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

Recent works succeeded to generate adversarial perturbations on the entireimage or the object of interests to corrupt CNN based object detectors. In thispaper, we focus on exploring the vulnerability of the Single Shot Module (SSM)commonly used in recent object detectors, by adding small perturbations topatches in the background outside the object. The SSM is referred to the RegionProposal Network used in a two-stage object detector or the single-stage objectdetector itself. The SSM is typically a fully convolutional neural networkwhich generates output in a single forward pass. Due to the excessiveconvolutions used in SSM, the actual receptive field is larger than the objectitself. As such, we propose a novel method to corrupt object detectors bygenerating imperceptible patches only in the background. Our method can find afew background patches for perturbation, which can effectively decrease truepositives and dramatically increase false positives. Efficacy is demonstratedon 5 two-stage object detectors and 8 single-stage object detectors on the MSCOCO 2014 dataset. Results indicate that perturbations with small distortionsoutside the bounding box of object region can still severely damage thedetection performance.

Exploring the Vulnerability of Single Shot Module in Object Detectors via Imperceptible Background Patches