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 ability to detect objects in images at varying scales has played apivotal role in the design of modern object detectors. Despite considerableprogress in removing hand-crafted components and simplifying the architecturewith transformers, multi-scale feature maps and/or pyramid design remain a keyfactor for their empirical success. In this paper, we show that this relianceon either feature pyramids or an hierarchical backbone is unnecessary and atransformer-based detector with scale-aware attention enables the plaindetector `SimPLR' whose backbone and detection head are both non-hierarchicaland operate on single-scale features. The plain architecture allows SimPLR toeffectively take advantages of self-supervised learning and scaling approacheswith ViTs, yielding competitive performance compared to hierarchical andmulti-scale counterparts. We demonstrate through our experiments that whenscaling to larger ViT backbones, SimPLR indicates better performance thanend-to-end segmentation models (Mask2Former) and plain-backbone detectors(ViTDet), while consistently being faster. The code will be released.

SimPLR: A Simple and Plain Transformer for Object Detection and Segmentation