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

Accurately estimating the 6D pose of objects is crucial for manyapplications, such as robotic grasping, autonomous driving, and augmentedreality. However, this task becomes more challenging in poor lightingconditions or when dealing with textureless objects. To address this issue,depth images are becoming an increasingly popular choice due to theirinvariance to a scene's appearance and the implicit incorporation of essentialgeometric characteristics. However, fully leveraging depth information toimprove the performance of pose estimation remains a difficult andunder-investigated problem. To tackle this challenge, we propose a novelframework called SwinDePose, that uses only geometric information from depthimages to achieve accurate 6D pose estimation. SwinDePose first calculates theangles between each normal vector defined in a depth image and the threecoordinate axes in the camera coordinate system. The resulting angles are thenformed into an image, which is encoded using Swin Transformer. Additionally, weapply RandLA-Net to learn the representations from point clouds. The resultingimage and point clouds embeddings are concatenated and fed into a semanticsegmentation module and a 3D keypoints localization module. Finally, weestimate 6D poses using a least-square fitting approach based on the targetobject's predicted semantic mask and 3D keypoints. In experiments on theLineMod and Occlusion LineMod datasets, SwinDePose outperforms existingstate-of-the-art methods for 6D object pose estimation using depth images. Thisdemonstrates the effectiveness of our approach and highlights its potential forimproving performance in real-world scenarios. Our code is athttps://github.com/zhujunli1993/SwinDePose.

Depth-based 6DoF Object Pose Estimation using Swin Transformer