Vyayam: Artificial Intelligence based Bicep Curl Workout Tacking System

Abstract

As a famous saying goes “Exercise not only changes our body it changes our mind, attitude, and mood”. Fitness is being a trend today. Everyone wants to be fit, beautiful, and healthy. But during this pandemic, everyone can't hire a trainer or go to a gym. Another option is wearable devices in which everyone can't afford it. This paper proposed an AI Trainer model. The proposed model used by anyone irrespective of their age and health condition. The AI Model uses Human Pose Estimation. It is a popular approach and it determines the position and orientation of the human body. This approach generates key points on the human body and based on that it creates a virtual skeleton in 2D dimension. The input is the live video which is taken from a person's webcam and the output is capturing landmarks or key points on the human body. The AI Trainer specifies the count and time of the settings the person needs to perform. It also specifies mistakes and feedback if any. This paper provides a methodology to use the pose estimation running on the CPU to find the correct points. Based on the points the gestures and other curls (biceps) are calculated. This paper proposes an approach using OpenCV to implement human pose estimation.