English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Street-to-aerial image geo-localization, which matches a query street-viewimage to the GPS-tagged aerial images in a reference set, has attractedincreasing attention recently. In this paper, we revisit this problem and pointout the ignored issue about image alignment information. We show that theperformance of a simple Siamese network is highly dependent on the alignmentsetting and the comparison of previous works can be unfair if they havedifferent assumptions. Instead of focusing on the feature extraction under thealignment assumption, we show that improvements in metric learning techniquessignificantly boost the performance regardless of the alignment. Withoutleveraging the alignment information, our pipeline outperforms previous workson both panorama and cropped datasets. Furthermore, we conduct visualization tohelp understand the learned model and the effect of alignment information usingGrad-CAM. With our discovery on the approximate rotation-invariant activationmaps, we propose a novel method to estimate the orientation/alignment between apair of cross-view images with unknown alignment information. It achievesstate-of-the-art results on the CVUSA dataset.

Revisiting Street-to-Aerial View Image Geo-localization and Orientation Estimation