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

We introduce FocalPose++, a neural render-and-compare method for jointlyestimating the camera-object 6D pose and camera focal length given a single RGBinput image depicting a known object. The contributions of this work arethreefold. First, we derive a focal length update rule that extends an existingstate-of-the-art render-and-compare 6D pose estimator to address the jointestimation task. Second, we investigate several different loss functions forjointly estimating the object pose and focal length. We find that a combinationof direct focal length regression with a reprojection loss disentangling thecontribution of translation, rotation, and focal length leads to improvedresults. Third, we explore the effect of different synthetic training data onthe performance of our method. Specifically, we investigate differentdistributions used for sampling object's 6D pose and camera's focal length whenrendering the synthetic images, and show that parametric distribution fitted onreal training data works the best. We show results on three challengingbenchmark datasets that depict known 3D models in uncontrolled settings. Wedemonstrate that our focal length and 6D pose estimates have lower error thanthe existing state-of-the-art methods.

FocalPose++: Focal Length and Object Pose Estimation via Render and Compare