EUVIMEDEUVIMED entdecken
Crossref

Closed-Box Adversarial Attack Method for Object Detection Under Multiview Conditions

Yun Zhang, Zhenhua Yu, Zheng Yin, Ou Ye, Xuya Cong, Houbing Herbert Song

IEEE Internet of Things Journal · 15.07.2025 · 12 (14) · 27886-27900

DOI: 10.1109/jiot.2025.3566950 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Internet of Things Journal, erschienen am 15.07.2025. Verfasst von Yun Zhang und 5 weiteren Autor:innen. Die Arbeit wurde bisher 5-mal zitiert (Crossref) und nennt 46 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Key Research and Development Program of Shaanxi Province, Youth Innovation Team of Shaanxi Universities.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep learning-based object detection has become an important application in industrial IoT. However, studies have shown that adversarial attacks may cause object detection to output incorrect detection results. Such vulnerabilities can threaten the robustness of object detection systems and lead to security problems. To address the issue of low attack effectiveness on target detection from different perspectives using the existing adversarial attack methods, this article proposes an adversarial attack method with multiview adaptive weight-balancing. First, a multiview channel is constructed for training, and the target features under different viewpoints are comprehensively considered to enhance the robustness of the attack method. Then, the model is optimized by combining the model shake drop and patch cut-out algorithms during the training process, so that the attack method no longer relies on a single model, thus enhancing its generalization ability. Finally, by dynamically adjusting the weights of each viewpoint, a weight-balancing strategy is constructed, which adaptively adjusts the preference of different perspectives during the training process to enhance the attack effect of the attack method in each viewpoint. To verify the performance of the method, experiments are conducted on multiple benchmarks, specifically the PKU-Reid dataset. Compared with the mainstream methods, the proposed method improves the attack success rate by 3.78% and 19.26% under glass-box and closed-box conditions, respectively, while reducing the mean average precision of the object detection model by 2.18% and 11.12%, respectively. The experimental results demonstrate that the proposed method effectively enhances attack performance on targets from different viewpoints and exhibits better viewpoint robustness.

Förderung

  • National Natural Science Foundation of China
  • Key Research and Development Program of Shaanxi Province
  • Youth Innovation Team of Shaanxi Universities
  • Key Scientific Research Program of Education Department of Shaanxi Province

Zitieren

APA

Zhang, Y., Yu, Z., Yin, Z., Ye, O., Cong, X., & Song, H. H. (2025). Closed-Box Adversarial Attack Method for Object Detection Under Multiview Conditions. IEEE Internet of Things Journal, 12(14), 27886-27900. https://doi.org/10.1109/jiot.2025.3566950

Vancouver

Zhang Y, Yu Z, Yin Z, Ye O, Cong X, Song HH. Closed-Box Adversarial Attack Method for Object Detection Under Multiview Conditions. IEEE Internet of Things Journal. 2025;12(14):27886-27900. doi:10.1109/jiot.2025.3566950

BibTeX
@article{Zhang2025,
  author = {Zhang, Yun and Yu, Zhenhua and Yin, Zheng and Ye, Ou and Cong, Xuya and Song, Houbing Herbert},
  title = {Closed-Box Adversarial Attack Method for Object Detection Under Multiview Conditions},
  journal = {IEEE Internet of Things Journal},
  year = {2025},
  volume = {12},
  number = {14},
  pages = {27886--27900},
  doi = {10.1109/jiot.2025.3566950}
}
Kostenlos starten →