Saved in:
Bibliographic Details
Main Authors: Han, Ting, Li, Fengjiao, Chen, Chunsong, Huang, Haoling, Chen, Yiping, Wu, Meiliu
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.04357
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917385706930176
author Han, Ting
Li, Fengjiao
Chen, Chunsong
Huang, Haoling
Chen, Yiping
Wu, Meiliu
author_facet Han, Ting
Li, Fengjiao
Chen, Chunsong
Huang, Haoling
Chen, Yiping
Wu, Meiliu
contents This paper proposes Spatially-Weighted CLIP (SW-CLIP), a novel framework for street-view geo-localization that explicitly incorporates spatial autocorrelation into vision-language contrastive learning. Unlike conventional CLIP-based methods that treat all non-matching samples as equally negative, SW-CLIP leverages Tobler's First Law of Geography to model geographic relationships through distance-aware soft supervision. Specifically, we introduce a location-as-text representation to encode geographic positions and replace one-hot InfoNCE targets with spatially weighted soft labels derived from geodesic distance. Additionally, a neighborhood-consistency regularization is employed to preserve local spatial structure in the embedding space. Experiments on a multi-city dataset demonstrate that SW-CLIP significantly improves geo-localization accuracy, reduces long-tail errors, and enhances spatial coherence compared to standard CLIP. The results highlight the importance of shifting from semantic alignment to geographic alignment for robust geo-localization and provide a general paradigm for integrating spatial principles into multimodal representation learning.
format Preprint
id arxiv_https___arxiv_org_abs_2604_04357
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Spatially-Weighted CLIP for Street-View Geo-localization
Han, Ting
Li, Fengjiao
Chen, Chunsong
Huang, Haoling
Chen, Yiping
Wu, Meiliu
Computer Vision and Pattern Recognition
This paper proposes Spatially-Weighted CLIP (SW-CLIP), a novel framework for street-view geo-localization that explicitly incorporates spatial autocorrelation into vision-language contrastive learning. Unlike conventional CLIP-based methods that treat all non-matching samples as equally negative, SW-CLIP leverages Tobler's First Law of Geography to model geographic relationships through distance-aware soft supervision. Specifically, we introduce a location-as-text representation to encode geographic positions and replace one-hot InfoNCE targets with spatially weighted soft labels derived from geodesic distance. Additionally, a neighborhood-consistency regularization is employed to preserve local spatial structure in the embedding space. Experiments on a multi-city dataset demonstrate that SW-CLIP significantly improves geo-localization accuracy, reduces long-tail errors, and enhances spatial coherence compared to standard CLIP. The results highlight the importance of shifting from semantic alignment to geographic alignment for robust geo-localization and provide a general paradigm for integrating spatial principles into multimodal representation learning.
title Spatially-Weighted CLIP for Street-View Geo-localization
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.04357