Master of Science
Date of Defense
Sanjiv Bhatia, Ph.D.
Badri Adhikari, Ph.D. (Chairperson)
Sharlee Climer, Ph.D.
Lav Gupta, Ph.D.
Yoga, a complementary health approach, according to a 2017 National Health Interview Survey by the Center for Disease Control and Prevention (CDC), is a choice of around 14.3% adults in the US. Kapalbhati pranayama, a yoga practice of alternating fast exhales and longer passive inhales, is understood to improve our health. Incorrect and irregular practices, however, can cause injuries and adverse effects. To avoid these undesired effects, it is essential to maintain a pace fit for the practitioner. In the absence of any tools to observe a pace of practice, this work develops a deep learning method that listens to a person’s breathing sound and outputs the pace/counts of kapalbhati breathings in real-time. Our deep learning model, based on our own accuracy definition, is 93% accurate. The model, when packaged as an application, can be used by practitioners with a plan/prescription to observe and adjust their pace and counts per session. In this paper, we elaborately discuss the data collection, cleaning, labeling, deep learning architecture and training, challenges in development, and quantitative evaluation of our results. We plan to release our code publicly along with an application that the public can use. We would like to caution that this work aims to provide an example tool for maximizing benefits and minimizing the adverse effects and injuries from kapalbhati practice, but not to promote yoga practice.
Shrestha, Bikash, "Pranayama Breathing Detection with Deep Learning" (2021). Theses. 369.