD) techniques to recommend three yoga poses suitable for the user’s emotional state. Yoga pose detection is achieved using YOLOv3, while yoga pose estimation and key point extraction are performed using the MoveNet model. Pose correction algorithms analyze key points to provide real-time feedback on posture alignment. YogaMood Harmony aims to promote emotional balance, stress relief, and physical wellness by empowering users to engage in yoga practices tailored to their individual needs. Through seamless integration of advanced technologies and traditional wellness practices, this research work fosters a harmonious synergy between mind, body, and technology, ultimately contributing to the user’s holistic well-being. The system was tested using a custom dataset of yoga poses, achieving up to 96.5% accuracy. These results underline its effectiveness in providing safe and adaptive yoga experiences suited to individual needs.