During the construction period, numerous site photographs are routinely captured; however, their use is largely limited to simple visual inspection of construction status. To enhance the practical utilization of such photographic information, this study proposes a 4D-based construction progress management system that visually evaluates schedule progress by integrating site photographs within a BIM-based information management framework. The proposed system synchronizes site photographs with corresponding 4D model images using coordinate linkage and applies deep learning–based object detection to identify matching construction elements. Construction progress is approximately estimated by analyzing bounding box overlap between detected elements in site photographs and planned elements in 4D model images. A case study conducted on a bridge construction project demonstrated that the trained model achieved an overall mAP@0.5 of 0.532, and that the proposed method enables intuitive and approximate progress evaluation. The results indicate that the proposed system can improve the usability of site photographs as supporting information for 4D-based construction progress management.
Park et al. (Wed,) studied this question.