Research and application of YOLOv8 model in detecting and counting available parking spaces in a car park

Abstract

Recently, the number of cars in major urban areas of Vietnam has increased rapidly. Therefore, information about the locations of parking lots and the number of available parking spaces is essential, helping drivers quickly find the most suitable parking spot. Many technologies and algorithms have been applied to recognize and classify objects, among which YOLOv8 cannot be overlooked. This technology can detect objects in real-time thanks to its fast processing speed, high accuracy, and ease of integration and deployment in practical systems. With the outstanding advantages of YOLOv8, this research aims to detect cars in parking lots using this technology. Then, it evaluates and analyzes the collected vehicle data to statistically count the available spots in car parking lots. The research results show that image recognition achieves an accuracy of 90 % or more, while car recognition from videos (real-time) only reaches 80 %.