A Unified Taxonomy and Multimodal Dataset for Events in Invasion Games

Publikation: Beitrag in Buch/Bericht/Sammelwerk/KonferenzbandAufsatz in KonferenzbandForschungPeer-Review

Autoren

  • Henrik Biermann
  • Jonas Theiner
  • Manuel Bassek
  • Dominik Raabe
  • Daniel Memmert
  • Ralph Ewerth

Organisationseinheiten

Externe Organisationen

  • Deutsche Sporthochschule Köln
  • Technische Informationsbibliothek (TIB) Leibniz-Informationszentrum Technik und Naturwissenschaften und Universitätsbibliothek
Forschungs-netzwerk anzeigen

Details

OriginalspracheEnglisch
Titel des SammelwerksMMSports'21: Proceedings of the 4th International Workshop on Multimedia Content Analysis in Sports
Herausgeber/-innenRainer Lienhart, Thomas B. Moeslund, Hideo Saito
Herausgeber (Verlag)Association for Computing Machinery (ACM)
Seiten1-10
Seitenumfang10
ISBN (elektronisch)9781450386708
ISBN (Print)978-1-4503-8670-8
PublikationsstatusVeröffentlicht - 20 Okt. 2021

Abstract

The automatic detection of events in complex sports games like soccer and handball using positional or video data is of large interest in research and industry. One requirement is a fundamental understanding of underlying concepts, i.e., events that occur on the pitch. Previous work often deals only with so-called low-level events based on well-defined rules such as free kicks, free throws, or goals. High-level events, such as passes, are less frequently approached due to a lack of consistent definitions. This introduces a level of ambiguity that necessities careful validation when regarding event annotations. Yet, this validation step is usually neglected as the majority of studies adopt annotations from commercial providers on private datasets of unknown quality and focuses on soccer only. To address these issues, we present (1) a universal taxonomy that covers a wide range of low and high-level events for invasion games and is exemplarily refined to soccer and handball, and (2) release two multi-modal datasets comprising video and positional data with gold-standard annotations to foster research in fine-grained and ball-centered event spotting. Experiments on human performance demonstrate the robustness of the proposed taxonomy, and that disagreements and ambiguities in the annotation increase with the complexity of the event. An I3D model for video classification is adopted for event spotting and reveals the potential for benchmarking. Datasets are available at: https://github.com/mm4spa/eigd

ASJC Scopus Sachgebiete

Zitieren

A Unified Taxonomy and Multimodal Dataset for Events in Invasion Games. / Biermann, Henrik; Theiner, Jonas; Bassek, Manuel et al.
MMSports'21: Proceedings of the 4th International Workshop on Multimedia Content Analysis in Sports. Hrsg. / Rainer Lienhart; Thomas B. Moeslund; Hideo Saito. Association for Computing Machinery (ACM), 2021. S. 1-10.

Publikation: Beitrag in Buch/Bericht/Sammelwerk/KonferenzbandAufsatz in KonferenzbandForschungPeer-Review

Biermann, H, Theiner, J, Bassek, M, Raabe, D, Memmert, D & Ewerth, R 2021, A Unified Taxonomy and Multimodal Dataset for Events in Invasion Games. in R Lienhart, TB Moeslund & H Saito (Hrsg.), MMSports'21: Proceedings of the 4th International Workshop on Multimedia Content Analysis in Sports. Association for Computing Machinery (ACM), S. 1-10. https://doi.org/10.1145/3475722.3482792
Biermann, H., Theiner, J., Bassek, M., Raabe, D., Memmert, D., & Ewerth, R. (2021). A Unified Taxonomy and Multimodal Dataset for Events in Invasion Games. In R. Lienhart, T. B. Moeslund, & H. Saito (Hrsg.), MMSports'21: Proceedings of the 4th International Workshop on Multimedia Content Analysis in Sports (S. 1-10). Association for Computing Machinery (ACM). https://doi.org/10.1145/3475722.3482792
Biermann H, Theiner J, Bassek M, Raabe D, Memmert D, Ewerth R. A Unified Taxonomy and Multimodal Dataset for Events in Invasion Games. in Lienhart R, Moeslund TB, Saito H, Hrsg., MMSports'21: Proceedings of the 4th International Workshop on Multimedia Content Analysis in Sports. Association for Computing Machinery (ACM). 2021. S. 1-10 doi: 10.1145/3475722.3482792
Biermann, Henrik ; Theiner, Jonas ; Bassek, Manuel et al. / A Unified Taxonomy and Multimodal Dataset for Events in Invasion Games. MMSports'21: Proceedings of the 4th International Workshop on Multimedia Content Analysis in Sports. Hrsg. / Rainer Lienhart ; Thomas B. Moeslund ; Hideo Saito. Association for Computing Machinery (ACM), 2021. S. 1-10
Download
@inproceedings{af3a8e4355fb4c56ba14d4e2393952e4,
title = "A Unified Taxonomy and Multimodal Dataset for Events in Invasion Games",
abstract = "The automatic detection of events in complex sports games like soccer and handball using positional or video data is of large interest in research and industry. One requirement is a fundamental understanding of underlying concepts, i.e., events that occur on the pitch. Previous work often deals only with so-called low-level events based on well-defined rules such as free kicks, free throws, or goals. High-level events, such as passes, are less frequently approached due to a lack of consistent definitions. This introduces a level of ambiguity that necessities careful validation when regarding event annotations. Yet, this validation step is usually neglected as the majority of studies adopt annotations from commercial providers on private datasets of unknown quality and focuses on soccer only. To address these issues, we present (1) a universal taxonomy that covers a wide range of low and high-level events for invasion games and is exemplarily refined to soccer and handball, and (2) release two multi-modal datasets comprising video and positional data with gold-standard annotations to foster research in fine-grained and ball-centered event spotting. Experiments on human performance demonstrate the robustness of the proposed taxonomy, and that disagreements and ambiguities in the annotation increase with the complexity of the event. An I3D model for video classification is adopted for event spotting and reveals the potential for benchmarking. Datasets are available at: https://github.com/mm4spa/eigd ",
keywords = "Datasets Event Detection, Events in Sports, Human Performance Analysis",
author = "Henrik Biermann and Jonas Theiner and Manuel Bassek and Dominik Raabe and Daniel Memmert and Ralph Ewerth",
note = "Funding Information: This project has received funding from the German Federal Ministry of Education and Research (BMBF – Bundesministerium f{\"u}r Bildung und Forschung) under 01IS20021A, 01IS20021B, and 01IS20021C. This research was supported by a grant from the German Research Council (DFG, Deutsche Forschungsgemeinschaft) to DM (grant ME 2678/30.1).",
year = "2021",
month = oct,
day = "20",
doi = "10.1145/3475722.3482792",
language = "English",
isbn = "978-1-4503-8670-8",
pages = "1--10",
editor = "Rainer Lienhart and Moeslund, {Thomas B.} and Hideo Saito",
booktitle = "MMSports'21: Proceedings of the 4th International Workshop on Multimedia Content Analysis in Sports",
publisher = "Association for Computing Machinery (ACM)",
address = "United States",

