ETD

• artificial intelligence
• computer vision
• data mining
• deep learning
• UML quality AI
• UML schema
• YOLO

This thesis focuses on the quality assessment of UML diagram layouts employing cutting-edge machine learning and computer vision techniques. The developed software, following a meticulous training phase, demonstrates the capability to assign a quality grade and furnish constructive feedback to designers upon submitting their schemas. State-of-the-art methodologies were employed, and the achieved results stand as a benchmark in addressing the challenges inherent to UML diagram layout assessment.

The framework leverages advanced tools such as YOLO, PyTorch, OpenCV, and Detecto, including the Faster R-CNN architecture. This amalgamation of frameworks contributes to the robustness and efficiency of the system, ensuring that the outcomes not only meet but also compare favorably with the current state-of-the-art solutions to this pervasive problem.