Tools
Tools
Aishwarya Pravin, Ubale (2023) Yoga Pose Multiclass Classification Using Machine Learning Models. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (3MB) | Preview |
Preview |
PDF (Configuration Manual)
Download (1MB) | Preview |
Abstract
An ancient practice - Yoga is substantially popular for its holistic benefits that involves physical postures and mind control practices. In today’s time, leveraging the advancements happening particularly in Machine learning and computer vision field has enhanced various aspects of Yoga. This research paper focuses Machine learning models to classify yoga poses using image recognition techniques. The dataset utilized for the same consists images of popularly known yoga poses. Performing augmentation on this dataset and through application of machine learning models - after performing hyper parameter optimisation, is utilized to perform the classification task. The evaluation of the experiments and performance of the models has been assessed using accuracy, Precision, Recall, F1-Score and discusses the implications of these findings in the context of yoga pose classification. The comparative study highlights the potential improvement in the performance of Support Vector Classifier than other models with accuracy of 87% where as Decision Tree Classifier being the least improved model performing average with 71% accuracy. When contrasted with other existing approaches that do not use augmentation as one of the prepossessing steps and also uses heavy in size deep learning models for yoga pose classification task, the experimental outcome demonstrates the superior accuracy and efficiency of the proposed methodology that utilizes combination of preprocessed - augmented and fine-tuned hyper parameters for machine learning models.
Actions (login required)
 |
View Item |