}

Download

TY - GEN

T1 - A Unified Taxonomy and Multimodal Dataset for Events in Invasion Games

AU - Biermann, Henrik

AU - Theiner, Jonas

AU - Bassek, Manuel

AU - Raabe, Dominik

AU - Memmert, Daniel

AU - Ewerth, Ralph

N1 - Funding Information: This project has received funding from the German Federal Ministry of Education and Research (BMBF – Bundesministerium für Bildung und Forschung) under 01IS20021A, 01IS20021B, and 01IS20021C. This research was supported by a grant from the German Research Council (DFG, Deutsche Forschungsgemeinschaft) to DM (grant ME 2678/30.1).

PY - 2021/10/20

Y1 - 2021/10/20

N2 - The automatic detection of events in complex sports games like soccer and handball using positional or video data is of large interest in research and industry. One requirement is a fundamental understanding of underlying concepts, i.e., events that occur on the pitch. Previous work often deals only with so-called low-level events based on well-defined rules such as free kicks, free throws, or goals. High-level events, such as passes, are less frequently approached due to a lack of consistent definitions. This introduces a level of ambiguity that necessities careful validation when regarding event annotations. Yet, this validation step is usually neglected as the majority of studies adopt annotations from commercial providers on private datasets of unknown quality and focuses on soccer only. To address these issues, we present (1) a universal taxonomy that covers a wide range of low and high-level events for invasion games and is exemplarily refined to soccer and handball, and (2) release two multi-modal datasets comprising video and positional data with gold-standard annotations to foster research in fine-grained and ball-centered event spotting. Experiments on human performance demonstrate the robustness of the proposed taxonomy, and that disagreements and ambiguities in the annotation increase with the complexity of the event. An I3D model for video classification is adopted for event spotting and reveals the potential for benchmarking. Datasets are available at: https://github.com/mm4spa/eigd

AB - The automatic detection of events in complex sports games like soccer and handball using positional or video data is of large interest in research and industry. One requirement is a fundamental understanding of underlying concepts, i.e., events that occur on the pitch. Previous work often deals only with so-called low-level events based on well-defined rules such as free kicks, free throws, or goals. High-level events, such as passes, are less frequently approached due to a lack of consistent definitions. This introduces a level of ambiguity that necessities careful validation when regarding event annotations. Yet, this validation step is usually neglected as the majority of studies adopt annotations from commercial providers on private datasets of unknown quality and focuses on soccer only. To address these issues, we present (1) a universal taxonomy that covers a wide range of low and high-level events for invasion games and is exemplarily refined to soccer and handball, and (2) release two multi-modal datasets comprising video and positional data with gold-standard annotations to foster research in fine-grained and ball-centered event spotting. Experiments on human performance demonstrate the robustness of the proposed taxonomy, and that disagreements and ambiguities in the annotation increase with the complexity of the event. An I3D model for video classification is adopted for event spotting and reveals the potential for benchmarking. Datasets are available at: https://github.com/mm4spa/eigd

KW - Datasets Event Detection

KW - Events in Sports

KW - Human Performance Analysis

UR - http://www.scopus.com/inward/record.url?scp=85119045626&partnerID=8YFLogxK

U2 - 10.1145/3475722.3482792

DO - 10.1145/3475722.3482792

M3 - Conference contribution

SN - 978-1-4503-8670-8

SP - 1

EP - 10

BT - MMSports'21: Proceedings of the 4th International Workshop on Multimedia Content Analysis in Sports

A2 - Lienhart, Rainer

A2 - Moeslund, Thomas B.

A2 - Saito, Hideo

PB - Association for Computing Machinery (ACM)

ER -