This summer project was done under the supervision of Professor Dr. Veena Bansal, Department of Industrial Management and Technology, IIT Kanpur.
Yoga is a group of physical, mental and spiritual practices that aims to control and still the mind. Studies have found that yoga practice lowers the resting heart rate, increases endurance, and can improve our maximum uptake of oxygen during exercise. Today there are enough medical and scientific evidences to show that the very fundamentals of brain activity, chemistry, and genetic content can be changed by practicing Yoga. Yoga practices increase the standard of life.
So, Yoga as a system needs much more involvement than any other thing that we do. We know that many people around us are getting ill, catching various diseases, eventually degrading their quality of life. They can indeed go to the doctor and consult them, but sometimes excessive intake of medicines is somewhat harmful to the body. That is where Yoga comes into the picture. We then came up with an innovative idea: why not create a Yoga Recommendation system for people suffering from some disease or having some health related problem.
We have used Deep Learning and NLP techniques to create a model that would take four words from the user as input and suggest to him top-seven asanas that match the benefit words. We first began by formulating our dataset. We created a dataset of about 329 asanas, including the benefits, contraindications, description, and procedure for each asana. Our biggest challenge was to search for correct and accurate information about each asana, so we looked up various Yoga books and collected our information. We then implemented the concept of word vectorization which is a classic approach of converting input data from text into vectors of real numbers. This technique is widely used in Natural Language Processing. A neural network will take the embeddings of the word as input and will predict the asanas in decreasing order of their probability of matching. Our model is unique as no such implementation like this has happened before. We aim to provide the user with the best asana recommendation according to his input which would be very handy for his health-related problem.
I am profoundly grateful to Dr. Veena Bansal, Department of Industrial and Management Engineering for her expert guidance and constant encouragement throughout the Summer Term. I would also like to thank her for giving me and my project partner this golden opportunity to do this project on the topic Yoga Asana Recommendation Model.
It helped me do much research in the related Machine Learning and Deep learning field and also taught me the importance of Yoga and exercise in one's life.
At last, I wanted to thank my project partner Ms. Richa Prakash Sachan to work with me on the assigned project and all other group members who helped me directly or indirectly during this course of work.