AI Integrated Web Application Development for OSM Change Detection: A Case Study of Luxembourg,Western Europe

Abstract. The rapid pace of urban growth demands continuous observation, especially in compact, high-development areas like Luxembourg. To effectively analyse land use transformations, scalable tools and an open, data-driven approach are essential for maintaining accuracy and clear visualization. The traditional methods of using remotely sensed data are bulky and difficult to work with, which requires a more easy and dynamic aspect. The present research introduces an advanced approach by creating an SQL algorithm to show the change detection of land use categories for 2016 and 2024 considering built-up areas. An AI-powered web application designed to detect and interpret changes in built-up using Open Street Map (OSM) data. By incorporating modern technologies with AI, particularly Large Language Model (LLMs), this platform provides a simple interface for retrieval of statistical data, also querying and analysing the spatial trends. The accuracy of built-up changes derived from OSM was validated by comparing them with a supervised Land Use Land Cover (LULC) classification generated from Sentinel-2 imagery. One of the most significant findings is that OSM data proved remarkably accurate, aligning closely with classified land-use maps derived from Sentinel-2 satellite imagery. OSM demonstrated exceptional detail in capturing urban features such as building outlines, road networks, and commercial zones with high spatial fidelity. The validation between Sentinel-2 imagery and OSM derived data strengthened confidence with an overall accuracy of 92.4% and a kappa coefficient of 0.89. Thus, despite its crowdsourced ori gins, OSM proves that it can be a reliable source for temporal land-use monitoring when properly validated and visualized.