pith. machine review for the scientific record. sign in

arxiv: 2602.07044 · v2 · submitted 2026-02-04 · 💻 cs.CV · cs.AI

Recognition: no theorem link

PipeMFL-240K: A Large-scale Dataset and Benchmark for Object Detection in Pipeline Magnetic Flux Leakage Imaging

Tianyi Qu , Songxiao Yang , Haolin Wang , Huadong Song , Xiaoting Guo , Wenguang Hu , Guanlin Liu , Honghe Chen
show 1 more author
Yafei Ou
Authors on Pith no claims yet

Pith reviewed 2026-05-16 08:12 UTC · model grok-4.3

classification 💻 cs.CV cs.AI
keywords PipeMFL-240Kmagnetic flux leakageobject detectionpipeline inspectiondefect detectionlarge-scale datasetbenchmarkdeep learning
0
0 comments X

The pith

PipeMFL-240K provides the first large-scale public dataset and benchmark for detecting defects in pipeline magnetic flux leakage images.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces a new dataset called PipeMFL-240K to support the development of deep learning models that can automatically find defects in magnetic flux leakage images of pipelines. It contains over 249,000 images with 200,000 bounding box annotations collected from 12 real pipelines covering 1,530 kilometers. Current object detection methods still struggle with the dataset's challenges, such as very small defects that take up only a few pixels, a long-tailed distribution across 12 categories, and high variation within classes. By making this data public with established baselines, the work aims to enable fair comparisons and speed up progress in using AI for pipeline safety and maintenance.

Core claim

The central claim is that PipeMFL-240K, comprising 249,320 images and 200,020 high-quality bounding-box annotations across 12 categories from 12 pipelines spanning 1,530 km, serves as the first public large-scale benchmark for object detection in pipeline MFL pseudo-color images. This benchmark reveals that state-of-the-art detectors underperform due to the data's extreme long-tailed distribution, prevalence of tiny objects, and substantial intra-class variability, thereby establishing a challenging testbed to drive future algorithmic improvements in MFL-based pipeline integrity assessment.

What carries the argument

The PipeMFL-240K dataset, a collection of real-world MFL pseudo-color images with bounding box annotations for 12 defect categories that poses specific detection challenges like tiny objects and class imbalance.

Load-bearing premise

The 249,320 images and their annotations from the 12 pipelines capture the typical complexity and variability found in actual field inspections of pipelines.

What would settle it

Re-labeling a random subset of the images by independent experts and finding significant disagreement with the provided bounding boxes, or observing that detectors trained on the dataset perform no better than random on new MFL data from additional pipelines.

Figures

Figures reproduced from arXiv: 2602.07044 by Guanlin Liu, Haolin Wang, Honghe Chen, Huadong Song, Songxiao Yang, Tianyi Qu, Wenguang Hu, Xiaoting Guo, Yafei Ou.

Figure 1
Figure 1. Figure 1: An overview of MFL detection, the key challenges it poses and the limitations of prior work. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Feature taxonomy and annotation characteristics of the PipeMFL-240K dataset. The figure illustrates the pipeline [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: (A) Overall object counts for each annotated [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative benchmark results on representative MFL samples. Predicted bounding boxes from different detectors are [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Dataset scale study results on YOLOv8-m, YOLO26-m and RF-DETR-Base, illustrating performance variations in [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Overview of data collection and acquisition [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Overview of data selection and filtering. [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Pattern visualization of damage-type categories in MFL imaging cases: MTL, CRC, GWA and SWA with MLN scene [PITH_FULL_IMAGE:figures/full_fig_p015_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Pattern visualization of component-type categories in MFL imaging cases: BRN, CAS, TEE, ESP, BND, SLE, VAL and [PITH_FULL_IMAGE:figures/full_fig_p016_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Corrosion density as a function of service age for [PITH_FULL_IMAGE:figures/full_fig_p017_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Qualitative benchmark results on representative damage samples (Part A). [PITH_FULL_IMAGE:figures/full_fig_p020_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Qualitative benchmark results on representative damage samples (Part B). [PITH_FULL_IMAGE:figures/full_fig_p021_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Qualitative benchmark results on representative damage samples (Part C). [PITH_FULL_IMAGE:figures/full_fig_p022_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Qualitative benchmark results on representative component samples (Part A). [PITH_FULL_IMAGE:figures/full_fig_p023_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Qualitative benchmark results on representative component samples (Part B). [PITH_FULL_IMAGE:figures/full_fig_p024_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Qualitative benchmark results on representative component samples (Part C). [PITH_FULL_IMAGE:figures/full_fig_p025_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: Qualitative benchmark results on representative tiny damage samples (Part A). [PITH_FULL_IMAGE:figures/full_fig_p026_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: Qualitative benchmark results on representative tiny damage samples (Part B). [PITH_FULL_IMAGE:figures/full_fig_p027_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: Qualitative benchmark results on representative tiny damage samples (Part C). [PITH_FULL_IMAGE:figures/full_fig_p028_19.png] view at source ↗
read the original abstract

