The construction industry, traditionally slow in adopting digital innovations, continues to face critical inefficiencies during the pre-construction phase—particularly in quantity takeoff (QTO), cost estimation, and architectural drawing analysis. This article explores the transformative potential of artificial intelligence (AI) and computer vision (CV) in automating these essential yet labor-intensive processes. The study reviews leading tools such as Togal, BlueBeam, PlanSwift, and OnScreen, which leverage machine learning to recognize patterns, detect symbols, and extract structured data from PDF-based construction documents. These technologies enable the accurate identification of architectural components—including walls, doors, plumbing fixtures, and parking areas—with precision exceeding 98%, significantly reducing human labor and errors. Despite these advancements, the article outlines existing limitations such as the reliance on high-quality annotated datasets, occasional segmentation inaccuracies, and the challenge of integrating AI-driven systems with traditional project management environments. Manual review is still required to ensure final accuracy. To address these gaps, the authors propose a scalable, modular cloud-based architecture that employs ResNet-50 with PointRend for fine-grained wall and door segmentation, YOLO-X for object detection, and Natural Language Processing (NLP) for analyzing textual content. The entire workflow is visualized via a flowchart diagram that guides the user through validation, segmentation, classification, error handling, and final report generation. The findings demonstrate that AI-enhanced QTO systems significantly improve planning accuracy, reduce risks, and foster collaboration—while also highlighting the need for continued innovation and broader industry adoption to fully unlock digital transformation in construction.
Коваленко et al. (Mon,) studied this question.