Analýza geometrie a obsazenosti parkovacích míst pomocí obrazů

Abstract

This master's thesis addresses the automated analysis of parking areas using computer vision and deep learning methods. The primary objective is to design and implement an experimental software solution focused on detecting parking space geometry without the need for manual annotation. The system analyzes camera footage and dynamically generates a map of parking spots based on the detection and spatial distribution of vehicles. The solution also includes the subsequent occupancy evaluation of these identified spaces. Modern approaches to object detection and image segmentation were utilized for the implementation. The resulting functionality of the proposed solutions was experimentally verified on real-world data, with an emphasis on robustness under various operating conditions.