Automatic Quantity Takeoff for Irregular Areas of Architectural, Engineering, and Construction Objects from 2D Drawings

Accurate cost estimation of construction projects relies on precise quantity takeoffs (QTOs), which conventionally has been conducted on 2D drawings, and despite the fast development of Building Information Modeling (BIM), remains a predominant practice. This study focuses on the development of a boundary detection algorithm utilizing computer vision techniques for extracting the shapes of infrastructure elements such as buildings and bridges from 2D PDF drawings. The aim is to enhance information extraction capabilities, particularly for QTOs of complex shapes such as the areas of irregular bridge decks. Utilizing computer vision technologies, the algorithm aims to accurately delineate the boundaries of various infrastructure components by eliminating redundant lines, such as measurement lines and dash lines. The proposed boundary detection algorithm achieved 94.1% accuracy in automatic area calculation, and about 99% time reduction compared to manual approach. The proposed approach holds promise for improving efficiency and accuracy in QTO processes, and offering potential solutions in more accurate QTOs.