Pipeline integrity is critical to industrial safety and environmental protection, with Magnetic Flux Leakage (MFL) detection being a primary non-destructive testing technology. Despite the promise of deep learning for automating MFL interpretation, progress toward reliable models has been constrained by the absence of a large-scale public dataset and benchmark, making fair comparison and reproducible evaluation difficult. We introduce \textbf{PipeMFL-240K}, a large-scale, meticulously annotated dataset and benchmark for complex object detection in pipeline MFL pseudo-color images. PipeMFL-240K reflects real-world inspection complexity and poses several unique challenges: (i) an extremely long-tailed distribution over \textbf{12} categories, (ii) a high prevalence of tiny objects that often comprise only a handful of pixels and (iii) substantial intra-class variability. The dataset contains \textbf{249,320} images and \textbf{200,020} high-quality bounding-box annotations, collected from 12 pipelines spanning approximately \textbf{1,530} km. Extensive experiments are conducted with state-of-the-art object detectors to establish baselines. Results show that modern detectors still struggle with the intrinsic properties of MFL data, highlighting considerable headroom for improvement, while PipeMFL-240K provides a reliable and challenging testbed to drive future research. As the first public dataset and the first benchmark of this scale and scope for pipeline MFL inspection, it provides a critical foundation for efficient pipeline diagnostics as well as maintenance planning and is expected to accelerate algorithmic innovation and reproducible research in MFL-based pipeline integrity assessment.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript introduces PipeMFL-240K, a dataset of 249,320 MFL pseudo-color images containing 200,020 bounding-box annotations across 12 object categories, collected from 12 real pipelines totaling approximately 1,530 km. It identifies domain-specific challenges including extreme long-tailed class imbalance, prevalence of tiny objects spanning only a few pixels, and high intra-class variability. Baseline experiments with state-of-the-art object detectors are reported, showing that current models continue to struggle on these data, and the work positions the release as the first public large-scale benchmark for MFL-based pipeline inspection to support reproducible research and algorithmic progress.

Significance. If the annotations are verifiably high-quality and the 12 pipelines are representative of operational MFL inspection conditions, the dataset would constitute a substantial contribution by removing the primary barrier to reproducible deep-learning research in this industrial NDT domain. It would enable systematic comparison of detectors on realistic long-tailed and small-object regimes and could directly support improved automated diagnostics for pipeline safety.

major comments (2)
  1. [Abstract and dataset construction] Abstract and dataset description: The central claim that PipeMFL-240K constitutes a 'reliable and challenging testbed' rests on the assertion of 'meticulously annotated' and 'high-quality' bounding boxes. No annotation protocol, number of annotators, inter-annotator agreement statistics, expert review process, or quantitative validation metrics are supplied. For a dataset whose stated difficulties are tiny objects and long-tailed classes, even moderate label noise would invalidate the baseline comparisons and the 'headroom for improvement' conclusion.
  2. [Experiments and baselines] Experimental section: The manuscript states that 'extensive experiments' were conducted with state-of-the-art detectors yet supplies neither the precise data splits (train/val/test ratios or pipeline-level partitioning), the full list of evaluated models, nor the primary quantitative metrics (mAP, AP per class, or small-object AP) that would allow readers to reproduce or extend the baselines. This information is load-bearing for the claim that modern detectors 'still struggle'.
minor comments (1)
  1. [Abstract] The abstract claims the dataset 'reflects real-world inspection complexity' without providing any quantitative comparison (e.g., object-size histograms or class-frequency tables) against prior smaller MFL datasets to substantiate representativeness.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive review. We agree that the manuscript requires additional details on annotation quality and experimental reproducibility to fully support its claims. We address each major comment below and will incorporate the requested information in the revised version.

read point-by-point responses

  1. Referee: [Abstract and dataset construction] Abstract and dataset description: The central claim that PipeMFL-240K constitutes a 'reliable and challenging testbed' rests on the assertion of 'meticulously annotated' and 'high-quality' bounding boxes. No annotation protocol, number of annotators, inter-annotator agreement statistics, expert review process, or quantitative validation metrics are supplied. For a dataset whose stated difficulties are tiny objects and long-tailed classes, even moderate label noise would invalidate the baseline comparisons and the 'headroom for improvement' conclusion.

