The rise of 2D vision-language models (VLMs) has en-abled new possibilities for language-driven 3D scene understanding tasks. Existing works focus on indoor scenes, or autonomous driving scenarios and typically validate against a pre-defined set of semantic object classes. In this work, we analyze the capabilities of vision-language models for large-scale urban 3D scene understanding, and propose new applications of VLMs that directly operate on aerial 3D reconstructions of cities. In particular, we address higher-level 3D scene understanding tasks such as population den-sity, building age, property prices, crime rate, and noise pollution. Our analysis reveals surprising zero-shot and few-shot performance of VLMs in urban environments.

Abstract We present Norma, a tool for the modeling and analysis of Relay-based Railways Interlocking Systems (RRIS). Norma is the result of a research project funded by the Italian Railway Network, to support the reverse engineering and migration to computer-based technology of legacy RRIS. The frontend fully supports the graphical modeling of Italian RRIS, with a palette of over two hundred basic components, stubs to abstract RRIS subcircuits, and requirements in terms of formal properties. The internal component based representation is translated into highly optimized Timed nuXmv models, and supports various syntactic and semantic checks based on formal verification, simulation and test case generation. Norma is experimentally evaluated, demonstrating the practical support for the modelers, and the effectiveness of the underlying optimizations.