Smart parking systems play a pivotal role in traffic optimization and decreasing congestion in urban areas. Traditional sensor-based approaches often face limitations in scalability and cost-effectiveness. In order to overcome these limitations, this paper explores the capabilities of model YOLOv9, a cutting-edge object detection model that is tailored for real-time applications. This paper proposes a novel approach that leverages YOLOv9’s efficiency and precision to enable real-time detection of available parking spaces. By leveraging its advanced features, it significantly improves the parking management system and enhance the overall convenience for drivers. Here, it compares the performance of YOLOv9 with its predecessor, YOLOv8, specifically for this task. Furthermore, the paper outlines the methodology for adapting YOLOv9 to detect available parking spaces in smart parking scenarios, providing practical insights into its implementation. The findings reveal that YOLOv9 model achieves an impressive accuracy rate of 95.8%, in comparison to accuracy value extracted of 95.1% from model YOLOv8. Additionally, YOLOv9 exhibits a remarkable 8.2% improvement in training time efficiency compared to YOLOv8, enhancing its suitability for real-time processing. With its exceptional accuracy and efficiency gains, YOLOv9 emerges as the preferred choice of model for detecting parking spaces, especially in real-time applications like smart parking system.