AI-Driven Prediction of Optimal Player Positions Based on Fundamental Skills in Netball

Abstract

This research presents an innovative, video-based performance monitoring system designed to assess and enhance the fundamental skills of netball players while predicting optimal playing positions. Using advanced computer vision and machine learning techniques, the system evaluates fundamental skills through automated video analysis. The framework integrates Convolutional Neural Networks (CNNs), with feature extraction powered by pre-trained ResNet50 models, and temporal sequence analysis using Long-Short-Term Memory (LSTM) networks. In addition, random forest classifiers, supported by SMOTE-balanced data sets, are used for accurate position prediction. The system also incorporates centroid-based motion tracking through OpenCV, enabling precise monitoring of player movements. Unlike traditional subjective coaching methods, data-driven insights offer tailored training recommendations. The predictive analytics component also anticipates player development trajectories and role assignment. By transforming conventional netball training into an AI-driven predictive process, this research provides a powerful tool for coaches and players, improving skill evaluation accuracy, optimizing player positioning, and fostering informed evidence-based decision-making. Ultimately, this innovation contributes to improved individual performance, more cohesive team dynamics, and increased competitive success.