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 latest generation of transformer-based vision models has proven to besuperior to Convolutional Neural Network (CNN)-based models across severalvision tasks, largely attributed to their remarkable prowess in relationmodeling. Deformable vision transformers significantly reduce the quadraticcomplexity of attention modeling by using sparse attention structures, enablingthem to incorporate features across different scales and be used in large-scaleapplications, such as multi-view vision systems. Recent work has demonstratedadversarial attacks against conventional vision transformers; we show thatthese attacks do not transfer to deformable transformers due to their sparseattention structure. Specifically, attention in deformable transformers ismodeled using pointers to the most relevant other tokens. In this work, wecontribute for the first time adversarial attacks that manipulate the attentionof deformable transformers, redirecting it to focus on irrelevant parts of theimage. We also develop new collaborative attacks where a source patchmanipulates attention to point to a target patch, which contains theadversarial noise to fool the model. In our experiments, we observe thataltering less than 1% of the patched area in the input field results in acomplete drop to 0% AP in single-view object detection using MS COCO and a 0%MODA in multi-view object detection using Wildtrack.

Attention Deficit is Ordered! Fooling Deformable Vision Transformers with Collaborative Adversarial Patches