GeoBridge: Decoupled Semantic Conditioning for Generative Image Geolocalization

2026-08-12

AI Summary

This work addresses the information loss and coarse granularity inherent in existing image geolocation methods when decoding multimodal semantic reasoning into continuous geographic coordinates. The authors propose GeoBridge, a novel framework that bridges a frozen multimodal large language model with a Riemannian spherical manifold head through a role-decoupled conditional mechanism, enabling geometrically consistent generative coordinate decoding. By explicitly decoupling semantic supervision from the conditioning interface, GeoBridge avoids interference from discrete labels on the continuous spherical representation and facilitates the integration of multi-granularity geographic priors. Evaluated on IM2GPS3K, the method achieves state-of-the-art accuracy rates of 38.67%, 52.89%, and 70.37% at 25/200/750 km thresholds, respectively, significantly outperforming current baselines.

Abstract

Multimodal large language models (MLLMs) have advanced image geolocalization mainly by improving how they reason about geographic cues. How that reasoning isdecoded into coordinates, however, has lagged behind. Predicting a place name for a geocoding API is discrete and lossy: it ignores image evidence and collapses multi-granular semantics into a coarse lookup. We argue that the bottleneck has shifted from what a model reasons to how that reasoning is represented for a continuous, geometry-aware decoder. We present GeoBridge, a role-decoupled conditioning mechanism that connects a frozen semantic MLLM to a frozen Riemannian flow-matching head that generates coordinates on the sphere. The central obstacle is arole conflict: supervising the condition with discrete semantic labels biases its representation toward class-discriminative geometry, at odds with the smooth manifold the generative head requires. GeoBridge keeps the semantic supervision decoupled from the condition interface: a separate projection forms the continuous condition the frozen head expects, injecting geographic priors without disturbing the spherical decoder. On IM2GPS3K, GeoBridge reaches 38.67/52.89/70.37 at the 25/200/750 km thresholds, improving over a place-name-to-API pipeline and reasoning-augmented direct prediction at these precision-relevant scales. GeoBridge is a decode-side algorithmic contribution, orthogonal and complementary to chain-of-thought reasoning. Code will be made publicly available.

Problem

• geolocalization
• semantic conditioning
• coordinate decoding
• multimodal reasoning
• continuous representation

Innovation

• decoupled semantic conditioning
• Riemannian flow-matching
• generative geolocalization
• spherical manifold
• frozen MLLM