    Authors: We agree that explicit documentation of the annotation process is essential to substantiate dataset quality, especially for tiny objects and long-tailed classes where label noise could affect conclusions. In the revised manuscript we will add a dedicated subsection detailing the annotation protocol, the number of annotators, inter-annotator agreement statistics (including IoU thresholds and agreement coefficients), the multi-stage expert review process by pipeline inspection specialists, and quantitative validation metrics such as precision-recall on a held-out expert-verified subset. These additions will directly address the concern and reinforce the reliability of the benchmark. revision: yes

  2. Referee: [Experiments and baselines] Experimental section: The manuscript states that 'extensive experiments' were conducted with state-of-the-art detectors yet supplies neither the precise data splits (train/val/test ratios or pipeline-level partitioning), the full list of evaluated models, nor the primary quantitative metrics (mAP, AP per class, or small-object AP) that would allow readers to reproduce or extend the baselines. This information is load-bearing for the claim that modern detectors 'still struggle'.

    Authors: We acknowledge that the experimental section currently omits the precise implementation details needed for reproducibility. In the revised manuscript we will specify the exact train/validation/test ratios, describe the pipeline-level partitioning strategy used to avoid cross-pipeline leakage, list all evaluated detectors with their configurations, and report the full set of primary metrics including overall mAP, per-class AP, and small-object AP (AP_S). These changes will allow readers to reproduce the baselines and will strengthen the evidence that current detectors continue to struggle on the dataset's intrinsic challenges. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical dataset release with no derivations or self-referential predictions

full rationale

The paper is a dataset introduction and empirical benchmark study. It reports collection of 249,320 images and 200,020 annotations from 12 real pipelines, then runs standard off-the-shelf detectors to produce baseline numbers. No equations, fitted parameters, predictions, or uniqueness theorems appear anywhere in the text. Claims of being the first large-scale public MFL dataset rest on the factual assertion of prior non-existence rather than any self-referential derivation. No self-citations are used to justify core methodology. The work is therefore self-contained against external benchmarks and receives the default non-circularity finding.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is an empirical dataset and benchmark paper with no mathematical derivations, so it introduces no free parameters, axioms, or invented entities.

pith-pipeline@v0.9.0 · 5619 in / 1201 out tokens · 43018 ms · 2026-05-16T08:12:42.743443+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

48 extracted references · 48 canonical work pages

  1. [1]

    Fatih Cagatay Akyon, Sinan Onur Altinuc, and Alptekin Temizel. 2022. Slicing Aided Hyper Inference and Fine-tuning for Small Object Detection.2022 IEEE International Conference on Image Processing (ICIP)(2022), 966–970. doi:10.1109/ ICIP46576.2022.9897990

    work page arXiv 2022

  2. [2]

    Bruce Brown and Jeremy Moloney. 2017. Under-deposit corrosion.Trends in oil and gas corrosion research and technologies(2017), 363–383

    work page 2017

  3. [3]

    Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexan- der Kirillov, and Sergey Zagoruyko. 2020. End-to-End Object Detection with Transformers. InComputer Vision – ECCV 2020, Andrea Vedaldi, Horst Bischof, Thomas Brox, and Jan-Michael Frahm (Eds.). Springer International Publishing, Cham, 213–229

    work page 2020

  4. [4]

    Tianheng Cheng, Lin Song, Yixiao Ge, Wenyu Liu, Xinggang Wang, and Ying Shan. 2024. YOLO-World: Real-Time Open-Vocabulary Object Detection. In 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 16901–16911. doi:10.1109/CVPR52733.2024.01599 9 Arxiv Preprint, February 04, 2026, 2602.07044 Tianyi Qu et al

    work page doi:10.1109/cvpr52733.2024.01599 2024

  5. [5]

    Mark Everingham, Luc Van Gool, Christopher Williams, John Winn, and Andrew Zisserman. 2010. The Pascal Visual Object Classes (VOC) challenge.International Journal of Computer Vision88 (06 2010), 303–338. doi:10.1007/s11263-009-0275-4

    work page doi:10.1007/s11263-009-0275-4 2010

  6. [6]

