9th International Workshop on Computer Vision in Sports (CVsports) at CVPR 2023

 

Congratulations to the Winner of the Best paper award sponsored by Sportlogiq:

SportsPose – A Dynamic 3D sports pose dataset. Christian K Ingwersen, Christian M Mikkelstrup, Janus Nørtoft Jensen, Morten Rieger Hannemose, and Anders Bjorholm Dahl.



Program, June 19. Room: West 214

9.00-9.05 Welcome
9.05-9.50 Invited talk: Claudio Silva – Towards Improved Understanding of Sports and E-Sports Data?
9.50-10.05 Oral 1: SoccerNet-Caption: Dense Video Captioning for Soccer Broadcasts Commentaries
10.05-10.20 Oral 2: Combining Physics and Deep Learning Models to Simulate the Flight of a Golf Ball
10.20-10.50 Coffee break
10.50-11.35 Invited talk: Allan Svejstrup – Combat Sports and Computer Vision, a match made in heaven?
11.35-11.50 Oral 3: A Scale-Invariant Trajectory Simplification Method for Efficient Data Collection in Videos
11.50-12.05 Oral 4: SPARTAN: Self-supervised Spatiotemporal Transformers Approach to Group Activity Recognition
12.05-13.30 Lunch break
13.30-14.15 Invited talk: Elie Zemmour – Using Machine Vision to Create Automatic, Real-time Sports Content in Various Formats
14.15-14.30 Oral 5: SportsPose – A Dynamic 3D sports pose dataset
14.30-15.00 Poster spotlights – 2 minutes per paper
15.00-16.30 Poster session + coffee
16.30-17.15 Results and talks from SoccerNet challenges 2023
17.15-17.30 Best paper award sponsored by Sportlogiq

Papers:


Oral/Poster Title Authors
Oral 1 SoccerNet-Caption: Dense Video Captioning for Soccer Broadcasts Commentaries Hassan Mkhallati (Université Libre de Bruxelles); Anthony Cioppa (University of Liège (ULiège)); Silvio Giancola (KAUST); Bernard Ghanem (KAUST); Marc Van Droogenbroeck (University of Liège)
Oral 2 Combining Physics and Deep Learning Models to Simulate the Flight of a Golf Ball William McNally (Dunlop Sports); Jacob Lambeth (Cleveland Golf); Dustin Brekke (Dunlop Sports Americas)
Oral 3 A Scale-Invariant Trajectory Simplification Method for Efficient Data Collection in Videos Yang Liu (Magicleap); Luiz Gustavo Hafemann (Ubisoft La Forge)
Oral 4 SPARTAN: Self-supervised Spatiotemporal Transformers Approach to Group Activity Recognition Naga Venkata Sai Raviteja Chappa (University of Arkansas); Pha Nguyen (University of Arkansas); Alexander H Nelson (University of Arkansas); Han-Seok Seo (University of Arkansas); Xin Li (West Virginia University); Page D Dobbs (University of Arkansas ); Khoa Luu (University of Arkansas)
Oral 5 SportsPose – A Dynamic 3D sports pose dataset Christian K Ingwersen (TrackMan); Christian M Mikkelstrup (Technical University of Denmark); Janus Nørtoft Jensen (Technical University of Denmark); Morten Rieger Hannemose (Technical University of Denmark); Anders Bjorholm Dahl (Technical University of Denmark)
Poster board 124 VARS: Video Assistant Referee System for Automated Soccer Decision Making from Multiple Views Jan Held (University of Liège); Anthony Cioppa (University of Liège (ULiège)); Silvio Giancola (KAUST); Abdullah J Hamdi (KAUST); Bernard Ghanem (KAUST); Marc Van Droogenbroeck (University of Liège)
Poster board 125 Towards Active Learning for Action Spotting in Association Football Videos Silvio Giancola (KAUST); Anthony Cioppa (University of Liège (ULiège)); Julia Georgieva (Curtin University); Johsan Billingham (FIFA); Andreas Serner (FIFA); Kerry Peek (University of Sydney); Bernard Ghanem (KAUST); Marc Van Droogenbroeck (University of Liège)
Poster board 126 Monocular 3D Human Pose Estimation for Sports Broadcasts using Partial Sports Field Registration Tobias Baumgartner (German Sport University Cologne); Stefanie Klatt (German Sport University Cologne)
Poster board 127 NeighborTrack: Single Object Tracking by Bipartite Matching with Neighbor Tracklets and Its Applications to Sports Yu-Hsi Chen (Institute of Information Science, Academia Sinica, Taiwan); Chien-Yao Wang (Institute of Information Science, Academia Sinica); Cheng-Yun Yang (Purdue University); Hung-Shuo Chang (Institute of Information Science, Academia Sinica ); Youn-Long Lin (National Tsing Hua University); Yung-Yu Chuang (National Taiwan University); Hong-Yuan Mark Liao (Institute of Information Science, Academia Sinica, Taiwan)
Poster board 128 Human Spine Motion Capture using Perforated Kinesiology Tape Hendrik Hachmann (Leibniz Universität Hannover); Bodo Rosenhahn (Leibniz University Hannover)
Online poster One-shot skeleton-based action recognition on strength and conditioning exercises Michael C Deyzel (Stellenbosch University); Rensu P Theart (University of Stellenbosch)
Poster board 129 All Keypoints You Need: Detecting Arbitrary Keypoints on the Body of Triple, High, and Long Jump Athletes Katja Ludwig (University of Augsburg); Julian Lorenz (University of Augsburg); Robin Schön (University of Augsburg); Rainer Lienhart (“Universitat Augsburg, Germany”)
Poster board 130 Visualizing Skiers’ Trajectories in Monocular Videos Matteo Dunnhofer (University of Udine); Luca Sordi (University of Udine); CHRISTIAN MICHELONI (University of Udine, Italy)
Poster board 131 TemPose, a new skeleton-based transformer model designed for fine-grained motion recognition in badminton Magnus A Ibh (IT university of Copenhagen); Stella Graßhof (IT University of Copenhagen); Dan Witzner (IT University of Copenhagen); Pascal Madeleine (Aalborg University)
Poster board 132 Homography based Player Identification in Live Sports Yash Pandya (Amazon); Kaustav Nandy (Amazon ); Shivam Agarwal (Amazon)
Poster board 133 Self-Supervised Video Interaction Classification using Image Representation of Skeleton Data Farzaneh Askari (University of McGill); Ruixi Jiang (McGill University); Zhiwei Li (McGill University); Jiatong Niu (Mcgill); Yuyan Shi (McGill University); James J. Clark (McGill University)

