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

The goal of this paper is to estimate the viewpoint for a novel object.Standard viewpoint estimation approaches generally fail on this task due totheir reliance on a 3D model for alignment or large amounts of class-specifictraining data and their corresponding canonical pose. We overcome thoselimitations by learning a reconstruct and align approach. Our key insight isthat although we do not have an explicit 3D model or a predefined canonicalpose, we can still learn to estimate the object's shape in the viewer's frameand then use an image to provide our reference model or canonical pose. Inparticular, we propose learning two networks: the first maps images to a 3Dgeometry-aware feature bottleneck and is trained via an image-to-imagetranslation loss; the second learns whether two instances of features arealigned. At test time, our model finds the relative transformation that bestaligns the bottleneck features of our test image to a reference image. Weevaluate our method on novel object viewpoint estimation by generalizing acrossdifferent datasets, analyzing the impact of our different modules, andproviding a qualitative analysis of the learned features to identify whatrepresentations are being learnt for alignment.

Novel Object Viewpoint Estimation through Reconstruction Alignment