EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

FF-DEIM: DEIM with Image Dehazing and Self-Supervised Pretraining for Catenary Support Component Detection

Lingzhi Zhang, Jinyong Huang, Guojin Qin, Jincheng Cao, Fei Fan iD, Hui Wang, Haonan Yang iD

Sensors · 06.08.2026 · 26 (15) · 5000

DOI: 10.3390/s26155000 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 06.08.2026. Verfasst von Lingzhi Zhang und 6 weiteren Autor:innen. Beteiligte Einrichtung: Hunan Provincial Engineering Technology Research Center for High-Speed Railway Operation Safety Assurance, Hunan Railway Professional Technology Coll…. Die Arbeit nennt 68 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, China Postdoctoral Science Foundation, Central Universities.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The catenary support component (CSC) is a key part of the electrified railway system, and its operational status directly affects railway operational safety. These components’ images are collected using inspection equipment and detected using computer vision techniques. However, catenary network inspection faces the following issues: (1) due to limitations in the equipment’s shooting angle and changes in viewing distance, the collected images contain multi-scale and multi-class problems, and (2) the railway environment is highly variable, and adverse weather conditions such as fog, rain, and low light affect the imaging devices, leading to degraded image quality. To address these issues, this paper proposes a novel detection framework, FF-DEIM, for detecting catenary support components. First, a dual-channel fusion network (DCFNet) is introduced, which significantly improves image quality by removing foreground interferences such as fog, raindrops, and dynamic blur. Second, a pretraining framework based on contrastive learning, mask image modeling with contrastive learning (MIMCL), is designed to enhance the model’s focus on key regions of the catenary network components, optimizing feature extraction capabilities and improving model convergence speed. Then, a feature-focusing pyramid network (FFPN) is proposed, which uses the focus feature module to fuse cross-level contextual features, enhancing the ability to capture local details and improving the model’s small object detection performance. Finally, a drone-based catenary network image dataset, including scenes with fog, rain, and low light, is constructed, and experiments validate the effectiveness of the proposed method.

Schlagwörter

Stichwörter der Autor:innen

catenary support components (CSCs)deep learningobject detectionself-supervised learning

Autor:innen und Institutionen

  • Hunan Provincial Engineering Technology Research Center for High-Speed Railway Operation Safety Assurance, Hunan Railway Professional Technology College, Zhuzhou 412001, China.
  • China Railway No. 5 Engineering Group Co., Ltd., Changsha 410017, China.
  • School of Electrical Engineering, Southwest Jiaotong University, Chengdu 610097, China.

ORCID-Profile: Fei Fan, Haonan Yang

Förderung

  • National Natural Science Foundation of China
  • China Postdoctoral Science Foundation
  • Central Universities

Zitieren

APA

Zhang, L., Huang, J., Qin, G., Cao, J., Fan, F., Wang, H., & Yang, H. (2026). FF-DEIM: DEIM with Image Dehazing and Self-Supervised Pretraining for Catenary Support Component Detection. Sensors, 26(15), 5000. https://doi.org/10.3390/s26155000

Vancouver

Zhang L, Huang J, Qin G, Cao J, Fan F, Wang H, et al. FF-DEIM: DEIM with Image Dehazing and Self-Supervised Pretraining for Catenary Support Component Detection. Sensors. 2026;26(15):5000. doi:10.3390/s26155000

BibTeX
@article{Zhang2026,
  author = {Zhang, Lingzhi and Huang, Jinyong and Qin, Guojin and Cao, Jincheng and Fan, Fei and Wang, Hui and Yang, Haonan},
  title = {FF-DEIM: DEIM with Image Dehazing and Self-Supervised Pretraining for Catenary Support Component Detection},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {15},
  pages = {5000},
  doi = {10.3390/s26155000}
}
Kostenlos starten →