Recognition of Yoga Poses Through an Interactive System with Kinect Device

In daily life, Yoga has become a well-known discipline around the world that keep people in good physical and mental health. As well, gesture recognition is a field of investigation that takes great importance for the self-training of various sports using acquisition techniques such as Kinect device. This research proposes an interactive system capable of recognizing 6 poses for learning Yoga that can track up to 6 people at the same time. It is also integrated with command voices to visualize the instructions and pictures about the poses to be performance for the user. In order to get a strong database for recognition, the system used Adaboost algorithm though the software development kit specially for Kinect. All data was trained by an expert yoga trainer and final database showed above 94.78% as maximum value for poses analyzed in terms of accuracy.