AI Object Detection of Parts in Engineering Drawings and a Concept of Integration in Engineering Education - AssistME

Oct 23, 2026

AI Object Detection of Parts in Engineering Drawings and a Concept of Integration in Engineering Education-AssistME explores how AI and VR can be integrated into engineering design education. It uses the Scaled-YOLOv4 object detection algorithm to recognize parts in engineering drawings, improving prediction quality through techniques such as tiling and post-logic adjustments to reach mAP up to 0.909. The detected information is exported to a VR environment via JSON and enables analysis of students’ individual design solutions.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

1 Introduction

2 AI object detection and VR in engineering and education

2.1 AI object detection - short overview

2.2 AI object detection in engineering

AI Object Detection of Parts in Engineering Drawings and a Concept of Integration in Engineering Education-AssistME Jan Lennart Kraske, Abdullah Doksanbir, Fabian Dillenhöfer, Bernd Künne Technical University Dortmund, Germany Abstract: This article shows how AI and VR technology can be implemented into engineering design education. AI is used to recognize parts in engineering design drawings. The recognition task is done by the Scaled-YOLOv4 object detection algorithm. Implementing adjustments such as tiling and post-logic adjustments regarding the predictions, the detection quality can be improved up to a mAP value of 0.909. The detection of the AI can be transferred into a VR environment via a JSON file. It is shown that the developed AI model is capable of analyzing student’s individual solutions. Keywords: Artificial Intelligence (AI), Object Detection, Virtual Reality (VR), Design Education, Engineering 1 Introduction AI object detection is used in many different applications such as autonomous driving, education and even in everyday life (Adiguzel et al., 2023; Atakishiyev et al., 2021; Dixit et al., 2022) . With the release of ChatGPT, many people were exposed to this technology for the first time and the interest in AI has been increasing ever since. It is also applied to different engineering tasks, for example for analyzing technical drawings. With the help of AI based object detection algorithms, it is possible to extract letters, numbers and symbols out of these images (Eylan et al., 2020, Sarkar et al. 2022) or classify certain objects inside the drawings (Dillenhöfer, 2023) . This work demonstrates the AI-based detection of parts in engineering drawings and introduces a concept for its integration into engineering education in section 2. Many engineering students, especially those in the earlier semesters, have issues imagining technical drawings in three dimensional space. To be a successful engineer it is essential to have well developed spatial skills (Duffy et al., 2018) . Spatial understanding of technical drawings requires experience, which many students lack. As a result, many avoidable mistakes are made in the design of mechanical devices such as gearboxes. In the particular system that is presented in this work, AI and virtual reality (VR) technology are combined to extract individual design solutions from technical drawings to visualize and gamify it in a VR environment. Han showed, that the use of VR has a positive influence on learning (Han et al., 2023) . The aim of this work is to investigate whether AI can be used to recognize components in technical drawings with sufficient accuracy to cover the wide variety of student design solutions. The recognized components need to be processed in such a way that they can be used in a VR environment. The approach intends to show the relation between drawing a 3D object and thereby help students understand their own designs better as well as understanding whole machines and their parts. For this purpose, related works are analyzed and the necessary quality requirements are derived. Later, the detection performance of the trained AI is analyzed and discussed. 2 AI object detection and VR in engineering and education In this section, the state ofthe art regarding AI object detection particularly in engineering is being introduced. The impact of new technologies integrated in education are shown and which impact they have on the learning process of students. 2.1 AI object detection-short overview Image recognition is a sub-area of image processing and its task is to identify an object or a feature (feature map) in animage or video (Chai et al., 2021). More and more AI object detection algorithms were revealed recently. One of the most popular object detection algorithms (ODA) is YOLO (You Only Look Once) (Redmon et al., 2016) . Object detection algorithms are divided into two classes. There are one-stage and two-stage algorithms. The RCNN

FAQs

How does the work recognize parts in engineering drawings?

It applies the Scaled-YOLOv4 object detection algorithm to identify parts in engineering design drawings. Quality is enhanced using adjustments such as tiling and post-logic corrections.

What performance level is reported for the object detection model?

The detection quality is improved up to an mAP value of 0.909.

How is the AI result used in the VR learning environment?

The detected output is transferred into a VR environment via a JSON file, allowing the system to analyze and visualize students’ individual design solutions.

Oct 23, 2026

Upload by The UK Girl