English | 简体中文
In this project, we have applied the YOLOv5 object detection model to track the players of the San Francisco 49ers NFL team. The goal of this project is to provide a platform for people around the world to access our tracking data and use it for other purposes such as individual player tracking and injury prevention, as well as game analysis and fan engagement.
We have used publicly available video footage of the San Francisco 49ers games for training the YOLOv5 model. The dataset contains video clips of games from different angles, which has been annotated to identify and track players on the field.
We have chosen the YOLOv5 model due to its superior performance in object detection tasks. The YOLOv5 model is an upgrade to the YOLOv4 model, which has been shown to outperform other object detection models in terms of accuracy and speed.
We provide pre-trained models and sample code to get started with the tracking. The code can be run on any computer with a GPU and Python 3 installed. Users can input their own video footage and get real-time tracking of players on the field.
Our tracking data can be used for various purposes such as:
*Individual player tracking: Coaches and trainers can use our tracking data to monitor the performance of individual players, such as their speed, acceleration, and movement patterns on the field.
*Injury prevention: The tracking data can be used to identify players who are at risk of injury due to overuse or poor movement patterns. Coaches can then take preventive measures to reduce the risk of injury.
*Game analysis: The tracking data can be used to analyze the performance of the team as a whole, such as the effectiveness of different plays and strategies.
*Fan engagement: Fans can use the tracking data to get a more immersive experience of the game, such as tracking their favorite player on the field.
We plan to continue improving our tracking models by incorporating more advanced techniques such as multi-object tracking and action recognition. We also plan to expand our dataset to include more games and teams from the NFL and other sports leagues.
We would like to thank the San Francisco 49ers organization for providing access to their game footage and data. We would also like to acknowledge the contributions of the YOLOv5 development team, whose code we have used in this project.
This project is licensed under the MIT License - see the LICENSE.md file for details.
Luxolo Kuhlane & Nicholas Stadler