A method to evaluate rope-jumping skills based on smartphone and OpenPose

Abstract:
Rope-jumping is a popular fitness sport which can help strengthen lower-limbs muscles and improve body’s equilibrium ability. And it has been included in Chinese middle school entrance examination in recent years. Thus, finding a method to evaluate rope-jumping skills has obvious practical significance. Motion analysis methods in previous studies are mainly based on wearable sensors and special cameras. Restricted to their expensive equipment and operation by trained personnel, these methods are difficult to spread in public. However, as Artificial Intelligence (AI) developed, human keypoints detection technique provides us a new approach to analyze human motion. In this paper, a method based on a smartphone’s built-in camera and open-source human key-points detection project OpenPose to evaluate rope-jumping was proposed. By collecting and analyzing the height curves captured by OpenPose from videos of three different jumping pace modes, we calculated four parameters include jumping pace, flight time, touchdown time and jumping height. Both intraclass correlation coefficient (ICC(3,2)) and multi-correlation coefficient (CMC) between results from our method and standard data group are bigger than 0.9. This indicates that our method has good validity and high reliability.