Classification of Surya Namaskar Yoga Asanas: A Sequential Combination of Predominant Poses for Physical and Mental Health

Abstract

Surya namaskar is one of the predominant forms of yoga which consists of a predefined sequence of poses to be practiced. By doing Surya namaskar, a performer gets benefited with strong bones, muscles and overall enhanced health condition. This yoga form has to be practiced under the guidance of yoga teacher as improper practicing of postures can lead to health issues. Because of their busy schedule people are not able to practice yoga by going to classes. This leads to necessity of having an automated system for self-learning of Surya namaskar. This research paper proposes an approach for self-learning of Surya namaskar with automated classification of poses by using machine learning and deep learning classifiers. In the first stage, posture images are preprocessed to extract the posture part by using bounding box technique. Different shape-based features, such as, SURF and HOG are obtained in the second phase. Third phase involves classification of Surya namaskar poses using Artificial Neural Networks (ANN) and K-Nearest Neighbor (KNN) classifiers. The results are also supported by using deep learning architecture such as Convolutional Neural Network (CNN). People can use the above model to perform Surya namaskar efficiently.