Adversarial Patch

About

We present a method to create universal, robust, targeted adversarial image patches in the real world. The patches are universal because they can be used to attack any scene, robust because they work under a wide variety of transformations, and targeted because they can cause a classifier to output any target class. These adversarial patches can be printed, added to any scene, photographed, and presented to image classifiers; even when the patches are small, they cause the classifiers to ignore the other items in the scene and report a chosen target class. To reproduce the results from the paper, our code is available at https://github.com/tensorflow/cleverhans/tree/master/examples/adversarial_patch

Tom B. Brown, Dandelion Mané, Aurko Roy, Martín Abadi, Justin Gilmer · 2017

Related benchmarks

Task	Dataset	Result
Untargeted Attack	Tongji (test)	ASR100	56
Untargeted Adversarial Attack	AISEC (test)	ASR97.9	56
Targeted Adversarial Attack	AISEC	ASR79.87	56
Targeted Attack	Tongji (test)	ASR99.51	56
Untargeted Attack	IITD	Attack Success Rate78.61	56
Targeted Attack	IITD (test)	ASR20.2	56
Image Classification	Classification Dataset	Top-1 Accuracy98	5
Object Detection	SAR-Ship-Dataset (test)	mAP5064	5