EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

YOLOv12-VSD: A Transfer-Learning-Assisted Real-Time Detection Algorithm for Vehicle Surface Defects

Haopu Liu iD, Dequn Zhao, Yu Li iD

Sensors · 06.05.2026 · 26 (9) · 2905

DOI: 10.3390/s26092905 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 06.05.2026. Verfasst von Haopu Liu und 2 weiteren Autor:innen. Beteiligte Einrichtung: Beijing Dublin International College, Beijing University of Technology, Beijing 100124, China. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 34 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: the National Key Research and Development Program of China, the Natural Science Foundation of China.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Vehicle surface defect detection faces three core challenges: classification–localization inconsistency for boundary-sensitive defects, insufficient multi-scale feature response across defect sizes, and cross-scenario generalization degradation caused by domain shift among production lines. This paper proposes YOLOv12-VSD, an improved detection algorithm addressing these issues through coordinated modifications at three levels. An IoU-aware classification loss aligns classification confidence with localization quality. A reparameterized convolution module at the P4 feature level (P4-RepC3) enriches intermediate-layer directional feature diversity without increasing inference cost. A multi-scale spatial pyramid pooling–fast structure at the P5 feature level (P5-SPPF) expands the effective receptive field for large-area defects. A three-stage transfer learning framework comprising source-domain pretraining, target-domain adaptation, and low-learning-rate refinement is further designed to reduce domain shift with limited annotations. Experiments show that YOLOv12-VSD achieves a mean Average Precision at IoU threshold 0.50 (mAP@50) of 0.715, the highest among six comparison models, with only 6.1M parameters and 17.1 giga floating-point operations per second (GFLOPs). After three-stage transfer, mAP@50 improves from 0.531 to 0.652, with training duration reduced by 64%.

Schlagwörter

Stichwörter der Autor:innen

IoU-aware classification lossYOLOv12deep learningobject detectionstructural reparameterizationtransfer learningvehicle surface defect

Autor:innen und Institutionen

  • Beijing Dublin International College, Beijing University of Technology, Beijing 100124, China.
  • School of Information Science and Technology, Beijing University of Technology, Beijing 100124, China.

ORCID-Profile: Haopu Liu, Yu Li

Förderung

  • the National Key Research and Development Program of China
  • the Natural Science Foundation of China

Zitieren

APA

Liu, H., Zhao, D., & Li, Y. (2026). YOLOv12-VSD: A Transfer-Learning-Assisted Real-Time Detection Algorithm for Vehicle Surface Defects. Sensors, 26(9), 2905. https://doi.org/10.3390/s26092905

Vancouver

Liu H, Zhao D, Li Y. YOLOv12-VSD: A Transfer-Learning-Assisted Real-Time Detection Algorithm for Vehicle Surface Defects. Sensors. 2026;26(9):2905. doi:10.3390/s26092905

BibTeX
@article{Liu2026,
  author = {Liu, Haopu and Zhao, Dequn and Li, Yu},
  title = {YOLOv12-VSD: A Transfer-Learning-Assisted Real-Time Detection Algorithm for Vehicle Surface Defects},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {9},
  pages = {2905},
  doi = {10.3390/s26092905}
}
Kostenlos starten →