Across four benchmark datasets and 25 vision-language models, GPT-4.1 is the most accurate geolocator, reaching 61% Recall@1km on social-media-like images while all models struggle on street-level imagery.
G^3: Geolocation via Guidebook Grounding
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
We demonstrate how language can improve geolocation: the task of predicting the location where an image was taken. Here we study explicit knowledge from human-written guidebooks that describe the salient and class-discriminative visual features humans use for geolocation. We propose the task of Geolocation via Guidebook Grounding that uses a dataset of StreetView images from a diverse set of locations and an associated textual guidebook for GeoGuessr, a popular interactive geolocation game. Our approach predicts a country for each image by attending over the clues automatically extracted from the guidebook. Supervising attention with country-level pseudo labels achieves the best performance. Our approach substantially outperforms a state-of-the-art image-only geolocation method, with an improvement of over 5% in Top-1 accuracy. Our dataset and code can be found at https://github.com/g-luo/geolocation_via_guidebook_grounding.
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Assessing the Geolocation Capabilities, Limitations and Societal Risks of Generative Vision-Language Models
Across four benchmark datasets and 25 vision-language models, GPT-4.1 is the most accurate geolocator, reaching 61% Recall@1km on social-media-like images while all models struggle on street-level imagery.