Street-Level Geolocalization Using Multimodal Large Language Models and Retrieval-Augmented Generation

Abstract

Street-level geolocalization is a challenging task requiring the fusion of visual landscape clues with spatial reference data. In this paper, we propose a novel multimodal framework that integrates state-of-the-art Vision-Language Models (VLMs) with Retrieval-Augmented Generation (RAG) to localize street-level imagery. By combining hierarchical spatial indexing and visual semantics, our approach significantly improves localization precision and provides interpretable reasoning traces for geographical predictions.

Yunus Serhat Bıçakçı

Assistant Professor

Assistant Professor specializing in GeoAI, Multimodal Vision-Language Models, and Spatial Data Science.