    Bo Feng, Jianbo Wu, Hongming Tu, Jian Tang, and Yihua Kang. 2022. A Review of Magnetic Flux Leakage Nondestructive Testing.Materials15, 20 (2022). doi:10. 3390/ma15207362

    work page 2022

  7. [7]

    Fucheng Han and Xianming Lang. 2023. A Fast Magnetic Flux Leakage Small Defect Detection Network.IEEE Transactions on Industrial Informatics19, 12 (2023), 11941–11948. doi:10.1109/TII.2023.3280950

    work page doi:10.1109/tii.2023.3280950 2023

  8. [8]

    Fucheng Han, Xianming Lang, and Mingyang Liu. 2023. An Anchor-Free Pipeline MFL Image Detection Method.IEEE Transactions on Instrumentation and Mea- surement72 (2023), 1–8. doi:10.1109/TIM.2023.3304688

    work page doi:10.1109/tim.2023.3304688 2023

  9. [9]

    Songling Huang, Lisha Peng, Hongyu Sun, and Shisong Li. 2023. Deep Learning for Magnetic Flux Leakage Detection and Evaluation of Oil & Gas Pipelines: A Review.Energies16, 3 (2023). doi:10.3390/en16031372

    work page doi:10.3390/en16031372 2023

  10. [10]

    Hwang, S

    K. Hwang, S. Mandayam, S.S. Udpa, L. Udpa, W. Lord, and M. Atzal. 2000. Charac- terization of gas pipeline inspection signals using wavelet basis function neural networks.NDT & E International33, 8 (2000), 531–545. doi:10.1016/S0963- 8695(00)00008-6

    work page doi:10.1016/s0963- 2000

  11. [11]

    Lin Jiang, Huaguang Zhang, Jinhai Liu, Xiangkai Shen, and Hang Xu. 2025. Mutual Supervision of MFL Heterogeneous Signals for Insufficient Sample Defect Detection on Pipeline Safety Operation.IEEE Transactions on Automation Science and Engineering22 (2025), 1714–1724. doi:10.1109/TASE.2024.3369659

    work page doi:10.1109/tase.2024.3369659 2025

  12. [12]

    2020.Ultralytics YOLOv5

    Glenn Jocher. 2020.Ultralytics YOLOv5. doi:10.5281/zenodo.3908559

    work page doi:10.5281/zenodo.3908559 2020

  13. [13]

    2023.Ultralytics YOLOv8

    Glenn Jocher, Ayush Chaurasia, and Jing Qiu. 2023.Ultralytics YOLOv8. https: //github.com/ultralytics/ultralytics

    work page 2023

  14. [14]

    2024.Ultralytics YOLO11

    Glenn Jocher and Jing Qiu. 2024.Ultralytics YOLO11. https://github.com/ ultralytics/ultralytics

    work page 2024

  15. [15]

    2026.Ultralytics YOLO26

    Glenn Jocher and Jing Qiu. 2026.Ultralytics YOLO26. https://github.com/ ultralytics/ultralytics

    work page 2026

  16. [16]

    Lecun, L

    Y. Lecun, L. Bottou, Y. Bengio, and P. Haffner. 1998. Gradient-based learning applied to document recognition.Proc. IEEE86, 11 (1998), 2278–2324. doi:10. 1109/5.726791

    work page 1998

  17. [17]

    Chi-Ming Lee, S Bond, and P Woollin. 2005. Preferential weld corrosion: Effects of weldment microstructure and composition. InCORROSION 2005. Association for Materials Protection and Performance, 1–16

    work page 2005

  18. [18]

    Chunhui Li, Yan Lyu, Xianglai Meng, Linghui Li, and Xiaochen Pu. 2025. Research on Intelligent Defect Recognition in Oil and Gas Pipeline Magnetic Flux Leakage Detection Based on YOLOv8.Chinese Journal of Scientific Instrument46, 2 (2025), 247–254

    work page 2025

  19. [19]

    Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Dollár. 2017. Focal Loss for Dense Object Detection. In2017 IEEE International Conference on Computer Vision (ICCV). 2999–3007. doi:10.1109/ICCV.2017.324

    work page doi:10.1109/iccv.2017.324 2017

  20. [20]

    Lawrence Zitnick

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C. Lawrence Zitnick. 2014. Microsoft COCO: Common Objects in Context. InComputer Vision – ECCV 2014, David Fleet, Tomas Pajdla, Bernt Schiele, and Tinne Tuytelaars (Eds.). Springer International Publishing, Cham, 740–755

    work page 2014

  21. [21]

