Many existing masonry structures show significant signs of deterioration, mostly due to the continuous effect of environmental phenomena and changes in load application over long periods. Thus, continuous structural health monitoring of such constructions is becoming more and more important. In this paper, we describe a proposed workflow to automate the generation of three-dimensional geometrical digital twin models of masonry structures. The workflow makes use of image-processing and deep-learning techniques for the detection of structural elements. Additionally, 3D-scanners (LiDAR)/photogrammetry is used for the generation of photorealistic replicas of the structure’s geometry. As a case study, a masonry arch bridge tested in the laboratory has been used. The bridge is 3 m in span and is composed of spandrel walls and limestone as backfill material. The workflows presented in this work have the potential to create a completely autonomous system for the monitoring and assessment of masonry structures in the field.