<![CDATA[Vehicle license plate detection is a crucial computer vision task for traffic monitoring, automated parking, and vehicle identification. This study evaluates the performance of a YOLO-based license plate detection system implemented in Python and executed on Google Colab to ensure reproducibility. A public dataset of vehicle images with variations in lighting conditions and viewing angles is used for testing. Performance is assessed using precision, recall, F1-score, mAP@0.5, and mAP@0.5:0.95. The results show a precision of 0.7653 and a recall of 0.6809, yielding an F1-score of 0.7206. The mAP@0.5 reaches 0.7776, while the mAP@0.5:0.95 drops to 0.3572. As a contribution, this work provides a simple and replicable baseline evaluation workflow for YOLO-based license plate detection using standard object-detection metrics. The large gap between mAP@0.5 and mAP@0.5:0.95 indicates that the model often detects the presence of license plates but struggles to localize them precisely under stricter IoU thresholds, highlighting localization sensitivity for small objects under real-world variations. These findings can guide future improvements through dataset diversification, augmentation, and higher-resolution training to enhance bounding box accuracy.]]>
Copyrights © 2026
