EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Adapting Vision–Language Models for Few-Shot Industrial Defect Detection

Chayanon Sub-r-pa, Rung-Ching Chen

Algorithms · 27.03.2026 · 19 (4) · 259

DOI: 10.3390/a19040259 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Algorithms, online erschienen am 27.03.2026. Verfasst von Chayanon Sub-r-pa und 1 weiteren Autor:innen. Die Arbeit wurde bisher 8-mal zitiert (Crossref) und nennt 33 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Science and Technology Council.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Automated surface defect detection often faces a “cold-start” problem due to limited annotated data for new anomalies. Traditional object detectors struggle to converge in such few-shot settings. To address this, we adapt Vision–Language Models (VLMs), specifically YOLO-World. We use semantic pre-training to mitigate data scarcity. We evaluate this approach on the MVTec AD dataset in bounding-box format. We use a strict 1:9 train-validation split, resulting in an average of 11.8 defect instances per category. YOLO-World surpasses traditional baselines, like YOLOv11s and YOLOv26s, in 12 of 15 categories. The optimized VLM pipeline achieves up to 64.9% mAP@50 on texture-heavy categories, such as Tile, with only nine training instances. Ablation studies show standard optimization techniques are limited under 10-shot constraints. We find a critical augmentation divide. Disabling spatial distortions (Mosaic) is vital to preserving rigid-object geometry. The Normalized Wasserstein Distance (NWD) improves the localization of microscopic anomalies. Varifocal Loss (VFL) often causes model collapse. Ultimately, VLMs offer a superior foundation for cold-start inspection but require carefully tailored pipelines for robustness.

Förderung

  • National Science and Technology Council

Zitieren

APA

Sub-r-pa, C., & Chen, R. (2026). Adapting Vision–Language Models for Few-Shot Industrial Defect Detection. Algorithms, 19(4), 259. https://doi.org/10.3390/a19040259

Vancouver

Sub-r-pa C, Chen R. Adapting Vision–Language Models for Few-Shot Industrial Defect Detection. Algorithms. 2026;19(4):259. doi:10.3390/a19040259

BibTeX
@article{Subrpa2026,
  author = {Sub-r-pa, Chayanon and Chen, Rung-Ching},
  title = {Adapting Vision–Language Models for Few-Shot Industrial Defect Detection},
  journal = {Algorithms},
  year = {2026},
  volume = {19},
  number = {4},
  pages = {259},
  doi = {10.3390/a19040259}
}
Kostenlos starten →