trAIner - An AI Fitness Coach Solution
Vaibhav Singh, Atharva Patade, Gaurang Pawar, Dhanashree Hadsul
Abstract
Physical activity and exercise can have immediate and long-term health benefits. But when it comes to working out, form is the most important thing. Poor form places undue emphasis on muscles, tendons, and ligaments, leading to strains and sprains. Good form reduces overcompensation and the likelihood of injury. This is one of the many reasons why people work under the guidance of a trainer. Whether one wants to develop an individualised program, or simply feel that they would benefit from the additional accountability, a personal trainer can be a great resource. Online coaching and virtual training have emerged as a staple in the fitness industry. Owing to technological advancements, with a suitable application, one can get constant reminders and a much-needed drive to focus more on fitness and nutrition. Over the years, AI has spread its roots in almost every functional area of business.Pose estimation is among the most popular solutions that AI has to offer; it is used to determine the position and orientation of the human body given an image containing a person. Our goal is to implement an automated fitness coach solution which performs all the tasks of a physical personal trainer. The app obtains users’ motion data by the use of a webcam, and then applies human pose estimation assisted with repetition counting and form evaluation via voice based real time feedback.