• Version
• Download 6
• File Size 300.79 KB
• File Count 1
• Create Date 23 December 2025
• Last Updated 23 December 2025

Automated Construction Site Analysis Using Images: A Review

 

 

Dr. Manjusha Tatiya, Vaishnavi Katikar, Samruddhi Bagal, Gajendra Thakur, Maharudra Ganjure

Artificial Intelligence and Data Science/ Indira College of Engineering and Management / Savitribai Phule Pune University, Pune, India

Corresponding Author Email: hodai_ds@indiraicem.ac.in vaishnavi.katikar@indiraicem.ac.in | ORCID: https://orcid.org/0009-0006-8784-3810

bagalsamruddhi22@gmail.com| ORCID: https://orcid.org/0009-0005-5994-9151

gajendrathakur1031@gmail.com | ORCID: https://orcid.org/0009-0006-5156-6196

maharudraganjure@gmail.com | ORCID: https://orcid.org/0009-0009-6055-9918

 

 

Abstract
Keeping construction projects on track is a major challenge for project managers. The outdated method of sending authorities to a site for manual inspection is inefficient, expensive, and difficult to scale across multiple projects. To address these problems, recent research has progressively turned to computer vision and machine learning to automate progress monitoring. This paper reviews the current state of these automated techniques, creating findings from key recent studies. Current research establishes significant success in using AI to analyze site images. Deep learning models like Convolutional Neural Networks (CNNs) are now extensively used for detecting construction stages and classifying materials. For more detailed tasks, such as identifying specific building components, object detection algorithms like YOLO and Mask R-CNN have also proven effective. Additionally, recent studies are actively addressing the unique challenges of monitoring progress in complex indoor environments. Despite these advancements, a notable gap remains between the capabilities of these AI models and the general needs of project management. The literature consistently highlights persistent challenges such as image obstructions, poor lighting conditions, and the need for more granular, activity-level tracking. This review consolidates the progress made in the field and highlights the critical next steps needed to bridge the gap from specialized AI tools to fully integrated, reliable construction monitoring platforms.

Keywords— Convolutional Neural Networks, Deep Learning, Object Detection, YOLO, Mask R-CNN, 3D Scene Reconstruction, Image Segmentation.

Download