New AI-assisted method cross-checks building design documents to help detect errors

by Seoul National University of Science and Technology

edited by Swati Mestri, reviewed by Robert Egan

This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility:

Add to Preferred Sources The proposed pipeline achieves greater than 90% accuracy in extracting information from structural design documents and in experiments identified 253 human errors, including 29 critical errors, demonstrating its potential for reducing the burden of manual checking. Credit: Associate Professor Jaewook Jeong from Seoul National University of Science and Technology Collaboration between different stakeholders is a key part of project-based industries like construction. A variety of documents are required for effective communication, and inconsistencies between them could lead to project delays and increased costs, making early detection essential.

Digital collaboration tools like building information modeling (BIM) have been proposed to address this challenge. BIM allows multiple designers to share and revise engineering drawings and documents, enabling early detection of inconsistencies. However, high initial costs, lack of skilled personnel and technical barriers limit its adoption, leaving many projects reliant on traditional, document-based collaboration.

When different stakeholders use different tools and document formats, information can be misinterpreted or inconsistently represented. This is particularly concerning for structural calculation documents and structural drawings. Such discrepancies can introduce human errors that, in severe cases, could compromise building safety. Preventing such errors, therefore, requires meticulous management.

To alleviate this burden, a research team led by associate professor Jaewook Jeong from the Department of Safety Engineering at Seoul National University of Science and Technology in South Korea has developed a new artificial intelligence (AI)-assisted automated pipeline for detecting human errors in structural design documents.

"Our AI-assisted framework automatically extracts, organizes and compares information from structural calculation documents and drawings to identify inconsistencies that may result from human error. Importantly, the system performs these checks without relying on building information modeling (BIM), making it particularly relevant to construction projects that continue to use conventional design documents," explains Jeong.

Their study was published in Engineering Applications of Artificial Intelligence.

The proposed automated pipeline starts with project data collection and preprocessing, during which necessary documents are acquired and low-quality information is pruned. It then extracts and standardizes structural information from calculation documents and drawings. For calculation documents, the DocLayNet-YOLO net model, a generalized document layout analysis model, first identifies regions containing target information. Next, key structural member information in the identified regions is extracted in text form using the Surya-OCR model. In the final step, this extracted information is merged and standardized into a single file.

For structural drawings, where multiple members are listed in schedules with varying row and column widths, member codes are identified first. OCR, together with Canny edge detection, Hough transform and morphological operations, is employed to locate table regions containing these member codes. The extracted information in these member code regions is then filtered and refined, followed by standardization.

The pipeline then validates the extracted information and compares data from the two document types. Member codes and direction are used as key identifiers to establish correspondence between structural calculation documents and drawings. The system detects errors, identifies and reports duplicate entries, and organizes the results into a consistent data set. The errors identified are then categorized. The researchers also developed a user-friendly, web-based application to deploy this pipeline.

In a case study involving a residential construction project in South Korea, the proposed system achieved information extraction accuracies of 95.56% for structural calculation documents and 90.47% for structural drawing documents. The consistency verification process identified 253 discrepancies, including 224 missing entries, 15 main reinforcement mismatches, three stirrup mismatches, eight dimensional discrepancies and three duplicate entries. These findings demonstrate the framework's potential to support more systematic and efficient structural design document verification.

"Our AI-assisted pipeline could help structural engineers, design consultants, contractors and construction supervision teams to identify discrepancies before they lead to construction errors, improving the reliability of buildings and infrastructure," remarks Jeong. "It could be particularly valuable for small- and medium-sized construction projects that cannot implement BIM."

Overall, this AI framework could help reduce the burden of manual checking and improve consistency in construction projects where BIM cannot be implemented. In the future, it could become integrated into routine engineering workflows, making advanced safety technologies more accessible.

Minsang Gu et al., Detection of human errors in construction structural design documents using automated extraction algorithm, Engineering Applications of Artificial Intelligence (2026). DOI: 10.1016/j.engappai.2026.116556 Key concepts AI-enabled digital twins Provided by Seoul National University of Science and Technology Who's behind this story? Swati Mestri Swati Mestri holds a bachelor's degree in Electronics Engineering and has worked as a content editor since 2019. She has experience editing research documents across technology, health care, and materials science, and has a particular interest in technology and space. Full profile →

Bachelor's in mathematical biology, Master's in creative writing. Well-traveled with unique perspectives on science and language. Full profile →