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

Omnidirectional 360{\deg} camera proliferates rapidly for autonomous robotssince it significantly enhances the perception ability by widening the field ofview(FoV). However, corresponding 360{\deg} depth sensors, which are alsocritical for the perception system, are still difficult or expensive to have.In this paper, we propose a low-cost 3D sensing system that combines anomnidirectional camera with a calibrated projective depth camera, where thedepth from the limited FoV can be automatically extended to the rest of therecorded omnidirectional image. To accurately recover the missing depths, wedesign an omnidirectional depth extension convolutional neuralnetwork(ODE-CNN), in which a spherical feature transform layer(SFTL) isembedded at the end of feature encoding layers, and a deformable convolutionalspatial propagation network(D-CSPN) is appended at the end of feature decodinglayers. The former resamples the neighborhood of each pixel in theomnidirectional coordination to the projective coordination, which reduces thedifficulty of feature learning, and the later automatically finds a propercontext to well align the structures in the estimated depths via CNN w.r.t. thereference image, which significantly improves the visual quality. Finally, wedemonstrate the effectiveness of proposed ODE-CNN over the popular 360D datasetand show that ODE-CNN significantly outperforms (relatively 33% reductionin-depth error) other state-of-the-art (SoTA) methods.

ODE-CNN: Omnidirectional Depth Extension Networks