Transferable Adversarial Examples for Anchor Free Object Detection

TL;DR

This paper introduces the first adversarial attack method targeting anchor-free object detectors, using category-wise attacks and semantic information to generate transferable adversarial examples that can also attack other detector types.

Contribution

It presents a novel attack approach specifically designed for anchor-free detectors, enhancing transferability to other models including anchor-based detectors.

Findings

Achieves state-of-the-art attack performance on benchmark datasets.

Demonstrates high transferability of adversarial examples across different detector architectures.

Validates effectiveness against both anchor-free and anchor-based object detectors.

Abstract

Deep neural networks have been demonstrated to be vulnerable to adversarial attacks: subtle perturbation can completely change prediction result. The vulnerability has led to a surge of research in this direction, including adversarial attacks on object detection networks. However, previous studies are dedicated to attacking anchor-based object detectors. In this paper, we present the first adversarial attack on anchor-free object detectors. It conducts category-wise, instead of previously instance-wise, attacks on object detectors, and leverages high-level semantic information to efficiently generate transferable adversarial examples, which can also be transferred to attack other object detectors, even anchor-based detectors such as Faster R-CNN. Experimental results on two benchmark datasets demonstrate that our proposed method achieves state-of-the-art performance and transferability.