    Jinhai Liu, Mingrui Fu, Feilong Liu, Jian Feng, and Kuangqing Cui. 2018. Window Feature-Based Two-Stage Defect Identification Using Magnetic Flux Leakage Measurements.IEEE Transactions on Instrumentation and Measurement67, 1 (2018), 12–23. doi:10.1109/TIM.2017.2755918

    work page doi:10.1109/tim.2017.2755918 2018

  22. [22]

    Jinhai Liu, Zhitao Wen, Xiangkai Shen, Fengyuan Zuo, Lin Jiang, and Huaguang Zhang. 2025. Online Pipeline Weld Defect Detection for Magnetic Flux Leak- age Inspection System via Lightweight Rotated Network.IEEE Transactions on Industrial Electronics72, 7 (2025), 7573–7584. doi:10.1109/TIE.2024.3503635

    work page doi:10.1109/tie.2024.3503635 2025

  23. [23]

    Shilong Liu, Zhaoyang Zeng, Tianhe Ren, Feng Li, Hao Zhang, Jie Yang, Qing Jiang, Chunyuan Li, Jianwei Yang, Hang Su, Jun Zhu, and Lei Zhang. 2025. Grounding DINO: Marrying DINO with Grounded Pre-training for Open-Set Object Detection. InComputer Vision – ECCV 2024, Aleš Leonardis, Elisa Ricci, Stefan Roth, Olga Russakovsky, Torsten Sattler, and Gül Varol...

    work page 2025

  24. [24]

    Reed, Cheng-Yang Fu, and Alexander C

    Wei Liu, Dragomir Anguelov, Dumitru Erhan, Christian Szegedy, Scott E. Reed, Cheng-Yang Fu, and Alexander C. Berg. 2016. SSD: Single Shot MultiBox Detector. InComputer Vision - ECCV 2016 - 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part I (Lecture Notes in Computer Science, Vol. 9905), Bastian Leibe, Jiri Matas...

    work page doi:10.1007/978-3-319-46448-0_2 2016

  25. [25]

    Qingmiao Ma, Weige Liang, and Peiyi Zhou. 2025. A Review on Pipeline In-Line Inspection Technologies.Sensors25, 15 (2025). doi:10.3390/s25154873

    work page doi:10.3390/s25154873 2025

  26. [26]

    Mariam Manzoor, Zahra Arabi Narei, Henry Leung, and Scott Miller. 2024. Multi- Modal Pipeline Defect Localization. InMuslims in ML Workshop co-located with NeurIPS 2024. https://openreview.net/forum?id=Mkkn1R6cCl

    work page 2024

  27. [27]

    Neubeck and L

    A. Neubeck and L. Van Gool. 2006. Efficient Non-Maximum Suppression. In 18th International Conference on Pattern Recognition (ICPR’06), Vol. 3. 850–855. doi:10.1109/ICPR.2006.479

    work page doi:10.1109/icpr.2006.479 2006

  28. [28]

    LD Paolinelli. 2018. Study of Phase Wetting in Three-Phase Oil-Water-Gas Horizontal Pipe Flow–Recommendations for Corrosion Risk Assessment. In NACE CORROSION. NACE, NACE–2018

    work page 2018

  29. [29]

    Xiang Peng, Uchenna Anyaoha, Zheng Liu, and Kazuhiko Tsukada. 2020. Analysis of Magnetic-Flux Leakage (MFL) Data for Pipeline Corrosion Assessment.IEEE Transactions on Magnetics56, 6 (2020), 1–15. doi:10.1109/TMAG.2020.2981450

    work page doi:10.1109/tmag.2020.2981450 2020

  30. [30]

    Joseph Redmon, Santosh Divvala, Ross Girshick, and Ali Farhadi. 2016. You Only Look Once: Unified, Real-Time Object Detection. In2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 779–788. doi:10.1109/CVPR. 2016.91

    work page doi:10.1109/cvpr 2016

  31. [31]

    Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun. 2017. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks.IEEE Transactions on Pattern Analysis and Machine Intelligence39, 6 (2017), 1137–1149. doi:10.1109/TPAMI.2016.2577031

    work page doi:10.1109/tpami.2016.2577031 2017

  32. [32]

    Isaac Robinson, Peter Robicheaux, Matvei Popov, Deva Ramanan, and Neehar Peri. 2026. RF-DETR: Neural Architecture Search for Real-Time Detection Trans- formers. InThe Fourteenth International Conference on Learning Representations (ICLR). https://openreview.net/forum?id=qHm5GePxTh

    work page 2026

  33. [33]

