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 dominant multi-camera 3D detection paradigm is based on explicit 3Dfeature construction, which requires complicated indexing of local image-viewfeatures via 3D-to-2D projection. Other methods implicitly introduce geometricpositional encoding and perform global attention (e.g., PETR) to build therelationship between image tokens and 3D objects. The 3D-to-2D perspectiveinconsistency and global attention lead to a weak correlation betweenforeground tokens and queries, resulting in slow convergence. We proposeFocal-PETR with instance-guided supervision and spatial alignment module toadaptively focus object queries on discriminative foreground regions.Focal-PETR additionally introduces a down-sampling strategy to reduce theconsumption of global attention. Due to the highly parallelized implementationand down-sampling strategy, our model, without depth supervision, achievesleading performance on the large-scale nuScenes benchmark and a superior speedof 30 FPS on a single RTX3090 GPU. Extensive experiments show that our methodoutperforms PETR while consuming 3x fewer training hours. The code will be madepublicly available.

Focal-PETR: Embracing Foreground for Efficient Multi-Camera 3D Object Detection