Adaptive OCR for Structured Data Extraction in Technical PDF Documents

The proposed research project aims to digitize information currently locked in image-only PDF files, which are common in government and engineering sectors but are difficult to search and analyze. These documents, which often include technical drawings, forms, and reports, cause inefficiencies and errors due to the need for manual review. To address this, the project will develop an adaptive system using advanced image recognition (OCR) and vision-language models to accurately extract structured data. This involves creating algorithms for content and layout analysis and building a user-friendly software tool to help people upload, view, and export this extracted information. For the participating institutions, the University of Alberta in Canada and the National Taiwan University of Science and Technology, this project strengthens their ongoing collaboration in digitalizing building processes and advances data-driven design and construction. The project’s outcomes will streamline document workflows, reduce manual effort, and improve data analysis for both academic research and real-world applications in engineering and government.

Faculty Supervisor:

Yuxiang Chen

Student:

Partner:

National Taiwan University of Science and Technology

Discipline:

Engineering

Sector:

Education

University:

University of Alberta

Program: