SoccerTrack Challenge 2025

Important Dates

• Competition development phase starts: April 14, 2025

• Competition final phase starts: June 13, 2025 (AoE)

• Team registration close:  July 5, 2025 (AoE)

• Competition ends: July 13, 2025, 11:59 p.m. (AOE)

• MMSport 25 Regular workshop papers submission:  July 11, 2025 

• Extended paper submission deadline for winners: August 1, 2025 

• Camera ready version for winners: August 11, 2025 

Rules for SoccerTrack Challenge 2025

A. Registration Required

Please submit a team registration form before you join this competition even if you are a solo challenger. This is required to get prizes. There are no limitations for the team size. However, if you are creating multiple teams within the same lab, be careful not to violate the rules of PRIVATE SHARING.

When you want to merge your team with another team, please contact the SoccerTrack Challenge Admin Team via email.

B. No Private Sharing Outside Teams

Privately sharing code or data outside of teams is not permitted. It’s OK to share code if made available to all participants on the forum.

C. External Dataset

You may use data other than the competition data to develop and test your submission. However, you will ensure the external data is publicly available to everyone without any cost.

Key differences from similar competitions

While competitions such as SoccerNet Game State Reconstruction 2024/2025 and SoccerNet Tracking 2022/2023 focus on tracking from broadcast footage, the SoccerTrack Challenge 2025 utilizes fixed-viewpoint video. This introduces unique challenges, as players appear smaller, and the number of visible players is higher compared to broadcast footage.

For this competition, we will use a dataset consisting of 10 soccer matches captured from a fixed viewpoint. This setup presents a new and more challenging task for participant teams, pushing the boundaries of player tracking and identification in non-broadcast soccer footage such as non-elite sports teams.

Detail

Datasets

The Dataset for this challenge are SoccerTrack v2 datasets (you can download from here). You can use all datasets listed below:

• Video (.mp4 in training,test, and challenge)

• Bounding boxes (training set only; .txt file in MOT format [https://motchallenge.net/instructions/])

If you submit a paper to use the SoccerTrack v2 dataset, please cite: 

@article{scott2025soccertrackv2,

  title={SoccerTrack v2: A Full-Pitch Multi-View Soccer Dataset for Game State Reconstruction}, 

  author={Atom Scott and Ikuma Uchida and Kento Kuroda and Yufi Kim and Keisuke Fujii},

  journal = {2508.01802},

  year    = {2025}

}

Evaluation

The goal of this challenge is accurate tracking of soccer players. Performance of your model will be evaluated based on HOTA (Higher Order Tracking Accuracy) score, which is a holistic and popular score in multi-object tracking (MOT). HOTA is designed to overcome many of the limitations of previous metrics. For example, you can see this web page for understanding.
During the testing and challenge phases, two previously unseen videos will be provided respectively, and the final evaluation will be based on the average HOTA score computed over these two videos.

Submission Format

To submit your results to this competition, you must construct a submission zip file containing two text files, each corresponding to one test video. (submission.zip)

Each file should be named using the format: answer_<video_name>.txt (e.g., answer_123456.txt).

Do not include any folders inside the zip file.
Please note that frame indices in the results should start from 1.

Registration and Submission

Before submitting your prediction, registration is required. You need to register your information to both (1) CodaLab and (2) Google Form. Please follow the procedures below.

• Solo challengers are also welcome! But don't forget to register as a solo team!

• No limitation to team size. (But be careful about violation of the NO PRIVATE SHARING policy when you make multiple teams.)

Steps for Registration

1. [Every Members] Make your codalab accounts. All members should have thier own accounts.

2. [Every Members] Register to this challenge from Participate tab in the codalab page.

3. [Team Leader] Submit the Google Form for team registration.

4. (Wait for a while... Admin team will check and aprove your team.)

5. You got email from codalab about approval and you can submit your prediction!

Codalab page: https://codalab.lisn.upsaclay.fr/competitions/22532

Google Form:  https://forms.gle/Uodjd5yPXU1pMFEA6

Sample code:  https://github.com/open-starlab/stc-2025
Contact:           soccertrack-challenge@g.sp.m.is.nagoya-u.ac.jp
Discord:          https://discord.gg/PnH2MDDaCf

Organizers: 

• Keisuke Fujii (Nagoya University)

• Li Yin  (Nagoya University)

• Qingrui Hu (Nagoya University)

SoccerTrack projects: 

• SoccerTrack (v1)  [Scott et al. 2022, CVSports]

• TeamTrack [Scott et al. 2024, CVSports]

• SoccerTrack (v2: current data): [Scott et al. 2025, arXiv]

Sponsor

• Red Dot Drone Japan