Tennis-Analysis-System is a system that analyzes tennis matches and provides player statistics and mini court visualizations and It utilizes YOLO (You Only Look Once) v8x for player detection and tracking, and fine-tunes YOLOv5 for ball detection and tracking. Additionally, a ResNet50 model is used for keypoint detection and tracking.
- Clone the repository:
git clone https://github.com/Sh-31/Tennis-Analysis-System.git- Install the required dependencies:
pip3 install -r requirements.txt (linux)
pip install -r requirements.txt (windows)- Run the main script:
python3 main.py (linux)
python main.py (windows)output_video.mp4
For ball detection, we used the tennis-ball-detection dataset from Roboflow. You can access it at the following link:
For keypoint detection, we used a dataset collected by yastrebksv. You can download it from the link below:
You can also download it using the code provided in the Tennis-Analysis-system\fine_tuning\explore.ipynb notebook.
| Epoch | Train/Box Loss | Train/Cls Loss | Train/Dfl Loss | Precision (B) | Recall (B) | Val/Box Loss | Val/Cls Loss |
|---|---|---|---|---|---|---|---|
| 95 | 2.052 | 0.94742 | 0.86513 | 0.76142 | 0.53465 | 1.9983 | 1.1024 |
| 96 | 1.9937 | 0.90107 | 0.83483 | 0.71385 | 0.46934 | 2.0944 | 1.1647 |
| 97 | 1.9396 | 0.88057 | 0.83543 | 0.73493 | 0.48515 | 2.1743 | 1.233 |
| 98 | 1.9461 | 0.92784 | 0.82794 | 0.72858 | 0.50497 | 2.1229 | 1.2019 |
| 99 | 1.8929 | 0.82987 | 0.83066 | 0.74648 | 0.49505 | 2.0971 | 1.1892 |
| 100 | 1.933 | 0.88627 | 0.84558 | 0.75446 | 0.49505 | 2.1242 | 1.1936 |
- Player detection and tracking using YOLOv8
- Ball detection and tracking using YOLOv5 (fine-tuned)
- Keypoint detection and tracking using a Resnet50 model (fine-tuned)
- Mini court visualizations with player and ball positions
- Player statistics calculation
- Mini-court coordination is not generic (It needs the true height of the player to convert the proportion of meters to pixels).