Get object pose from one input image and a reference using 3D hypothesis verification with Zhao
Get object pose from one input image and a reference using 3D hypothesis verification with Zhao
3D-Aware Hypothesis & Verification for Generalizable Relative Object Pose Estimation
arXiv paper abstract https://arxiv.org/abs/2310.03534
arXiv PDF paper https://browse.arxiv.org/pdf/2310.03534.pdf
Prior methods … of generalizable object pose estimation highly rely on having dense views of the unseen object. By contrast … address … where only a single reference view of the object is available.
… goal then is to estimate the relative object pose between this reference view and a query image that depicts the object in a different pose.
In this scenario, robust generalization is imperative due to the presence of unseen objects during testing and the large-scale object pose variation between the reference and the query.
… present a new hypothesis-and-verification framework, in which … generate and evaluate multiple pose hypotheses, ultimately selecting the most reliable one as the relative object pose.
To measure reliability, … introduce a 3D-aware verification that explicitly applies 3D transformations to the 3D object representations learned from the two input images.
… superior accuracy of … approach in relative pose estimation and its robustness in large-scale pose variations, when dealing with unseen objects.
Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website
LinkedIn https://www.linkedin.com/in/morris-lee-47877b7b
