Multimodal explanation of urban growth predictions using Vision-Language Models

By Giovanna Venuti / 23/04/2026 

Teacher: Giovanna Venuti

Supervisor: Ing. Gianluca Murdaca – MindEarth

Description: The project aims to develop a prototype for the automatic explanation of urban growth predictions generated by machine learning models on geospatial data. Starting from historical built-up maps, probability maps of future growth, and simple morphological indicators, the student will create a pipeline capable of producing structured textual explanations about the location, shape, and spatial coherence of the predicted growth areas.

Difficult level: Medium – High

Requirements: Python, basic GIS, raster analysis, machine learning,
interest in explainable AI and multimodal model

Scroll to Top