<![CDATA[Finding an empty parking spot in open areas, particularly in busy locations such as shopping centers, remains a significant challenge. This study proposes a real-time system for detecting outdoor parking space availability using the YOLOv8 algorithm, selected for its speed and accuracy in object detection. The dataset consists of 131 annotated images, expanded through three augmentation techniques (rotation, shearing, and flipping) to increase variability. Model training was performed with multiple hyperparameter configurations and evaluated using precision, recall, F1-score, accuracy, and mAP@50. The best configuration, obtained with the Adam optimizer, achieved 96.74% precision, 99.06% recall, 99.17% mAP@50, and 77.91% accuracy. While the system performed effectively and responsively in real-time daytime scenarios, a key limitation is its reduced performance under nighttime conditions due to low visibility and image noise.This research contributes by demonstrating YOLOv8’s potential to improve real-time detection of parking spaces, particularly through handling occlusions and lighting variations, which remain challenges in outdoor environments.]]>
Copyrights © 2025
