GeoFidelity-Bench: Evaluating Segment-Level Geographic Fidelity in Text-to-Image Street-View Generation
2026-06-22 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors created GeoFidelity-Bench, a test to see if text-to-image models can generate images that match specific road segments rather than just general city scenes. They found that adding street and neighborhood names helps models produce more locally plausible images, but it does not significantly improve matching exact road segments. Using GPS coordinates in prompts did not provide clear benefits. Real images confirmed that the test can detect true segment differences. Overall, the study shows current models struggle to generate faithful images for very specific streets.
text-to-image modelsstreet-view generationGeoFidelity-BenchMapillary imagesOpenStreetMapgeographic fidelitycity-level contextGPS coordinatesimage retrievallocal plausibility
Authors
Kaizhen Tan, Hanzhe Hong, Siru Tao
Abstract
Text-to-image models can generate visually plausible city streets, but whether their outputs correspond to a requested road segment rather than a generic city prior remains unclear. We introduce GeoFidelity-Bench, a reference-panel benchmark for segment-conditioned geographic fidelity in street-view generation. It contains 7,117 curated Mapillary images covering 109 named OpenStreetMap road segments in 25 cities across six continents. For each generated panel, the benchmark ranks the target reference panel against panels from the nearest segment in the same city, other segments in the same city, and segments from other cities, making local discrimination rather than absolute target similarity the primary test. We evaluate six open-weight text-to-image generators under city-only, street-and-neighborhood, and GPS-augmented prompts. Adding street and neighborhood names is associated with an increase of 5.5 percentage points in top-1 retrieval accuracy over city-only prompts, with a 95% confidence interval from 3.4 to 7.7 percentage points. However, the similarity margin between the target and the nearest segment in the same city remains near zero, indicating that local names improve broad local plausibility more than exact segment identity. Prompts that keep the city fixed but use incorrect street or neighborhood names further show that only part of the gain depends on the correct local names, while appending raw GPS coordinates as ordinary text yields no statistically clear additional benefit. Held-out real-image queries successfully recover segment identity, showing that the curated references contain usable segment-level signal. GeoFidelity-Bench thus reveals a persistent gap between city- or neighborhood-plausible street-view generation and faithful generation for a specific road segment.