EUVIMEDEUVIMED entdecken
Crossref

DSANet-KD: Dual Semantic Approximation Network via Knowledge Distillation for Rail Surface Defect Detection

Wujie Zhou, Jiankang Hong, Xiaoxiao Ran, Weiqing Yan, Qiuping Jiang

IEEE Transactions on Intelligent Transportation Systems · 2024-10 · 25 (10) · 13849-13862

DOI: 10.1109/tits.2024.3385744 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Intelligent Transportation Systems, 2024-10. Verfasst von Wujie Zhou und 4 weiteren Autor:innen. Die Arbeit wurde bisher 44-mal zitiert (Crossref) und nennt 78 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Zhejiang Provincial Natural Science Foundation of China, Natural Science Foundation of Ningbo Municipality.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Owing to the development of convolutional neural networks (CNNs), the detection of defects on rail surfaces has significantly improved. Although existing methods achieve good results, they incur huge computational and parameter costs associated with CNNs. The usual approach to this problem is to design lightweight models that meet the needs of real-world applications; however, the performance is often compromised. To address the aforementioned problems, we designed a dual semantic approximation network via knowledge distillation (DSANet-KD, a student model with knowledge distillation) for rail surface defect detection; it focuses on both foreground and background knowledge and obtains more accurate prediction results. This model comprises an adaptive 3D spatial integration module, feature-optimization decoding module, and dual semantic approximation knowledge-distillation framework. Specifically, we employed a thoroughly trained teacher defect detection network equipped with dual semantic approximation information as an experienced teacher to guide the training of a student defect detection network. Experimental results showed that the proposed DSANet-KD achieved better accuracy with a smaller number of parameters than the state-of-the-art methods. To demonstrate the generalizability of DSANet-KD, experiments were conducted on a publicly available RGBD-SOD dataset, whose source code is available at: https://github.com/hjklearn/DSANet-KD.

Förderung

  • National Natural Science Foundation of China
  • Zhejiang Provincial Natural Science Foundation of China
  • Natural Science Foundation of Ningbo Municipality

Zitieren

APA

Zhou, W., Hong, J., Ran, X., Yan, W., & Jiang, Q. (2024). DSANet-KD: Dual Semantic Approximation Network via Knowledge Distillation for Rail Surface Defect Detection. IEEE Transactions on Intelligent Transportation Systems, 25(10), 13849-13862. https://doi.org/10.1109/tits.2024.3385744

Vancouver

Zhou W, Hong J, Ran X, Yan W, Jiang Q. DSANet-KD: Dual Semantic Approximation Network via Knowledge Distillation for Rail Surface Defect Detection. IEEE Transactions on Intelligent Transportation Systems. 2024;25(10):13849-13862. doi:10.1109/tits.2024.3385744

BibTeX
@article{Zhou2024,
  author = {Zhou, Wujie and Hong, Jiankang and Ran, Xiaoxiao and Yan, Weiqing and Jiang, Qiuping},
  title = {DSANet-KD: Dual Semantic Approximation Network via Knowledge Distillation for Rail Surface Defect Detection},
  journal = {IEEE Transactions on Intelligent Transportation Systems},
  year = {2024},
  volume = {25},
  number = {10},
  pages = {13849--13862},
  doi = {10.1109/tits.2024.3385744}
}
Kostenlos starten →