SCRT ‐ DETR : A Lightweight Structural Degradation‐Aware Detector With Spatial‐Frequency Enhancement for Robust Pipeline Defect Detection
Concurrency and Computation: Practice and Experience · 2026-11 · 38 (21)
DOI: 10.1002/cpe.71014 ↗Kurz gefasst
Fachartikel in der Zeitschrift Concurrency and Computation: Practice and Experience, online erschienen am 23.10.2026. Verfasst von Wei Liu und 5 weiteren Autor:innen. Die Arbeit nennt 46 Literaturangaben. Gefördert von: National Science and Technology Major Project.
Abstract (Originalsprache)
Regular monitoring of pipeline defects is crucial for the safe operation of municipal environmental infrastructure. However, tiny structural degradation, topological discontinuities, low‐light interference, and the slender morphology of defects pose significant challenges for accurate detection. In complex pipeline environments, existing detectors generally suffer from insufficient structural‐frequency perception and morphology‐insensitive localization. To address these issues, this paper proposes SCRT‐DETR, a lightweight structural degradation‐aware end‐to‐end detector for pipeline defect detection. The model introduces a spatial‐frequency collaborative FastSFS‐Net backbone, a topology‐preserving SODNP feature pyramid, and a morphology‐sensitive localization loss, Focaler‐MPDIoU, designed for slender defects under low‐light conditions. Experimental results show that the proposed method achieves an mAP50 of 81.02%, outperforming the RT‐DETR baseline by 2.96 percentage points, while reducing the number of parameters by 30.2%. The model strikes a favorable balance between accuracy and efficiency, demonstrating strong potential for real‐world pipeline inspection applications.
Förderung
- National Science and Technology Major Project
Zitieren
APA
Liu, W., Tian, X., Huang, D., Chen, K., Liang, J., & Huang, K. (2026). SCRT ‐ DETR : A Lightweight Structural Degradation‐Aware Detector With Spatial‐Frequency Enhancement for Robust Pipeline Defect Detection. Concurrency and Computation: Practice and Experience, 38(21). https://doi.org/10.1002/cpe.71014
Vancouver
Liu W, Tian X, Huang D, Chen K, Liang J, Huang K. SCRT ‐ DETR : A Lightweight Structural Degradation‐Aware Detector With Spatial‐Frequency Enhancement for Robust Pipeline Defect Detection. Concurrency and Computation: Practice and Experience. 2026;38(21). doi:10.1002/cpe.71014
BibTeX
@article{Liu2026,
author = {Liu, Wei and Tian, Xueyong and Huang, Dengke and Chen, Kai and Liang, Jiyan and Huang, Kaituo},
title = {SCRT ‐ DETR : A Lightweight Structural Degradation‐Aware Detector With Spatial‐Frequency Enhancement for Robust Pipeline Defect Detection},
journal = {Concurrency and Computation: Practice and Experience},
year = {2026},
volume = {38},
number = {21},
doi = {10.1002/cpe.71014}
}
EUVIMED