Automated monitoring of parking infrastructure using machine learning and computer vision methods

Background. The article examines the current problem of parking space shortage in modern urban conditions. The authors have developed an innovative solution based on an automated monitoring system using computer vision and deep learning methods. A comprehensive analysis of existing global analogues of parking management systems is carried out, highlighting their competitive advantages and significant limitations. As a methodological basis, a detailed process model presented in BPMN 2.0 notation is proposed, which includes a description of the solution architecture, video data processing algorithms and a neural network training methodology. Particular attention is paid to the development of a specialized reporting template that provides a visual representation of statistical data on the occupancy of parking spaces in real time.

Purpose. The purpose is to improve the efficiency of parking infrastructure management through the implementation of intelligent algorithms for automatic recognition.

Materials and methods. The study employed a comprehensive scientific approach incorporating machine learning techniques, systems theory, systems analysis and synthesis, along with analytical and statistical methods.

Results. The work uses a set of modern methods, including machine learning technologies (with an emphasis on the use of the YOLOv8m model), principles of system analysis and synthesis, as well as methods of statistical data processing.