This study proposes an efficient workflow for generating accurate discrete rigid block models of unreinforced masonry (URM) walls with openings from vision-based data. The proposed methodology uses AI-assisted object detection to identify the masonry units within an image of a target wall section (− 1 × 1 m). Statistical masonry texture characteristics (e.g., unit size distribution) and a masonry quality index parameter (e.g., vertical joint alignment) are automatically extracted and applied to generate a digital representation of the target wall. From this information, discrete blocks representing the expanded masonry units are generated row by row and used within the discrete element method (DEM) framework. The proposed workflow is applied to create a digital model of a URM wall in Kemptville, Ontario, Canada. A pushover analysis is performed in DEM, and the predicted failure mechanisms are used to complete an additional macro-block analysis. The results of this study demonstrate the potential of the proposed pipeline, which is shown to successfully capture the meso-scale texture and construction quality of the selected masonry façade. Emphasis is given to accurately representing the masonry wall texture and the “as-is” (or in situ) configuration in the adopted discontinuum-based computational modeling strategy. Therefore, the proposed workflow presents an efficient and accurate alternative to conventional CAD-based drafting techniques.