Prediction of Volleyball Trajectory Using Skeletal Motions of Setter Player

In this paper, we present a method that predicts the ball trajectory of a volleyball toss 0.3 s before the actual toss by observing the motion of the setter player. We input 3D data of body joints obtained using Kinect into a simple neural network, and 2D data estimated using OpenPose is used for comparison. We created simple neural networks for the two players and tested them. The trajectory of a volleyball toss is properly predicted by the proposed method and the error of the toss trajectory was approximately equal to the size of the ball. This technology can provide a new spectating experience in sports by superimposing the predicted images onto a live broadcast. We also show that this method can be used to identify the important body parts that contribute to the toss prediction. A professional volleyball analyst stated that this technology can be used for analyzing the peculiarities of opponent players.