Video-based Yogasan classification for the musculoskeletal disorder using the Cervus trail dependent multiclass support vector machine

Abstract

ABSTRACT Yogasana cures many diseases, one of which is a musculoskeletal disorder that may cause severe pain, bone injury, and so on. Here, Trikonasan, Uttanasan, Vrikshasana, Balasan, Tadasan and Virabhadrasan are the six different Yogasans considered for the treatment of musculoskeletal disorder, where acquiring better accuracy is still a challenge. Hence, this research introduces an efficient Yogasan classification method using the hybrid optimisation-based multiclass Support Vector Machine classifier (Cervus trail-multiSVM), in which the hybrid kernel function is designed using the proposed Cervus-trail optimisation to enhance the classificationaccuracy. The sproposed Cervus-Trail optimisation algorithm is modelled through integrating the rumbling and striving characteristic behaviour of the Cervus with the trailing behaviour of the Maritimus. The effectiveness of the classification relies on defining the kernel fusion factors of the multiSVM classifier with the proposed Cervus trail optimisation. Importantly, in this research, the Yogasan classification is performed using the yogasan videos that could render an effective diagnosis for the musculoskeletal disorder. The performance of the proposed method in terms of accuracy, F-Measure, precision, recall, and sensitivity is found to be 93.08%, 92.94%, 92.91%, 96.33% and 92.97%, respectively.