Football Player Detection and Tracking Using Deep Learning Approach

Detection and Tracking of Football Players in Video Clips addresses the challenge of accurately identifying and tracking football players amid dynamic movements, changing lighting conditions and frequent occlusions. This study utilizes advanced computer vision techniques specifically YOLOv5 for object detection and BYTETrack for object tracking. YOLOv5 efficiently detects objects by dividing the input image into a grid and predicting bounding boxes and class probabilities which is essential for precise multi-object detection. BYTETrack improves tracking accuracy by maintaining low-score non-background boxes to ensure effective association across frames. The performance of the system is evaluated using metrics such as confusion matrix, precision, recall, and F1 score which provides insights into its accuracy and reliability. The proposed work demonstrates how integrating cutting-edge computer vision technologies can enhance sports analytics, improve the understanding of player movements, optimize coaching strategies and advancing tactical analysis. The model demonstrates strong overall performance with a precision of 0.904, recall of 0.864 and mAP50 of 0.895. It particularly excels detection of players with a precision of 0.969, recall of 0.987, mAP50 of 0.993 and mAP50-95 of 0.86 while showing varied accuracy across different classes.