Player Tracking in Bird-Eye Diagram Through Basketball Video

The title of our project is Player Tracking in Bird-Eye Diagram Through Basketball Video

In the sport of basketball, analyzing gameplay and opposing team strategies can be challenging due to tilted camera angles in footage. Instant playbacks offer coaches a better perspective, enabling them to assess match dynamics, make real-time adaptations, and analyze games more effectively. This enhances decision-making, team success, and fan engagement. YOLOv5, a popular multi-object detection model, along with robust court mapping with ResNet18, plays a crucial role in accurately identifying players and positioning them on the detected court.

Original	Board

Related Paper:

• Maglo, Adrien, Astrid Orcesi, and Quoc-Cuong Pham. "KaliCalib: A Framework for Basketball Court Registration." Proceedings of the 5th International ACM Workshop on Multimedia Content Analysis in Sports. 2022.[PDF]

---

📚 1. Installation

Create a virtual environment:

virtualenv venv
source venv/bin/activate

Install the dependancies:

pip install -r requirements.txt

⚙️ 2. Run the program

Default run

python bird_eye_video.py

Add arguments

Change model

python bird_eye_video.py --modelPath models/model_challenge.pth

Change input video

python bird_eye_video.py --inputPath input/demo.mp4

Change weight

python bird_eye_video.py --weightPath yolov5/runs/train/exp2/weights/best.pt

Change floortexture

python bird_eye_video.py --floorTexturePath input/floor_texture/concrete.jpg

🛠️ 3. Dataset

The dataset used in this study includes bounding box locations of players and referees, as well as court vertex locations. Due to challenges in collecting data from actual NBA games, data was gathered from NBA 2k19, ensuring controlled scenes with visible sidelines. This dataset improves court detection, player tracking, and model generalization.

The self-annotated dataset is opensourced in Roboflow

🌟 4. Results

NBA 2K19 data

Original	Homography	Board

Real Game data

Original	Homography	Board

🎬 5. Documentation

• Project Document: [PDF]

• Project Powerpoint: [PowerPoint]

🏅 6. Reference

[1] P. Kaarthick, “An automated player detection and tracking in basketball game,” Computers, Materials Continua, vol. 58, pp. 625–639, 01 2019.

[2] D. Farin, S. Krabbe, P. With, and W. Effelsberg, “Robust camera calibration for sport videos using court models,” vol. 5307, 01 2004, pp. 80–91.

[3] W.-L. Lu, “Learning to track and identify players from broadcast sports videos,” Ph.D. dissertation, University of British Columbia, 2011. Link

[4] G. Jocher, “YOLOv5 by Ultralytics,” May 2020. Link

[5] A. Maglo, A. Orcesi, and Q.-C. Pham, “Kalicalib: A framework for basketball court registration,” in Proceedings of the 5th International ACM Workshop on Multimedia Content Analysis in Sports, 2022, pp. 111–116.

[6] M. Martinez, C. Sitawarin, K. Finch, L. Meincke, A. Yablonski, and A. Kornhauser, “Beyond grand theft auto v for training, testing and enhancing deep learning in self driving cars,” arXiv preprint arXiv:1712.01397, 2017.

[7] P. Dwivedi. (2019) March madness — analyze video to detect players, teams, and who attempted the basket. Link

[8] Stephan. (2019) Open source sports video analysis using machine learning. Link

📫 7. Contact

• Tien-Li Lin, Email: [email protected]
• Yi-Cheng Liu, Email: [email protected]
• Wei-Cheng Chiang, Email: [email protected]

Please cite the author's paper if you use the code in your work.

@inproceedings{maglo2022kalicalib,
  title={KaliCalib: A Framework for Basketball Court Registration},
  author={Maglo, Adrien and Orcesi, Astrid and Pham, Quoc-Cuong},
  booktitle={Proceedings of the 5th International ACM Workshop on Multimedia Content Analysis in Sports},
  pages={111--116},
  year={2022}
}