Fast Local Attack: Generating Local Adversarial Examples for Object Detectors

TL;DR

This paper introduces a method for generating targeted local adversarial perturbations for anchor-free object detectors, improving attack efficiency and transferability compared to global perturbation methods.

Contribution

It presents a novel local attack approach leveraging semantic information, specifically designed for anchor-free detectors, with enhanced transferability to other detector types.

Findings

Achieves higher attack success rate with less computation

Perturbations transfer effectively to anchor-based detectors

Method outperforms global perturbation approaches in local attack scenarios

Abstract

The deep neural network is vulnerable to adversarial examples. Adding imperceptible adversarial perturbations to images is enough to make them fail. Most existing research focuses on attacking image classifiers or anchor-based object detectors, but they generate globally perturbation on the whole image, which is unnecessary. In our work, we leverage higher-level semantic information to generate high aggressive local perturbations for anchor-free object detectors. As a result, it is less computationally intensive and achieves a higher black-box attack as well as transferring attack performance. The adversarial examples generated by our method are not only capable of attacking anchor-free object detectors, but also able to be transferred to attack anchor-based object detector.