CITTA YOGA
(COURSE: Intelligent Interactive Products)
Practising yoga at home can easily be done because little equipment and space are required. The risk is however that there is no instructor to correct you when poses are performed incorrectly, which may result in injuries. In this paper, we propose Citta Yoga, a digital yoga instructor that gives instructions and corrections on your pose and breathing. This is done through the use of a microphone and a linear SVC to check if people do not hold their breath. Furthermore, a webcam is used to check the posture of the user in three yoga poses and alerts him or her when performed incorrectly. For this OpenPose and an SVM classifier were used. The breathing can be recognized with 87.93% accuracy and the pose with 91% accuracy if high-quality images are used. These results can help to create programs that corrects the user to prevent injuries.
Reflection
Looking back, this was the very first project which was entirely a collaborative project in terms of software development. Collaborating was done through GitHub, which I had no prior experience with. The result achieves a high level of complex functionality by combining Weka, Processing and Python.
This was also the second project where I improved my academic writing after the IRIS research project.
It would set the tone for my future work.