Geo-spatial Artificial Intelligence System for Supporting Land Use Land Cover in Smart Cities Decision Mansoura City Use Case Study الذكاء الاصطناعى وعلوم البيانات المكانية ودورها فى علاج مشكلات النقل الحضرى فى مدينة المنصورة | ||
| مجلة کلية الآداب (الزقازيق) | ||
| Volume 50, Issue 114, July 2025 | ||
| Document Type: أبحاث أکادیمیة | ||
| DOI: 10.21608/artzag.2025.453511 | ||
| Authors | ||
| HANI ABDELBASSET ElSHAFEI1; Shohdi Abdelhamid Abdelkader Alkhwaga2; Mohamed Sabry Fouad Sariah3; Asmaa NematAllah Abdelshafi Alshrabasy4 | ||
| 1Faculty of Arts - Mansoura University | ||
| 2Professor of economic geography and geographic information systems Department of Geography and Geographic Information Systems - Faculty of Arts - Mansoura University | ||
| 3Department of computers and automatic control systems Mansoura University - Faculty of Engineering | ||
| 4Department of Geography and Geographic Information Systems Mansoura University - Faculty of Arts | ||
| Abstract | ||
| Smart cities are rapidly evolving environments that require innovative tools for sustainable urban development and land management. One of the key challenges in this context is understanding and monitoring Land Use Land Cover (LULC) changes to inform planning and policy decisions. Traditional methods of analyzing spatial data are often time-consuming, fragmented, and inaccessible to non-technical decision-makers. To address these limitations, this project introduces an AI-powered Geo-Spatial Assistant that leverages modern artificial intelligence, remote sensing, and natural language technologies. At the core of the system is a two-step approach that ensures robust, actionable geospatial intelligence. In the first step, a DeepLabV3+ semantic segmentation model is developed and trained to perform high-resolution LULC classification using satellite imagery. This model automates the detection of land cover patterns—such as urban expansion, vegetation areas, and water body distribution—providing accurate, up-to-date spatial data for urban planners. The second step integrates the classified LULC data into a Retrieval-Augmented Generation (RAG) pipeline, which allows users to pose natural language questions about spatial dynamics and receive AI-generated, context-rich responses. This empowers city planners, policymakers, and researchers to ask complex questions—like “Which areas of Mansoura are experiencing the highest urban sprawl?”—and instantly receive insights supported by both geospatial layers and retrieved domain knowledge. The complete platform is deployed as a full-stack FastAPI web application, offering interactive map visualization, intelligent Q&A, and comprehensive reporting capabilities. A case study in Mansoura City demonstrates how the system supports real-world smart city initiatives by delivering timely, data-driven recommendations. This two-step AI-enhanced workflow significantly improves accessibility to geospatial analysis, paving the way for more sustainable, evidence-based decision-making in urban governance. | ||
|
Statistics Article View: 80 |
||
