Integrated Framework Using Computer Vision and Ultra-Wide Band Techniques for Progress Reporting in Construction Projects

Timely delivery of construction projects within budget requires timely and reliable data-driven progress reporting. Proper analysis of the collected data is essential for efficient extraction of beneficial information from that data for better understanding of project progress and its status. The captured and analyzed data allow responsible parties to compare the as-built condition with the as-planned state and identify the needed corrective actions during construction phase. Digital cameras are used in the developed framework in data collection. The captured images are then utilized for recognition and identification of objects on construction job sites using deep neural network learning algorithms. The identified objects are then localized, i.e., their respective 3D coordinates are identified by using location tracking technologies of tags that are attached to the objects. This paper presents an integrated framework by using a novel deep learning-based object recognition algorithm and Ultra-Wide Band (UWB) technology to automatically recognize and localize the tracked building objects on project jobsites for better assessment of their installation status. The paper focuses on the developed automated integration of characteristics of recognized objects and their respective locations. The output report provides complete and reliable information on each tracked object including its visual condition, ID, location, object class, and captured time. This framework can overcome the limitations of each individual technology used in the integration process. And accordingly, improve accuracy and reliability by visual validation of tracked building objects in indoor environment and facilitate timely decision making in management of construction projects.