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

Recent volumetric 3D reconstruction methods can produce very accurateresults, with plausible geometry even for unobserved surfaces. However, theyface an undesirable trade-off when it comes to multi-view fusion. They can fuseall available view information by global averaging, thus losing fine detail, orthey can heuristically cluster views for local fusion, thus restricting theirability to consider all views jointly. Our key insight is that greater detailcan be retained without restricting view diversity by learning a view-fusionfunction conditioned on camera pose and image content. We propose to learn thismulti-view fusion using a transformer. To this end, we introduce VoRTX, anend-to-end volumetric 3D reconstruction network using transformers forwide-baseline, multi-view feature fusion. Our model is occlusion-aware,leveraging the transformer architecture to predict an initial, projective scenegeometry estimate. This estimate is used to avoid backprojecting image featuresthrough surfaces into occluded regions. We train our model on ScanNet and showthat it produces better reconstructions than state-of-the-art methods. We alsodemonstrate generalization without any fine-tuning, outperforming the samestate-of-the-art methods on two other datasets, TUM-RGBD and ICL-NUIM.

VoRTX: Volumetric 3D Reconstruction With Transformers for Voxelwise View Selection and Fusion