Fooling Object Detectors: Adversarial Attacks by Half-Neighbor Masks
This work addresses the problem of adversarial attacks on object detection systems, which is a critical security concern for users of these systems.
This paper proposes a Half-Neighbor Masked Projected Gradient Descent (HNM-PGD) attack to generate strong perturbations that can fool various object detection systems. The method achieved a top 1% ranking in the CIKM 2020 AnalytiCup Competition.
Although there are a great number of adversarial attacks on deep learning based classifiers, how to attack object detection systems has been rarely studied. In this paper, we propose a Half-Neighbor Masked Projected Gradient Descent (HNM-PGD) based attack, which can generate strong perturbation to fool different kinds of detectors under strict constraints. We also applied the proposed HNM-PGD attack in the CIKM 2020 AnalytiCup Competition, which was ranked within the top 1% on the leaderboard. We release the code at https://github.com/YanghaoZYH/HNM-PGD.