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

How to effectively represent camera pose is an essential problem in 3Dcomputer vision, especially in tasks such as camera pose regression and novelview synthesis. Traditionally, 3D position of the camera is represented byCartesian coordinate and the orientation is represented by Euler angle orquaternions. These representations are manually designed, which may not be themost effective representation for downstream tasks. In this work, we propose anapproach to learn neural representations of camera poses and 3D scenes, coupledwith neural representations of local camera movements. Specifically, the camerapose and 3D scene are represented as vectors and the local camera movement isrepresented as a matrix operating on the vector of the camera pose. Wedemonstrate that the camera movement can further be parametrized by a matrixLie algebra that underlies a rotation system in the neural space. The vectorrepresentations are then concatenated and generate the posed 2D image through adecoder network. The model is learned from only posed 2D images andcorresponding camera poses, without access to depths or shapes. We conductextensive experiments on synthetic and real datasets. The results show thatcompared with other camera pose representations, our learned representation ismore robust to noise in novel view synthesis and more effective in camera poseregression.

Learning Neural Representation of Camera Pose with Matrix Representation of Pose Shift via View Synthesis