A Gym Monitoring System Using YOLOv5

The technique of predicting the location of human body parts and joints in a given video is referred to as articulated body pose estimation. We have proposed a new Gym monitoring system using deep learning concepts, in which working process has been divided into two steps. In first step, actions are tracked using one stage-based YOLOv5 algorithm which follows the rule of the social distance monitoring system. In the second step, recognized the threshold angle by pose estimation algorithms (media-pipe). Now we calculate the angle at joints of human pose, threshold angles may be different for different actions and based on these threshold angles count number of repeated actions. A display screen counts the number of repeated actions. If repeated actions are uncorrected then it will not count.