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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Recently, the pure camera-based Bird's-Eye-View (BEV) perception provides afeasible solution for economical autonomous driving. However, the existingBEV-based multi-view 3D detectors generally transform all image features intoBEV features, without considering the problem that the large proportion ofbackground information may submerge the object information. In this paper, wepropose Semantic-Aware BEV Pooling (SA-BEVPool), which can filter outbackground information according to the semantic segmentation of image featuresand transform image features into semantic-aware BEV features. Accordingly, wepropose BEV-Paste, an effective data augmentation strategy that closely matcheswith semantic-aware BEV feature. In addition, we design a Multi-ScaleCross-Task (MSCT) head, which combines task-specific and cross-task informationto predict depth distribution and semantic segmentation more accurately,further improving the quality of semantic-aware BEV feature. Finally, weintegrate the above modules into a novel multi-view 3D object detectionframework, namely SA-BEV. Experiments on nuScenes show that SA-BEV achievesstate-of-the-art performance. Code has been available athttps://github.com/mengtan00/SA-BEV.git.

SA-BEV: Generating Semantic-Aware Bird's-Eye-View Feature for Multi-view 3D Object Detection