Leveraging YOLOv8 for Real-Time Parking Space Detection

This research develops a smart parking system based on computer vision integrated with the Firebase platform, utilizing the YOLOv8 algorithm and the OpenCV library. This system is designed to detect vehicles and determine the availability status of parking slots in real time through video recording analysis. The dataset was obtained from recorded parking activity videos, then processed using Roboflow for data labeling, and fine-tuned on Google Colab. The resulting model shows a vehicle detection accuracy of up to 99.1% with high stability, as indicated by the training loss graph. The process of mapping the parking area is carried out through an OpenCV-based application for the identification and configuration of parking slots. The status of each parking slot is automatically updated to the Firebase Realtime Database, enabling efficient historical data recording and information management. Testing shows the system's reliable performance under various lighting conditions, with high capabilities in vehicle detection and real-time data management. This study makes a significant contribution to the development of parking management solutions through the integration of technology that supports operational efficiency and high accuracy.