README.md

Adversarial-attack-paper-recommendation

​ [ catalogue ]

[TOC]

​ Adversarial attack is becoming a hot research area. However there are not easy to deep understand it. Now reading paper is the best way to understand adversarial attack. There's actually been a lot of papers on it. In this repositorie I will recommend some papers about different areas of adversarial attack.

Attack about object detectors

• Exploring the vulnerability of single shot module in object detectors via imperceptible background patches.

  Paper url: https://arxiv.org/abs/1809.05966

• Dpatch: An adversarial patch attack on object detectors.

  Paper url: https://arxiv.org/abs/1806.02299

  Code url: https://github.com/veralauee/DPatch?utm_source=catalyzex.com

Augmentation to make attack robust

​ Different from the digital world, the attack will be disable caused by viewing angle, brightness, rotation, and so on.So there are many aumentation to make attack robust in the real world.

• Adversarial t-shirt! evading person detectors in a physical world.

  Propose TPS transforms

  Paper url: https://arxiv.org/abs/1910.11099

  Code url: https://www.catalyzex.com/redirect?url=https://github.com/jandress94/adversarial_tshirt

Attack In the real world

• Adversarial t-shirt! evading person detectors in a physical world.

  Propose TPS transforms；invite a T-shirt which can not be detected by object detectors

  Paper url: https://arxiv.org/abs/1910.11099

  Code url: https://www.catalyzex.com/redirect?url=https://github.com/jandress94/adversarial_tshirt

• Making an Invisibility Cloak: Real World Adversarial Attacks on Object Detectors

  propose a new loss function to produce patch;Print the patch on the cloth. People who wear this cloth can not be detected by object detectors in the real world

  Paper url: https://arxiv.org/abs/1810.05206

• Fooling automated surveillance cameras: adversarial patches to attack person detection

  If you want to research producing adversarial patches, you should read this paper first. This Paper are a work about producing adversarial patches. Meanwhile I think this paper's git repository is the best git repository in all of the adversarial papers.

  Paper url: https://arxiv.org/abs/1904.08653

  Code url: https://www.catalyzex.com/redirect?url=https://github.com/Zhang-Jack/adversarial_yolo2