EUVIMEDEUVIMED entdecken
Open Access · CC BY 3.0

Ontology-Aware Structure-Matched Learning for Explainable PCB Defect Localization and Discovery

Cheng Peng, Fangqiang Hu

International Journal on Semantic Web and Information Systems · 02.10.2026 · 22 (1) · 1-25

DOI: 10.4018/ijswis.423768 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal on Semantic Web and Information Systems, erschienen am 02.10.2026. Verfasst von Cheng Peng und 1 weiteren Autor:innen. Die Arbeit nennt 39 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 3.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Printed circuit board (PCB) defect inspection from mixed unlabeled data requires semantic modeling of structural context and manufacturing constraints. Existing supervised methods depend on extensive annotations, whereas conventional unsupervised detectors often assume clean normal data and offer limited interpretability. The authors propose Ontology-Aware Structure-Matched Defect Discovery (OA-SMDD) for explainable localization and category discovery. A lightweight PCB ontology organizes structural entities, relations, and constraints. OA-SMDD masks the query center, retrieves regions with similar surrounding layouts, and aggregates matched features into a position-specific reference. Deviations from the reference generate anomaly responses and initial bounding boxes, while difference features support confidence-guided clustering and one-step refinement. On the DsPCBSD+ and PCB-IND v4 datasets, OA-SMDD achieves AP50:95 scores of 44.26% and 38.52%, adjusted Rand index scores of 64.38% and 53.42%, and DA-mAP50 scores of 59.87% and 49.64%, respectively.

Zitieren

APA

Peng, C., & Hu, F. (2026). Ontology-Aware Structure-Matched Learning for Explainable PCB Defect Localization and Discovery. International Journal on Semantic Web and Information Systems, 22(1), 1-25. https://doi.org/10.4018/ijswis.423768

Vancouver

Peng C, Hu F. Ontology-Aware Structure-Matched Learning for Explainable PCB Defect Localization and Discovery. International Journal on Semantic Web and Information Systems. 2026;22(1):1-25. doi:10.4018/ijswis.423768

BibTeX
@article{Peng2026,
  author = {Peng, Cheng and Hu, Fangqiang},
  title = {Ontology-Aware Structure-Matched Learning for Explainable PCB Defect Localization and Discovery},
  journal = {International Journal on Semantic Web and Information Systems},
  year = {2026},
  volume = {22},
  number = {1},
  pages = {1--25},
  doi = {10.4018/ijswis.423768}
}
Kostenlos starten →