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 Transformer-based 3D object detectors learn point cloud featureseither from point- or voxel-based representations. However, the former requirestime-consuming sampling while the latter introduces quantization errors. Inthis paper, we present a novel Point-Voxel Transformer for single-stage 3Ddetection (PVT-SSD) that takes advantage of these two representations.Specifically, we first use voxel-based sparse convolutions for efficientfeature encoding. Then, we propose a Point-Voxel Transformer (PVT) module thatobtains long-range contexts in a cheap manner from voxels while attainingaccurate positions from points. The key to associating the two differentrepresentations is our introduced input-dependent Query Initialization module,which could efficiently generate reference points and content queries. Then,PVT adaptively fuses long-range contextual and local geometric informationaround reference points into content queries. Further, to quickly find theneighboring points of reference points, we design the Virtual Range Imagemodule, which generalizes the native range image to multi-sensor andmulti-frame. The experiments on several autonomous driving benchmarks verifythe effectiveness and efficiency of the proposed method. Code will be availableat https://github.com/Nightmare-n/PVT-SSD.

PVT-SSD: Single-Stage 3D Object Detector with Point-Voxel Transformer