| dc.description.abstract | Yoga is one of the popular sports in Indonesia, and it is also one of the sports that involves various movements suitable for all age groups. Yoga encompasses unique movements. The movements in yoga are believed to promote balance and health for the body as well as the mind. However, in practice, mistakes in these movements can result in injuries of varying intensity. Therefore, a system is needed to detect each yoga movement, providing information about the movement's name and ensuring users' confidence by verifying the accuracy of their poses. This research utilizes the You Only Look Once version 5 (YOLOv5) method for real-time detection of yoga movements, consisting of five classes: Bridge Pose, Downward Dog Pose, Shoulder Stand Pose, Tree Pose, and Plank. The dataset comprises 1,000 data points divided into 80% for training data and 20% for validation data. Additionally, 75 data points were used for testing purposes. The utilization of the YOLOv5 method in detecting yoga movements yielded an accuracy of 92%. This outcome indicates that the system developed with this method performs well in detecting yoga movements. | en_US |
