The LBW (Leg Before Wicket) Review System is a project aimed at detecting and analyzing ball movements in cricket matches, particularly focusing on LBW scenarios. The system utilizes the YOLOv8 algorithm for ball detection, trained on a diverse dataset consisting of various cricket balls (pink, red, white) and other balls such as tennis balls.
The YOLOv9 model has been trained on a comprehensive dataset available on Roboflow. This dataset includes a wide range of cricket ball images captured from different angles and under various lighting conditions, enhancing the model's ability to accurately detect balls during matches.
https://universe.roboflow.com/ahmed-ws/ball-detection-3afaq/model/5
The performance of the trained YOLOv8 model is evaluated using the following metrics:
- Mean Average Precision (mAP): 91.9%
- Precision: 86.3%
- Recall: 94.8%
Some Overfitting
To utilize the LBW Review System:
- Clone the repository:
git clone https://github.com/ahmedembeddedx/lbw-review-system - Install the necessary dependencies.
- Run the provided scripts for inference on cricket match videos or live streams.
- Analyze the output for LBW scenarios and review decisions.
Contributions to the LBW Review System are welcome! If you have suggestions for improvements or encounter any issues, feel free to open an issue or submit a pull request.
This project is licensed under the Apache License