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

The recent trend for multi-camera 3D object detection is through the unifiedbird's-eye view (BEV) representation. However, directly transforming featuresextracted from the image-plane view to BEV inevitably results in featuredistortion, especially around the objects of interest, making the objects blurinto the background. To this end, we propose OA-BEV, a network that can beplugged into the BEV-based 3D object detection framework to bring out theobjects by incorporating object-aware pseudo-3D features and depth features.Such features contain information about the object's position and 3Dstructures. First, we explicitly guide the network to learn the depthdistribution by object-level supervision from each 3D object's center. Then, weselect the foreground pixels by a 2D object detector and project them into 3Dspace for pseudo-voxel feature encoding. Finally, the object-aware depthfeatures and pseudo-voxel features are incorporated into the BEV representationwith a deformable attention mechanism. We conduct extensive experiments on thenuScenes dataset to validate the merits of our proposed OA-BEV. Our methodachieves consistent improvements over the BEV-based baselines in terms of bothaverage precision and nuScenes detection score. Our codes will be published.

OA-BEV: Bringing Object Awareness to Bird's-Eye-View Representation for Multi-Camera 3D Object Detection