The availability of data related to the player positions on the court is of paramount importance for technical and tactical analyses in basketball. Given the relevance of this problem, the aim of this study was to evaluate a video-based framework for automatic 3D localization of multiple basketball players. The method proved feasible in the context of official games. Indeed, the obtained root-mean-square error (RMSE) associated with the estimation of the players position was 0.18 m, which, in view of the dimensions of one- half basketball court (14 m x 15 m), attests the accuracy of the introduced framework
© Copyright 2016 Proceedings of the 2016 International Association of Computer Science in Sport (IACSS) Conference. Veröffentlicht von FEF/UNICAMP. Alle Rechte vorbehalten.