TY - JOUR

T1 - Toward Automatically Labeling Situations in Soccer

AU - Fassmeyer, Dennis

AU - Anzer, Gabriel

AU - Bauer, Pascal

AU - Brefeld, Ulf

N1 - This publication was funded by the Open Access Publication Fund of Leuphana University Lüneburg. We would like to thank the German Football Association (DFB) for providing the data for this study.

PY - 2021/11/3

Y1 - 2021/11/3

N2 - We study the automatic annotation of situations in soccer games. At first sight, this translates nicely into a standard supervised learning problem. However, in a fully supervised setting, predictive accuracies are supposed to correlate positively with the amount of labeled situations: more labeled training data simply promise better performance. Unfortunately, non-trivially annotated situations in soccer games are scarce, expensive and almost always require human experts; a fully supervised approach appears infeasible. Hence, we split the problem into two parts and learn (i) a meaningful feature representation using variational autoencoders on unlabeled data at large scales and (ii) a large-margin classifier acting in this feature space but utilize only a few (manually) annotated examples of the situation of interest. We propose four different architectures of the variational autoencoder and empirically study the detection of corner kicks, crosses and counterattacks. We observe high predictive accuracies above 90% AUC irrespectively of the task.

AB - We study the automatic annotation of situations in soccer games. At first sight, this translates nicely into a standard supervised learning problem. However, in a fully supervised setting, predictive accuracies are supposed to correlate positively with the amount of labeled situations: more labeled training data simply promise better performance. Unfortunately, non-trivially annotated situations in soccer games are scarce, expensive and almost always require human experts; a fully supervised approach appears infeasible. Hence, we split the problem into two parts and learn (i) a meaningful feature representation using variational autoencoders on unlabeled data at large scales and (ii) a large-margin classifier acting in this feature space but utilize only a few (manually) annotated examples of the situation of interest. We propose four different architectures of the variational autoencoder and empirically study the detection of corner kicks, crosses and counterattacks. We observe high predictive accuracies above 90% AUC irrespectively of the task.

KW - labeling situations

KW - soccer

KW - sports analytics

KW - tracking data

KW - variational autoencoders

KW - Informatics

KW - Business informatics

UR - http://www.scopus.com/inward/record.url?scp=85119428473&partnerID=8YFLogxK

U2 - 10.3389/fspor.2021.725431

DO - 10.3389/fspor.2021.725431

M3 - Journal articles

C2 - 34805978

AN - SCOPUS:85119428473

VL - 3

JO - Frontiers in Sports and Active Living

JF - Frontiers in Sports and Active Living

SN - 2642-9367

M1 - 725431

ER -