Knowledge Base Repository
In addition to research papers, the Design Society is developing several valuable resources for those interested in the study of design. These include a repository of PhD theses, a library of case studies and transcripts of design activities, and an archive of our newsletters. Please note that these resources are accessible exclusively to Design Society members.
TechMB: Exploring the Potential of Vision Language Models for Interpreting Technical Drawings
Leonhard Kunz, Mario Klostermeier, Kokulan Thanabalan, Tatjana Legler, Martin Ruskowski
Type:
Year:
2025
Editor:
Dieter Krause, Kristin Paetzold, Sandro Wartzack
Author:
Series:
DfX
Institution:
Rheinland-Pfälzische Technische Universität Kaiserslautern-Landau
Section:
Structural Analysis, Simulation & Testing
Page(s):
10
DOI number:
Abstract:
Vision Language Models (VLMs) have gained widespread adoption among end users. Their versatility also sparked interest in applying them to more domain-specific challenges. This paper investigates the principal suitability of small-scale VLMs in the task of evaluating the manufacturability of parts based on a technical drawing by providing the Technical drawings for Manufacturability Benchmark (TechMB). A selection of small-scale VLMs is then tested using this benchmark. The results indicate that the models show potential for text extraction and interpretation of domain-specific terminology. However, they struggle with the reasoning about the manufacturing of the depicted parts and partly even with the delivery of concise and precise answers necessary for the targeted task.
Keywords:
