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

Visual localization is a core component in many applications, includingaugmented reality (AR). Localization algorithms compute the camera pose of aquery image w.r.t. a scene representation, which is typically built fromimages. This often requires capturing and storing large amounts of data,followed by running Structure-from-Motion (SfM) algorithms. An interesting, andunderexplored, source of data for building scene representations are 3D modelsthat are readily available on the Internet, e.g., hand-drawn CAD models, 3Dmodels generated from building footprints, or from aerial images. These modelsallow to perform visual localization right away without the time-consumingscene capturing and model building steps. Yet, it also comes with challenges asthe available 3D models are often imperfect reflections of reality. E.g., themodels might only have generic or no textures at all, might only provide asimple approximation of the scene geometry, or might be stretched. This paperstudies how the imperfections of these models affect localization accuracy. Wecreate a new benchmark for this task and provide a detailed experimentalevaluation based on multiple 3D models per scene. We show that 3D models fromthe Internet show promise as an easy-to-obtain scene representation. At thesame time, there is significant room for improvement for visual localizationpipelines. To foster research on this interesting and challenging task, werelease our benchmark at v-pnk.github.io/cadloc.

Visual Localization using Imperfect 3D Models from the Internet