The skill and performance of athletes is more and more represented by numbers. Technical devices are utilized to assist and monitor practices and games. In this regard, the objective of this study was to develop an IMU-based algorithm to recognize jump shots in arbitrary basketball motion sequences. For the extraction, a convolutional neural network was trained on the classification task. The leave-one-subject-out cross-validation of the network showed values of over 0.970 for recall and precision and an area under the curve of 0.995 for the receiver operating characteristic curve. The recognition algorithm represents the first step towards future motion analysis incorporated in a tool which may enable the individual player to self-evaluate their shooting mechanics and improve their shooting performance.
© Copyright 2020 ISBS Proceedings Archive (Michigan). Northern Michigan University. Published by International Society of Biomechanics in Sports. All rights reserved.