BadmintonTracker

Tracking players and the shuttlecock on the badminton court; a potential continuation of the BasketballTracker project.

The “BadmintonTracker” project aims to develop software for detecting and analyzing the position and speed of badminton players, as well as the shuttlecock, using images captured by tactical cameras during official matches. This project lies at the intersection of software engineering, computer vision, and sports analytics.

The project’s objectives are divided into two phases:

Software Development: Software will be designed that applies advanced image processing and computer vision techniques. This software will consist of detection and tracking algorithms that accurately identify the position and trajectory of the players and the shuttlecock.

Data Extraction: Kinematic parameters such as the position, velocity, and acceleration of the players and the ball will be extracted and calculated in real time. The variables will be structured and stored in a format suitable for subsequent quantitative and qualitative analysis.

Regarding the project methodology, images from tactical cameras used during badminton matches will be utilized. Preprocessing techniques, such as noise filtering, distortion correction, and lighting normalization, will be applied. Next, algorithms based on convolutional neural networks (CNNs) will be implemented to detect and segment the players and the shuttlecock in each frame. Tracking techniques such as the Kalman filter and the SORT (Simple Online and Realtime Tracking) algorithm will be used to maintain continuous tracking of objects throughout the video sequence. Data association techniques will also be implemented to resolve occlusion issues and temporary detection losses. Finally, the velocity and acceleration of the players and the shuttlecock will be calculated using numerical differentiation methods and time-series analysis. If possible, algorithms will be developed to detect critical events such as smashes, drop shots, and rapid movements based on kinematic thresholds.


TOP