Tracking football players with a conventional mobile device camera

Abstract

One of many important tasks in computer vision applications is real time object detection and tracking. Although classical tracking algorithms work accurately in perfectly arranged conditions, tracking objects in real-life situations poses challenges due to the multitude of variable conditions. However, efficient object tracking is a difficult task due to the multitude of the conditions met. Significant problems arise in computer vision systems, especially under the presence of noise, scene illumination, object shape variation and total or partial occlusion. The purpose of this paper is to develop a method for effective object tracking and detection in complex scenes, such as sports games. The proposed approach would enable to accurately track targets in motion in real scenes.