SoccerTrack#

News#

(2023/02/08) We have finished fixing the datasets and have also uploaded extra footage with complete annotations! The dataset can be downloaded from Kaggle.
(2022/11/03) After receving reports of erroneous data, we have fixed and reuploaded a majority of SoccerTrack. We are also adding videos with visualized bounding boxes so that you can be sure that the data is good. The visualizations can be found in the viz_results directory under Top-view/Wide-view (see Kaggle). We have created a spreadsheet to keep everyone updated on our progress. Spreadsheet Link

A Dataset and Tracking Algorithm for Soccer with Fish-eye and Drone Videos.

Project page
Paper
Tracking Algorithm (Work In Progress)
Documentation (Work In Progress)

Dataset Details#

-/-	Wide-View Camera	Top-View Camera	GNSS
Device	Z CAM E2-F8	DJI Mavic 3	STATSPORTS APEX 10 Hz
Resolution	8K (7,680 × 4,320 pixel)	4K (3,840 × 2,160 pixesl)	Abs. err. in 20-m run: 0.22 ± 0.20 m
FPS	30	30	10
Player tracking	✅	✅	✅
Ball tracking	✅	✅
Bounding box	✅	✅
Location data	✅	✅	✅
Player ID	✅	✅	✅

All data in SoccerTrack was obtained from 11-vs-11 soccer games between college-aged athletes. Measurements were conducted after we received the approval of Tsukuba university’s ethics committee, and all participants provided signed informed permission. After recording several soccer matches, the videos were semi-automatically annotated based on the GNSS coordinates of each player.

Below are low resolution samples from the soccertrack dataset we plan to release. The actual dataset will contains (drone) and 8K (fisheye) footage!

Drone Video#

https://user-images.githubusercontent.com/22371492/178085041-a8a2de85-bcd3-4c81-8b81-5ca93dbd4336.mp4

Fisheye Video#

https://user-images.githubusercontent.com/22371492/178085027-5d25781d-e3ed-4791-ad14-141b58187dcf.mp4

Note The resolution for the fisheye camera may change after calibration.

Dataset Download#

All the data can be downloaded from Kaggle!

For more details on how to use the dataset, please see the section “Dataset Preparation“.

Install#

pip#

The software can be installed using pip.

pip install soccertrack

You will neeed to install the following dependencies:

pip install torch torchvision pytorch-lightning

To use torch reid, you will need to install the following dependencies:

pip install git+https://github.com/KaiyangZhou/deep-person-reid.git

We recommed using poetry to handle dependencies. So install poetry and run the following command:

poetry install
poetry run pip install torch torchvision pytorch-lightning
poetry run pip install git+https://github.com/KaiyangZhou/deep-person-reid.git

Note The software is currently in development so it will break and change frequently!

Contributing#

See the Contributing Guide for more information.

Papers#

SoccerTrack:
A Dataset and Tracking Algorithm for Soccer with Fish-eye and Drone Videos

Atom Scott*, Ikuma Uchida*, Masaki Onishi, Yoshinari Kameda, Kazuhiro Fukui, Keisuke Fujii

Presented at CVPR Workshop on Computer Vision for Sports (CVSports'22). *Authors contributed equally.

Citation#

@inproceedings{scott2022soccertrack,
  title={SoccerTrack: A Dataset and Tracking Algorithm for Soccer With Fish-Eye and Drone Videos},
  author={Scott, Atom and Uchida, Ikuma and Onishi, Masaki and Kameda, Yoshinari and Fukui, Kazuhiro and Fujii, Keisuke},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={3569--3579},
  year={2022}
}

Acknowledgements#

Part of the tracking module has been adapted from motpy. We would like to thank the authors for their work.