This study aims to address the persistent challenges in traditional process of construction project management, which includes ineffective resource allocation, productivity monitoring and performance evaluation by proposing a computer vision-based solution.
An integrated system combining the trained YOLOv8x for object detection and ByteTrack for resource tracking, along with a user-friendly graphical user interface, is developed to support real-time analysis of as-planned vs as-built resource variations (manpower and equipment). To evaluate the effectiveness of developed system and assess its impact on project productivity and performance, a case study construction project is used.
The system achieved 95.38% accuracy in resource detection during its deployment period in the case study project, enabling precise tracking and real-time productivity analysis. This resulted in a two-day schedule gain and cost savings of Rs 112,561, highlighting its potential to optimize resource utilization and improve project outcome.
By automating resource tracking and linking it with productivity and performance metrics, this research provides a way for more efficient, cost-effective and technology-driven project management practices and sets the foundation for advanced monitoring of construction projects.
While prior studies have explored the development of computer vision-based systems for resource monitoring and construction management, there is a lack of empirical studies quantifying their influence on productivity and performance of the construction project. To address this gap, this research introduces a novel approach that integrates data from a developed resource tracking system with productivity and performance metrics to evaluate its influence on a real-world construction project.
