Media Pipe Based Virtual Fitness Assistance

In this modern world, almost every service is expected in online mode. Fitness Service is one among them. To satisfy the demand, many mobile and web applications were developed and deployed for public access. All the applications give the information on exercises, but do not verify whether user is performing them or not. Along with providing information about various exercises, this project introduces a novel approach of using pre-trained human pose estimation model named “Pose” of Google’s Media Pipe library, integrated with a web application, to capture live video feed from the user device, estimate user pose and perform numerical computations to find angle between joints selected as per exercise chosen, determining whether user is really performing exercise or not.