Stealthy Vehicle Adversarial Camouflage Texture Generation Based on Neural Style Transfer

Adversarial attacks that mislead deep neural networks (DNNs) into making incorrect predictions can also be implemented in the physical world. However, most of the existing adversarial camouflage textures that attack object detection models only consider the effectiveness of the attack, ignoring the...

Full description

Saved in:
Bibliographic Details
Main Authors: Wei Cai, Xingyu Di, Xin Wang, Weijie Gao, Haoran Jia
Format: Article
Language:English
Published: MDPI AG 2024-10-01
Series:Entropy
Subjects:
Online Access:https://www.mdpi.com/1099-4300/26/11/903
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850145206551183360
author Wei Cai
Xingyu Di
Xin Wang
Weijie Gao
Haoran Jia
author_facet Wei Cai
Xingyu Di
Xin Wang
Weijie Gao
Haoran Jia
author_sort Wei Cai
collection DOAJ
description Adversarial attacks that mislead deep neural networks (DNNs) into making incorrect predictions can also be implemented in the physical world. However, most of the existing adversarial camouflage textures that attack object detection models only consider the effectiveness of the attack, ignoring the stealthiness of adversarial attacks, resulting in the generated adversarial camouflage textures appearing abrupt to human observers. To address this issue, we propose a style transfer module added to an adversarial texture generation framework. By calculating the style loss between the texture and the specified style image, the adversarial texture generated by the model is guided to have good stealthiness and is not easily detected by DNNs and human observers in specific scenes. Experiments have shown that in both the digital and physical worlds, the vehicle full coverage adversarial camouflage texture we create has good stealthiness and can effectively fool advanced DNN object detectors while evading human observers in specific scenes.
format Article
id doaj-art-a33bf8f092c84dfbab9585d593211280
institution OA Journals
issn 1099-4300
language English
publishDate 2024-10-01
publisher MDPI AG
record_format Article
series Entropy
spelling doaj-art-a33bf8f092c84dfbab9585d5932112802025-08-20T02:28:09ZengMDPI AGEntropy1099-43002024-10-01261190310.3390/e26110903Stealthy Vehicle Adversarial Camouflage Texture Generation Based on Neural Style TransferWei Cai0Xingyu Di1Xin Wang2Weijie Gao3Haoran Jia4The Third Faculty of Xi’an Research Institute of High Technology, Xi’an 710064, ChinaThe Third Faculty of Xi’an Research Institute of High Technology, Xi’an 710064, ChinaThe Third Faculty of Xi’an Research Institute of High Technology, Xi’an 710064, ChinaThe Third Faculty of Xi’an Research Institute of High Technology, Xi’an 710064, ChinaThe Third Faculty of Xi’an Research Institute of High Technology, Xi’an 710064, ChinaAdversarial attacks that mislead deep neural networks (DNNs) into making incorrect predictions can also be implemented in the physical world. However, most of the existing adversarial camouflage textures that attack object detection models only consider the effectiveness of the attack, ignoring the stealthiness of adversarial attacks, resulting in the generated adversarial camouflage textures appearing abrupt to human observers. To address this issue, we propose a style transfer module added to an adversarial texture generation framework. By calculating the style loss between the texture and the specified style image, the adversarial texture generated by the model is guided to have good stealthiness and is not easily detected by DNNs and human observers in specific scenes. Experiments have shown that in both the digital and physical worlds, the vehicle full coverage adversarial camouflage texture we create has good stealthiness and can effectively fool advanced DNN object detectors while evading human observers in specific scenes.https://www.mdpi.com/1099-4300/26/11/903physical attackneural style transferstealthy adversarial attackwhite-box attackobject detection
spellingShingle Wei Cai
Xingyu Di
Xin Wang
Weijie Gao
Haoran Jia
Stealthy Vehicle Adversarial Camouflage Texture Generation Based on Neural Style Transfer
Entropy
physical attack
neural style transfer
stealthy adversarial attack
white-box attack
object detection
title Stealthy Vehicle Adversarial Camouflage Texture Generation Based on Neural Style Transfer
title_full Stealthy Vehicle Adversarial Camouflage Texture Generation Based on Neural Style Transfer
title_fullStr Stealthy Vehicle Adversarial Camouflage Texture Generation Based on Neural Style Transfer
title_full_unstemmed Stealthy Vehicle Adversarial Camouflage Texture Generation Based on Neural Style Transfer
title_short Stealthy Vehicle Adversarial Camouflage Texture Generation Based on Neural Style Transfer
title_sort stealthy vehicle adversarial camouflage texture generation based on neural style transfer
topic physical attack
neural style transfer
stealthy adversarial attack
white-box attack
object detection
url https://www.mdpi.com/1099-4300/26/11/903
work_keys_str_mv AT weicai stealthyvehicleadversarialcamouflagetexturegenerationbasedonneuralstyletransfer
AT xingyudi stealthyvehicleadversarialcamouflagetexturegenerationbasedonneuralstyletransfer
AT xinwang stealthyvehicleadversarialcamouflagetexturegenerationbasedonneuralstyletransfer
AT weijiegao stealthyvehicleadversarialcamouflagetexturegenerationbasedonneuralstyletransfer
AT haoranjia stealthyvehicleadversarialcamouflagetexturegenerationbasedonneuralstyletransfer