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

3D semantic scene completion (SSC) is an ill-posed perception task thatrequires inferring a dense 3D scene from limited observations. Previouscamera-based methods struggle to predict accurate semantic scenes due toinherent geometric ambiguity and incomplete observations. In this paper, weresort to stereo matching technique and bird's-eye-view (BEV) representationlearning to address such issues in SSC. Complementary to each other, stereomatching mitigates geometric ambiguity with epipolar constraint while BEVrepresentation enhances the hallucination ability for invisible regions withglobal semantic context. However, due to the inherent representation gapbetween stereo geometry and BEV features, it is non-trivial to bridge them fordense prediction task of SSC. Therefore, we further develop a unifiedoccupancy-based framework dubbed BRGScene, which effectively bridges these tworepresentations with dense 3D volumes for reliable semantic scene completion.Specifically, we design a novel Mutual Interactive Ensemble (MIE) block forpixel-level reliable aggregation of stereo geometry and BEV features. Withinthe MIE block, a Bi-directional Reliable Interaction (BRI) module, enhancedwith confidence re-weighting, is employed to encourage fine-grained interactionthrough mutual guidance. Besides, a Dual Volume Ensemble (DVE) module isintroduced to facilitate complementary aggregation through channel-wiserecalibration and multi-group voting. Our method outperforms all publishedcamera-based methods on SemanticKITTI for semantic scene completion.

Bridging Stereo Geometry and BEV Representation with Reliable Mutual Interaction for Semantic Scene Completion