Best paper award:

A best paper award of 1000 CAD is sponsored by Sportlogiq

 

Invited speakers

  • Claudio Silva, Institute Professor at Center for Data Science, New York University – Sports Data Science: Towards Improved Understanding of Sports and E-Sports Data?

While there has always been interest in analyzing sports data, this research area has received significantly more attention in recent years due to both the recognition of the importance of objective statistics and the proliferation of available data. For conventional sports, new technology has enabled the capture of game play at unprecedented levels of detail, including the tracking of positions of all players and game events at all times. And for E-sports, data is starting to pour in at increasing rates. Instead of being starved for data, analysts now have access to volumes of highly accurate gameplay data. This data deluge requires the development of novel visualization and machine learning tools and is leading to major new developments in sports data science. In this talk, we will cover recent developments in this area and the enabling technologies. We will review our early work on the development of the Statcast Baseball Metrics Engine (BME); then introduce new approaches to analyze data directly from videos, including efforts to build new acquisition hardware for sports; and cover new techniques for player valuation. This is joint work with the current and past members of the NYU Sports Analytics team, including Daniel Cervone, Carlos Dietrich, Marcos Lage, Jianzhe Lin, Jorge P. Ono, Guido Petri, Yurii Piadyk, Guande Wu, Peter Xenopoulos, and Shuya Zhao.

  • Allan Svejstrup, CEO at Jabbr.ai – Combat Sports and Computer Vision, a match made in heaven?

Fighting is said to be the oldest sport known to man, and even today sports like Boxing and MMA are among the most popular sports globally. The modern practice of having judges delegated to pick their subjective winner of a fight has been great for safety, but it has also led to fighting being ripe with corruption and controversy. At Jabbr, we’re bringing transparency to both the fighters and to the 500 million fans out there enjoying combat sports. In this talk, we’ll discuss some of the challenges we’ve faced applying Computer Vision and Action-Activity Recognition techniques to fighting, as well as how we think Computer Vision could potentially transform the entire industry from the grass root level and all the way up.

  • Elie Zemmour, Senior computer vision researcher at WSC-sports – Using Machine Vision to Create Automatic, Real-time Sports Content in Various Formats

Watching live sports has always been a popular recreational activity. In recent years, with so much sports content available, most people, and specifically younger generations, don’t have the spare time to keep up with all the live events. That’s why highlights videos and short-form content gained such popularity. Sports fans want to get their favorite content in real-time, and in digestible portions. Our mission at WSC-Sports is to automatically generate highlights videos and any kind of digestible sports content for any sports type and on any platform, allowing our clients to create more personalized content, faster, and in real-time. In this talk, I will share some of the challenges of generating sports content automatically for over 21 sports types, focus on how different approaches of video understanding and trimming help us solve various problems in creating short-form content of many kinds, and discuss how computer vision is changing live sports entirely.

 

Competition – SoccerNet 

  1. Player understanding: player tracking, player re-identification, and jersey number recognition
  2. Field understanding: pitch localization and camera calibration
  3. Broadcast video understanding: action spotting and dense video captioning

Read more about the challenges: https://www.soccer-net.org/challenges/2023

 

Important dates

  • Paper submission deadline: March 6, extended deadline: March 9, 2023 (11.59pm Pacific Time)
  • Notification of acceptance: March 28 March 31, 2023
  • Camera ready deadline: April 6 April 12, 2023
  • Workshop date: June 19, 2023