Remote architectural modeling faces persistent challenges due to the reliance on expert-driven shape grammar encoding and insufficient realism in reconstruction. To address these issues in networked and resource-constrained environments, we propose a symmetry-driven procedural modeling framework tailored for cloud-edge collaborative deployment. The framework supports seven fundamental architectural block types and twenty textures, enabling users to declaratively define building layouts via high-level shape grammar. Leveraging semantic segmentation of façade images powered by large visual models, our system automatically generates shape grammar expressions and adapts to diverse network conditions through edge-assisted processing. Comparative experiments demonstrate superior realism and customization capabilities over existing methods. Furthermore, a deployment across 100 residential areas in Beijing achieved 100% reconstruction coverage, validating the framework’s robustness, scalability, and potential for integration in smart city modeling and communication-aware digital twin systems.
Wang et al. (Fri,) studied this question.