Construction Site Layout Planning and Progress Monitoring Using Computer Vision on Drone Images

Abstract: Recent advancements in construction methods necessitate automated technologies for designing, planning, and monitoring workplaces. Construction site layout planning and control is often manual and infrequent. This framework applies deep learning and computer vision techniques to automate continuous monitoring and inspection of protective safety barriers separating workers from vehicle traffic. Orthomosaic maps, generated from georeferenced aerial images, assist in training a deep learning model. Automatic barrier detection links to the original site layout baseline. A custom-built layout analysis algorithm identifies changes over time. This approach enables practitioners to assess site safety, track modifications, and implement necessary improvements effectively.