EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Adaptive Momentum Teacher Reconstruction for Unsupervised Rail Surface Anomaly Detection

Yixin Song, Zelong You, Ruifeng Ding, Runliang Tian, Dongyu Fan

Inventions · 17.11.2026 · 11 (6) · 126

DOI: 10.3390/inventions11060126 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Inventions, online erschienen am 17.11.2026. Verfasst von Yixin Song und 4 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Labeled rail surface defect samples are scarce, defect types are diverse, and pixel-level annotations are costly to obtain, which limits the practical applicability of supervised learning methods. To address identity mapping, unstable reconstruction under limited training data, and abrupt gradient changes caused by hard-threshold mining in reconstruction-based unsupervised anomaly detection, we propose Adaptive Momentum Teacher Reconstruction (AMTR). AMTR comprises three complementary modules: the Adaptive Perturbation Bottleneck (APB) dynamically scales perturbations according to feature magnitude and removes the attention-input residual connection in the first decoder block to suppress identity mapping; Feature Discrepancy-Guided Gradient Mining (FDGM) uses continuous position-wise weights to modulate backward gradients; and Momentum Teacher Consistency Learning (MTCL) employs an exponential moving average teacher to provide stable reconstruction targets. On the evaluated RSDDs-AUG split, AMTR achieves the highest mean result for each of the seven metrics among the compared methods. In three independent runs, it achieves mean image-level AUROC and pixel-level AUPRO values of 99.30±0.04% and 77.98±0.51%, respectively. Ablation studies provide evidence for the contributions of the proposed modules. The experimental evaluation is conducted on the RSDDs-AUG dataset, while generalization to other rail datasets and inspection conditions remains to be further investigated.

Zitieren

APA

Song, Y., You, Z., Ding, R., Tian, R., & Fan, D. (2026). Adaptive Momentum Teacher Reconstruction for Unsupervised Rail Surface Anomaly Detection. Inventions, 11(6), 126. https://doi.org/10.3390/inventions11060126

Vancouver

Song Y, You Z, Ding R, Tian R, Fan D. Adaptive Momentum Teacher Reconstruction for Unsupervised Rail Surface Anomaly Detection. Inventions. 2026;11(6):126. doi:10.3390/inventions11060126

BibTeX
@article{Song2026,
  author = {Song, Yixin and You, Zelong and Ding, Ruifeng and Tian, Runliang and Fan, Dongyu},
  title = {Adaptive Momentum Teacher Reconstruction for Unsupervised Rail Surface Anomaly Detection},
  journal = {Inventions},
  year = {2026},
  volume = {11},
  number = {6},
  pages = {126},
  doi = {10.3390/inventions11060126}
}
Kostenlos starten →