Are Open-Vocabulary Models Ready for Detection of MEP Elements on Construction Sites

TL;DR

This paper evaluates the effectiveness of open-vocabulary vision-language models versus fine-tuned object detectors for detecting MEP elements on construction sites, revealing that specialized models still outperform general models in this domain.

Contribution

It provides a comparative analysis of open-vocabulary models and fine-tuned detectors for construction site MEP detection using a robotic platform.

Findings

Fine-tuned lightweight models outperform vision-language models in this task.

Vision-language models lack the specialization needed for construction site MEP detection.

The study offers insights into model selection for robotic construction site monitoring.

Abstract

The construction industry has long explored robotics and computer vision, yet their deployment on construction sites remains very limited. These technologies have the potential to revolutionize traditional workflows by enhancing accuracy, efficiency, and safety in construction management. Ground robots equipped with advanced vision systems could automate tasks such as monitoring mechanical, electrical, and plumbing (MEP) systems. The present research evaluates the applicability of open-vocabulary vision-language models compared to fine-tuned, lightweight, closed-set object detectors for detecting MEP components using a mobile ground robotic platform. A dataset collected with cameras mounted on a ground robot was manually annotated and analyzed to compare model performance. The results demonstrate that, despite the versatility of vision-language models, fine-tuned lightweight models still largely outperform them in specialized environments and for domain-specific tasks.