An SVM-based Adaptive Data Refinement and Augmentation Mechanism: A Use Case Study for Football Talent Recognition

This paper introduces an adaptive data refinement and augmentation mechanism for football talent recognition. The proposed system utilizes machine learning and wearable technologies to automate the assessment of player talent during training. Various machine learning algorithms are employed to analyze metrics such as acceleration and rotation, captured by motion sensors, to objectively evaluate talent. As a use case, the system was tested on professional football players during shooting training sessions, initially achieving a promising classification accuracy of 91% with the Support Vector Machine (SVM) algorithm. However, model overfitting concerns required a data augmentation technique, which improved accuracy to 96%.