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

Detection and tracking of moving objects is an essential component inenvironmental perception for autonomous driving. In the flourishing field ofmulti-view 3D camera-based detectors, different transformer-based pipelines aredesigned to learn queries in 3D space from 2D feature maps of perspectiveviews, but the dominant dense BEV query mechanism is computationallyinefficient. This paper proposes Sparse R-CNN 3D (SRCN3D), a novel two-stagefully-sparse detector that incorporates sparse queries, sparse attention withbox-wise sampling, and sparse prediction. SRCN3D adopts a cascade structurewith the twin-track update of both a fixed number of query boxes and latentquery features. Our novel sparse feature sampling module only utilizes local 2Dregion of interest (RoI) features calculated by the projection of 3D queryboxes for further box refinement, leading to a fully-convolutional anddeployment-friendly pipeline. For multi-object tracking, motion features, queryfeatures and RoI features are comprehensively utilized in multi-hypotheses dataassociation. Extensive experiments on nuScenes dataset demonstrate that SRCN3Dachieves competitive performance in both 3D object detection and multi-objecttracking tasks, while also exhibiting superior efficiency compared totransformer-based methods. Code and models are available athttps://github.com/synsin0/SRCN3D.

SRCN3D: Sparse R-CNN 3D for Compact Convolutional Multi-View 3D Object Detection and Tracking