Search papers, labs, and topics across Lattice.
This paper introduces a novel black-box attack method called \method{}, which effectively generates budget-aware adversarial patches by optimizing their location, texture, and size while considering query constraints. The approach employs a Contextual Thompson-Sampling strategy for efficient patch placement and NES-style pixel updates, allowing for dynamic growth of the patch only when progress stalls. Evaluations demonstrate that \method{} achieves significant suppression of object detection models, particularly highlighting the trade-offs between query budgets and visual footprint, thereby providing actionable insights into adversarial patch effectiveness across various detectors.
Adversarial patches can be optimized for maximum impact with minimal queries, revealing critical trade-offs in their design and effectiveness against object detectors.
Adversarial patches pose a practical threat to modern object detectors. Prior work shows vulnerability, but three gaps limit actionable insight: (i) few \emph{score-based black-box} attacks \emph{jointly} optimize patch \emph{location, texture, and size} under tight query budgets; (ii) success is rarely tied to the patch's \emph{visual footprint}; and (iii) evaluations often conflate EOT robustness with plain-view suppression. We present \method{}, a query-efficient, budget-adaptive black-box attack that couples a lightweight \emph{Contextual Thompson-Sampling} placer with NES-style pixel updates, growing the patch only when progress stalls. Reporting is anchored by a \emph{strict plain-image} suppression test; EOT is audited but never used as a substitute for success, and optional appearance/printability weights expose strength--visibility trade-offs. Across YOLOv5, Faster R-CNN, and YOLOS, \method{} achieves strong suppression on CNN-based detectors and substantial suppression on the transformer-based detector, using compact patches and exposing clear query--footprint trade-offs relative to fixed-size and heuristic baselines. A print--capture pilot further shows transfer across unseen physical objects and viewpoints.
