Football Player Tracking and Performance Analysis Using the OpenCV Library

Abstract

In recent years, the performance evaluation for footballers in field as traditional way is turned to the electronic platforms. This is as a relative approach to the traditional ones that suffer from human errors or need special designed equipment and expensive software. Therefore, there is a need for simple and accurate electronic system to evaluate the performance of football players at training and matches sessions, this help coaches to work better during training and to enhance the identified weakness and guide training tasks based on quantative information. The optical tracking techniques are employed mainly for processing real-time and recorded football matches to track a footballer and getting the important data. In this paper, an optical tracking model is proposed for tracking footballers in non-real-time matches’ videos using OpenCV library. Different image processing operations have been performed to include the conversions, filtering, feature extraction and identification. Moreover, the proposed model collects the tracking parameters of a footballer to be saved in a built dataset, which represents the main contribution of the work. The obtained parameters include footballer speed, crossed distance, and activity counts of each player. The proposed model is tested over match videos with maximum 64 seconds time period with 1598 frames. The tracking of a footballer is proved by its high accuracy and applied on 15 videos to create new dataset.