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

We present in this paper a novel query formulation using dynamic anchor boxesfor DETR (DEtection TRansformer) and offer a deeper understanding of the roleof queries in DETR. This new formulation directly uses box coordinates asqueries in Transformer decoders and dynamically updates them layer-by-layer.Using box coordinates not only helps using explicit positional priors toimprove the query-to-feature similarity and eliminate the slow trainingconvergence issue in DETR, but also allows us to modulate the positionalattention map using the box width and height information. Such a design makesit clear that queries in DETR can be implemented as performing soft ROI poolinglayer-by-layer in a cascade manner. As a result, it leads to the bestperformance on MS-COCO benchmark among the DETR-like detection models under thesame setting, e.g., AP 45.7\% using ResNet50-DC5 as backbone trained in 50epochs. We also conducted extensive experiments to confirm our analysis andverify the effectiveness of our methods. Code is available at\url{https://github.com/SlongLiu/DAB-DETR}.

DAB-DETR: Dynamic Anchor Boxes are Better Queries for DETR