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

Learning neural implicit representations has achieved remarkable performancein 3D reconstruction from multi-view images. Current methods use volumerendering to render implicit representations into either RGB or depth imagesthat are supervised by multi-view ground truth. However, rendering a view eachtime suffers from incomplete depth at holes and unawareness of occludedstructures from the depth supervision, which severely affects the accuracy ofgeometry inference via volume rendering. To resolve this issue, we propose tolearn neural implicit representations from multi-view RGBD images throughvolume rendering with an attentive depth fusion prior. Our prior allows neuralnetworks to perceive coarse 3D structures from the Truncated Signed DistanceFunction (TSDF) fused from all depth images available for rendering. The TSDFenables accessing the missing depth at holes on one depth image and theoccluded parts that are invisible from the current view. By introducing a novelattention mechanism, we allow neural networks to directly use the depth fusionprior with the inferred occupancy as the learned implicit function. Ourattention mechanism works with either a one-time fused TSDF that represents awhole scene or an incrementally fused TSDF that represents a partial scene inthe context of Simultaneous Localization and Mapping (SLAM). Our evaluations onwidely used benchmarks including synthetic and real-world scans show oursuperiority over the latest neural implicit methods. Project page:https://machineperceptionlab.github.io/Attentive_DF_Prior/

Learning Neural Implicit through Volume Rendering with Attentive Depth Fusion Priors