Meta Models of Yoga gestures by ACCF and SCHF with ML techniques

Abstract

This Yoga is a set of techniques that involve both the body and the mind, and its roots may be traced back to ancient India. Its purpose is to harmonise the body, the intellect, and the soul. The practise of yoga as both an art and a science for maintaining a healthy lifestyle has seen explosive growth in popularity over the past few decades all around the world. People who practised yoga during the lockdowns exhibited lower levels of stress, anxiety, and sadness, according to a number of studies, including the most recent COVID-19 pandemic times. For those interested in leading a more physically and mentally fit life, the ancient practice of yoga comes highly recommended. When practising a yoga asana, it is of the utmost significance to keep the correct posture the entire time. In this study, we make use of transfer learning from human posture estimation models to classify yoga postures. The collected images are used to train a meta model like, Classification via Regression(CVR), and Iterative Classifier optimizer(ICO) after image feature extraction techniques(Auto Color Correlogram Filter and Simple Histogram Filter), which is then applied to the task of determining which yogasanas are being performed. The SCHF+ICO gives an optimal outcome compare with other models.