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

Roadside camera-driven 3D object detection is a crucial task in intelligenttransportation systems, which extends the perception range beyond thelimitations of vision-centric vehicles and enhances road safety. While previousstudies have limitations in using only depth or height information, we findboth depth and height matter and they are in fact complementary. The depthfeature encompasses precise geometric cues, whereas the height feature isprimarily focused on distinguishing between various categories of heightintervals, essentially providing semantic context. This insight motivates thedevelopment of Complementary-BEV (CoBEV), a novel end-to-end monocular 3Dobject detection framework that integrates depth and height to construct robustBEV representations. In essence, CoBEV estimates each pixel's depth and heightdistribution and lifts the camera features into 3D space for lateral fusionusing the newly proposed two-stage complementary feature selection (CFS)module. A BEV feature distillation framework is also seamlessly integrated tofurther enhance the detection accuracy from the prior knowledge of thefusion-modal CoBEV teacher. We conduct extensive experiments on the public 3Ddetection benchmarks of roadside camera-based DAIR-V2X-I and Rope3D, as well asthe private Supremind-Road dataset, demonstrating that CoBEV not only achievesthe accuracy of the new state-of-the-art, but also significantly advances therobustness of previous methods in challenging long-distance scenarios and noisycamera disturbance, and enhances generalization by a large margin inheterologous settings with drastic changes in scene and camera parameters. Forthe first time, the vehicle AP score of a camera model reaches 80% onDAIR-V2X-I in terms of easy mode. The source code will be made publiclyavailable at https://github.com/MasterHow/CoBEV.

CoBEV: Elevating Roadside 3D Object Detection with Depth and Height Complementarity