Object Hider: Adversarial Patch Attack Against Object Detectors

TL;DR

This paper introduces two novel adversarial patch algorithms to attack object detectors, demonstrating high effectiveness and transferability, and achieving top results in a major adversarial challenge competition.

Contribution

It proposes heatmap-based and consensus-based adversarial patch algorithms specifically targeting object detection models, advancing attack methods in this domain.

Findings

Proposed methods are highly effective against state-of-the-art detectors.

The algorithms demonstrate strong transferability and generality.

Achieved top 7 placement in a large adversarial challenge competition.

Abstract

Deep neural networks have been widely used in many computer vision tasks. However, it is proved that they are susceptible to small, imperceptible perturbations added to the input. Inputs with elaborately designed perturbations that can fool deep learning models are called adversarial examples, and they have drawn great concerns about the safety of deep neural networks. Object detection algorithms are designed to locate and classify objects in images or videos and they are the core of many computer vision tasks, which have great research value and wide applications. In this paper, we focus on adversarial attack on some state-of-the-art object detection models. As a practical alternative, we use adversarial patches for the attack. Two adversarial patch generation algorithms have been proposed: the heatmap-based algorithm and the consensus-based algorithm. The experiment results have shown that the proposed methods are highly effective, transferable and generic. Additionally, we have applied the proposed methods to competition "Adversarial Challenge on Object Detection" that is organized by Alibaba on the Tianchi platform and won top 7 in 1701 teams. Code is available at: https://github.com/FenHua/DetDak