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 research shows that neural networks models used for computer vision(e.g., YOLO and Fast R-CNN) are vulnerable to adversarial evasion attacks. Mostof the existing real-world adversarial attacks against object detectors use anadversarial patch which is attached to the target object (e.g., a carefullycrafted sticker placed on a stop sign). This method may not be robust tochanges in the camera's location relative to the target object; in addition, itmay not work well when applied to nonplanar objects such as cars. In thisstudy, we present an innovative attack method against object detectors appliedin a real-world setup that addresses some of the limitations of existingattacks. Our method uses dynamic adversarial patches which are placed atmultiple predetermined locations on a target object. An adversarial learningalgorithm is applied in order to generate the patches used. The dynamic attackis implemented by switching between optimized patches dynamically, according tothe camera's position (i.e., the object detection system's position). In orderto demonstrate our attack in a real-world setup, we implemented the patches byattaching flat screens to the target object; the screens are used to presentthe patches and switch between them, depending on the current camera location.Thus, the attack is dynamic and adjusts itself to the situation to achieveoptimal results. We evaluated our dynamic patch approach by attacking theYOLOv2 object detector with a car as the target object and succeeded inmisleading it in up to 90% of the video frames when filming the car from a wideviewing angle range. We improved the attack by generating patches that considerthe semantic distance between the target object and its classification. We alsoexamined the attack's transferability among different car models and were ableto mislead the detector 71% of the time.

Dynamic Adversarial Patch for Evading Object Detection Models