Multi-Modal Posture Recognition System for Healthcare Applications
Siddarth Sreeni, Hari S R, R. Harikrishnan Unnithan, V. Sreejith · 2018
Access to good health care systems varies and health information technology plays a major role in improving the quality of health care and rehabilitation. Use of yoga or similar exercises for such rehabilitation is common. Incorrectly stretching or doing wrong postures can be detrimental to the person’s health. In this paper, we describe a multi-modal approach to analyse, learn and correct postures using recent advancements in machine learning technology. A combination of a 3-D depth map of the user and inertial measurement units with nine Degrees of Freedom (DOF) were used to gather real time postures of the user to measure the accuracy of different postures performed. To accurately measure the postures, a predictive system using machine learning is implemented. Experiments were conducted with volunteers performing multiple postures. The proposed system could accurately identify the postures.