EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

DefectBoostDETR: Wheel hub defect detection algorithm based on feature enhancement and multi-scale fusion

Zaien Ying, Huimin Mao, Yang Li

AIP Advances · 01.05.2026 · 16 (5)

DOI: 10.1063/5.0333264 ↗

Kurz gefasst

Fachartikel in der Zeitschrift AIP Advances, online erschienen am 12.05.2026. Verfasst von Zaien Ying und 2 weiteren Autor:innen. Die Arbeit nennt 40 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In practical applications, a balance needs to be made based on energy grade requirements. In this paper, a high-precision wheel hub defect detection algorithm, DefectBoostDETR, is proposed to solve the problems of low efficiency and accuracy of traditional wheel hub surface defect detection methods. The algorithm amplifies low-contrast flaws within 10 ms via the STARNet backbone, reconciles scale variance on curved surfaces by the multi-scale fusion module, preserves micro-defect edges through Haar-wavelet down-sampling, and adapts to hub curvature using learnable positional encodings without non-maximum suppression post-processing. The experimental results show that compared with the baseline RT-DETR model, the DefectBoostDETR model shows a better performance in terms of mAP@0.5. At the same time, the number of model parameters and floating-point operations were reduced by 41.6% and 42.4%, respectively. In the comparative experiments of the other seven different types of models, DefectBoostDETR showed the best performance in terms of detection accuracy and speed, realizing the comprehensive improvement of hub defect detection accuracy, efficiency, and lightweight. The DefectBoostDETR model proposed in this study shows significant advantages in the wheel hub defect detection task, providing a new solution for industrial defect detection.

Zitieren

APA

Ying, Z., Mao, H., & Li, Y. (2026). DefectBoostDETR: Wheel hub defect detection algorithm based on feature enhancement and multi-scale fusion. AIP Advances, 16(5). https://doi.org/10.1063/5.0333264

Vancouver

Ying Z, Mao H, Li Y. DefectBoostDETR: Wheel hub defect detection algorithm based on feature enhancement and multi-scale fusion. AIP Advances. 2026;16(5). doi:10.1063/5.0333264

BibTeX
@article{Ying2026,
  author = {Ying, Zaien and Mao, Huimin and Li, Yang},
  title = {DefectBoostDETR: Wheel hub defect detection algorithm based on feature enhancement and multi-scale fusion},
  journal = {AIP Advances},
  year = {2026},
  volume = {16},
  number = {5},
  doi = {10.1063/5.0333264}
}
Kostenlos starten →