7th International Workshop on Computer Vision in Sports (CVsports) at CVPR 2021

Workshop program

All talks will be posted as pre-recorded videos, availble throughout the conference week. Links for talks and Q&A meetings are available at the CVPR platform for registered attendees.

Invited talks (videos available throughout the conference week):

  • Pascal Fua, Professor, Computer Vision Laboratory, EPFL.
  • Remo Ziegler, SVP Product Manager, Vizrt.
  • Adrià Arbués Sangüesa, PhD felllow, Universitat Pompeu Fabra.
  • Mehrsan Javan, CTO and co-founder, Sportlogiq.

See titles and abstracts below.

 

Oral presentations (videos available throughout the conference week):

  • 12 peer-reviewed papers, see list of titles and authors below.

 

Live sessions, June 25:

8.40-9.00 (PT) / 11.40-12.00 (EDT) / 17.40-18.00 (CET) : Live Q&A with Pascal Fua and Remo Ziegler

10.00-10.30 (PT) / 13.00-13.30 (EDT) / 19.00-19.30 (CET): Live Q&A with paper authors

11.10-11.30 (PT) / 14.10-14.30 (EDT) / 20.10-20.30 (CET): Live Q&A with Adrià Arbués Sangüesa and Mehrsan Javan

11.30-11.45 (PT) / 14.30-14.45 (EDT) / 20.30-20.45 (CET): Best paper award sponsored by Sportlogiq

Invited speakers

Pascal Fua, Professor, Computer Vision Laboratory, EPFL

Title: Athletic 3D Pose Estimation

Abstract: While supervised body pose estimation is rapidly becoming a mature field, the bottleneck remains the availability of sufficiently large training datasets, which cannot be guaranteed for many kinds of human motions. An effective way to address this is to leverage unsupervised data to learn a low-dimensional representation of poses. Then,  it only takes very little annotated data to train a regressor to predict 3D poses from this representation. In this talk I will discuss several classes of techniques that rely on unsupervised learning objectives that exploit the availability of multi-view footage and of video sequences for this purpose. I will demonstrate the benefits of this approach to track skiers, divers, and ball players.

Slides: Download

 

Adrià Arbués Sangüesa, PhD felllow, Universitat Pompeu Fabra

Title: Bringing Computer Vision to the Court

Abstract: Computer Vision is a really powerful tool to bring meaningful insights and statistics in the context of sport. However, it is sometimes difficult to implement the trained models in real-case scenarios, since there is a missing connection between research and the actual court. Stemming from a basketball baseline, this talk explains the complete pipeline that starts with data gathering, details the creation of metrics and models, and ends up with communication and implementation tips. By building this bridge between analysts and coaches and a smooth communication flow, the performance of teams can be potentially improved

Slides: Download

 

Remo Ziegler, SVP Product Manager, Vizrt

Title: AI opportunities in Sports Broadcast

Abstract: The sports broadcast industry is a very exiting industry to work in. It is very fast paced and everybody tries to come up with new ways to excite the audience on the quest to maximize eyeballs. Additionally, to the fast paced nature of this industry, broadcasting is in the middle of a transition from linear TV to on-demand, customer centric driven content, which has an impact on the nature of production, automation, monetization and many more. In many of those steps understanding and visualizing data in a more automated fashion is key and thus the need for AI imminent. Let us take a glimpse at some of the challenges and opportunities of this field.

 

Mehrsan Javan, CTO and co-founder, Sportlogiq

Title: Sport Analytics: Turn Visual Data Into Insights

Abstract: Sports analytics is about observing, understanding, and describing the game in an intelligent manner. In practice, most of the focus has been on visual perception to take the video data and extract tracking data and game events. However, turning the incomplete data into actionable insights for the clubs has always been a challenge. This talk focuses on the use of broadcast feed for sport analytics, covers the components of a vision system for data acquisition, provides examples of how Sportlogiq captures the data from broadcast videos and turns them into useful insights for the clubs to make better decisions

