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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Given a single scene image, this paper proposes a method of Category-level 6DObject Pose and Size Estimation (COPSE) from the point cloud of the targetobject, without external real pose-annotated training data. Specifically,beyond the visual cues in RGB images, we rely on the shape informationpredominately from the depth (D) channel. The key idea is to explore the shapealignment of each instance against its corresponding category-level templateshape, and the symmetric correspondence of each object category for estimatinga coarse 3D object shape. Our framework deforms the point cloud of thecategory-level template shape to align the observed instance point cloud forimplicitly representing its 3D rotation. Then we model the symmetriccorrespondence by predicting symmetric point cloud from the partially observedpoint cloud. The concatenation of the observed point cloud and symmetric onereconstructs a coarse object shape, thus facilitating object center (3Dtranslation) and 3D size estimation. Extensive experiments on thecategory-level NOCS benchmark demonstrate that our lightweight model stillcompetes with state-of-the-art approaches that require labeled real-worldimages. We also deploy our approach to a physical Baxter robot to performgrasping tasks on unseen but category-known instances, and the results furthervalidate the efficacy of our proposed model. Code and pre-trained models areavailable on the project webpage.

SAR-Net: Shape Alignment and Recovery Network for Category-level 6D Object Pose and Size Estimation