Zero-shot Vision-Language Reranking for Cross-View Geolocalization

ArXi:2603.27251v1 Announce Type: cross Cross-view geolocalization (CVGL) systems, while effective at retrieving a list of relevant candidates (high Recall), often fail to identify the single best match (low Top-1 accuracy). This work investigates the use of zero-shot Vision-Language Models (VLMs) as rerankers to address this gap. We propose a two-stage framework: state-of-the-art (SOTA) retrieval followed by VLM reranking. We systematically compare two strategies: (1) Pointwise (scoring candidates individually) and (2) Pairwise (comparing candidates relatively.