Slides: Download

Accepted papers

  • Temporally-Aware Feature Pooling for Action Spotting in Soccer Broadcasts. Silvio Giancola (KAUST), Bernard Ghanem (KAUST)
  • Toward Improving The Visual Characterization of Physical Activities with Abstracted Scene Graphs. Amir M Rahimi (HRL Laboratories LLC), Kevin Lee (HRL Laboratories LLC), Amit Agarwal (HRL Laboratories LLC), Hyukseong Kwon (HRL Laboratories), Rajan Bhattacharyya (HRL Laboratories)
  • SoccerNet-v2: A Dataset and Benchmarks for Holistic Understanding of Broadcast Soccer Videos. Adrien Deliege (University of Liege), Anthony Cioppa (University of Liège (ULiège)), Silvio Giancola (KAUST), Meisam Jamshidi Seikavandi (Aalborg University), Jacob Velling Dueholm (Aalborg University), Kamal Nasrollahi (Aalborg University, Denmark), Bernard Ghanem (KAUST), Thomas B. Moeslund (Aalborg University), Marc Van Droogenbroeck (University of Liege)
  • Detecting and Matching Related Objects with One Proposal Multiple Predictions. Yang Liu (SPORTLOGiQ), Luiz Gustavo Hafemann (Sportlogiq Inc), Michael Jamieson (Sportloqiq), Mehrsan Javan (SPORTLOGiQ Inc., McGill University)
  • Contrastive Learning for Sports Video: Unsupervised Player Classification. Maria Koshkina (York University), Hemanth Pidaparthy (York University), James Elder (York University)
  • Camera Calibration and Player Localization in SoccerNet-v2 and Investigation of their Representations for Action Spotting. Anthony Cioppa (University of Liège (ULiège)), Adrien Deliege (University of Liege), Floriane Magera (EVS Broadcast Equipment), Silvio Giancola (KAUST), Olivier Barnich (EVS Broadcast Equipment), Bernard Ghanem (KAUST), Marc Van Droogenbroeck (University of Liege)
  • DeepDarts: Modeling Keypoints as Objects for Automatic Scorekeeping in Darts using a Single Camera. William McNally (University of Waterloo), Pascale B Walters (University of Waterloo), Kanav Vats (University of Waterloo), Alexander Wong (University of Waterloo), John McPhee (University of Waterloo)
  • LoL-V2T: Large-Scale Esports Video Description Dataset. Tsunehiko Tanaka (Waseda University), Edgar Simo-Serra (Waseda University)
  • Puck localization and multi-task event recognition in broadcast hockey videos. Kanav Vats (University of Waterloo), Mehrnaz Fani (University of Waterloo), David A Clausi (University of Waterloo), John Zelek (University of Waterloo)
  • Table Tennis Stroke Recognition Using Two-Dimensional Human Pose Estimation. Kaustubh Milind Kulkarni (Independent), Sucheth Shenoy (Independent)
  • Automatic Play Segmentation of Hockey Videos. Hemanth Pidaparthy (York University), Michael Dowling (Queens University), James Elder (York University)
  • Automated Tackle Injury Risk Assessment in Contact-Based Sports – A Rugby Union Example. Zubair Martin (University of Cape Town ), Sharief Hendricks (University of Cape Town), Amir Patel (University of Cape Town)

Motivation

Sports is said to be the social glue of society. It allows people to interact irrespective of their social status, age etc. With the rise of the mass media, a significant quantity of resources has been channeled into sports in order to improve understanding, performance and presentation. For example, areas like performance assessment, which were previously mainly of interest to coaches and sports scientists are now finding applications in broadcast and other media, driven by the increasing use of on-line sports viewing which provides a way of making all sorts of performance statistics available to viewers. Computer vision has recently started to play an important role in sports as seen in for example football where computer vision-based graphics in real-time enhances different aspects of the game. Computer vision algorithms have a huge potential in many aspects of sports ranging from automatic annotation of broadcast footage, through to better understand of sport injuries, and enhanced viewing. So far, the use of computer vision in sports has been scattered between different disciplines.

 

Call for papers

The ambition of this workshop is to bring together practitioners and researchers from different disciplines to share ideas and methods on current and future use of computer vision in sports. To this end we welcome computer vision-based research contributions as well as best-practice contributions focusing on the following (and similar) topics:

  • estimation of position and motion of cameras and participants in sports
  • tracking people and objects in sports
  • activity recognition in sports
  • event detection in sports
  • spectator monitoring
  • annotation and indexing in sports
  • graphical effects in sports
  • analysis of injuries in sports
  • performance assessment in sports
  • alternative sensing in sports (beyond the visible spectrum)
  • tactics analysis in sports
  • automatic narration and captioning in sports
  • training assistance in sports
  • augmented/virtual reality in sports

 

Important dates

  • Submission deadline: March 7 March 14, 2021 (11.59pm Pacific Time)
  • Notification of acceptance: April 1 April 7, 2021
  • Camera ready deadline: April 18, 2021
  • Workshop date: June 25, 2021

 

Submission instructions

Policies and guidelines (same as for CVPR): Link

Submission: Link

Accepted papers will be published in the CVPR workshop proceedings on IEEE Xplore and as open access on http://openaccess.thecvf.com/.

 

Best paper award

1000$ sponsored by Sportlogiq.

 

Program committee:

Adrien Deliege, University of Liege, Belgium
Anthony Cioppa, University of Liege, Belgium
Bastian Wandt, Leibniz University Hannover, Germany
Dan Gutfreund, IBM Research
Dan Mikami, NTT, Japan
Hideo Saito, Keio University, Japan
Jean-Yves Guillemaut, University of Surrey, UK
Jesse Davis, KU Leuven, Belgium
Jianhui Chen, University of British Columbia, Canada
Jonas Theiner, Universität Hannover, Germany
Li Cheng, University of Alberta, Canada
Marc Van Droogenbroeck, University of Liege, Belgium
Michael Jamieson, Sportlogiq, USA
Moritz Einfalt, University of Augsburg, Germany
Nicola Mosca, CNR ISSIA, Italy
Rainer Lienhart, Universitat Augsburg, Germany
Rajkumar Theagarajan, University of California, Riverside, USA
Sergio Escalera, Computer Vision Center (UAB) & University of Barcelona, Spain
Silvio Giancola, KAUST, Saudi Arabia
Simon Denman, Queensland University of Technology, Australia
Steven Schwarcz, University of Maryland College Park, USA

 

Organizers

Rikke Gade, Aalborg University, Denmark
Thomas Moeslund, Aalborg University, Denmark
Graham Thomas, BBC, UK
Adrian Hilton, University of Surrey, UK
Jim Little, University of British Columbia, Canada
Michele Merler, IBM Research, USA

 

Previous editions of CVsports:

 

Related publications:

 

Overall Meeting Sponsors  CVF logo