Title Block Detection and Information Extraction for Enhanced Building Drawings Search

Abstract: The architecture, engineering, and construction (AEC) industry still heavily relies on information stored in drawings. However, information extraction (IE) from building drawings is time-consuming and costly, especially when dealing with historical buildings. Drawing search can be simplified by leveraging the information stored in the title block. The work proposes an proposes a novel title block detection and IE pipeline by integrating Faster R-CNN and GPT-4o which enables advanced search and outperforms existing methods. An extensible domain-expert annotated dataset is produced via a novel AEC-friendly annotation pipeline that lays the foundation for future work.