    Ruiz, L E Mujica, M

    M. Ruiz, L E Mujica, M. Quintero, J. Florez, and S. Quintero. 2015. Magnetic Flux Leakage and Principal Component Analysis for metal loss approximation in a pipeline.Journal of Physics Conference Series628, 1 (2015)

    work page 2015

  34. [34]

    Xiangkai Shen, Jinhai Liu, Lin Jiang, Xiaoyuan Liu, and Huaguang Zhang. 2024. A Novel Weld Defect Detection Method for Intelligent Magnetic Flux Leakage De- tection System via Contextual Relation Network.IEEE Transactions on Industrial Electronics71, 6 (2024), 6304–6314. doi:10.1109/TIE.2023.3294578

    work page doi:10.1109/tie.2023.3294578 2024

  35. [35]

    Yan Shi, Chao Zhang, Rui Li, Maolin Cai, and Guanwei Jia. 2015. Theory and application of magnetic flux leakage pipeline detection.Sensors15, 12 (2015), 31036–31055

    work page 2015

  36. [36]

    Huadong Song, Qi Xiao, Gang Wang, Jianpeng Zhang, and Wenguang Hu. 2023. A Composite Approach of Electromagnetic Acoustic Transducer and Eddy Cur- rent for Inner and Outer Corrosion Defects Detection.IEEE Transactions on Instrumentation and Measurement72 (2023), 1–11. doi:10.1109/TIM.2023.3238689

    work page doi:10.1109/tim.2023.3238689 2023

  37. [37]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Ł ukasz Kaiser, and Illia Polosukhin. 2017. Attention is All you Need. InAdvances in Neural Information Processing Systems, I. Guyon, U. Von Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett (Eds.), Vol. 30. Curran Associates, Inc. https://proc...

    work page 2017

  38. [38]

    Jinzhu Wang and Zegen Wang. 2010. Defect Localization in Long-distance Oli and Gas Pipelines: A Survey (Chinese).Journal of Oil and Gas Technology32, 1 (2010), 372–374

    work page 2010

  39. [39]

    Qiannan Wang, Jianhua Tang, Jinhai Liu, Huaguang Zhang, Yifu Ren, Zhitao Wen, and Xiangkai Shen. 2025. A Novel Geometry-Aware Context Selection Network for Small Defect Detection Under Magnetic Flux Leakage Signals.IEEE Transactions on Instrumentation and Measurement74 (2025), 1–12. doi:10.1109/ TIM.2025.3571089

    work page arXiv 2025

  40. [40]

    Jianbo Wu, Hui Fang, Xiaoming Huang, Hui Xia, Yihua Kang, and Chaoqing Tang. 2017. An Online MFL Sensing Method for Steel Pipe Based on the Magnetic Guiding Effect.Sensors17, 12 (2017). doi:10.3390/s17122911

    work page doi:10.3390/s17122911 2017

  41. [41]

    Lang Xianming, Zheng Hao, Song Huadong, Liu Jinhai, Guo Xiaoting, Meng Qiang, and Yuan Haitao. 2021. Weld Defect Recognition Method of Pipeline Based on Improved Least Squares Twin Support Vector Machine. In2021 29th Mediterranean Conference on Control and Automation (MED). 500–505. doi:10. 1109/MED51440.2021.9480203

    work page arXiv 2021

  42. [42]

    Zhenchang Xu, Kuirong Liu, and Kewei Gu, Xiaolu Gao. 2023. Image recogni- tion model of pipeline magnetic flux leakage detection based on deep learning. Corrosion Reviews41, 6 (2023), 689–701

    work page 2023

  43. [43]

    Lijian Yang, Zhujun Wang, and Songwei Gao. 2020. Pipeline Magnetic Flux Leakage Image Detection Algorithm Based on Multiscale SSD Network.IEEE Transactions on Industrial Informatics16, 1 (2020), 501–509. doi:10.1109/TII.2019. 2926283

    work page doi:10.1109/tii.2019 2020

  44. [44]

    Veysel Yuksel, Yusuf Engin Tetik, Mahmut Omer Basturk, Onur Recepoglu, Kur- sad Gokce, and Mehmet Ali Cimen. 2023. A Novel Cascaded Deep Learning Model for the Detection and Quantification of Defects in Pipelines via Magnetic Flux Leakage Signals.IEEE Transactions on Instrumentation and Measurement72 (2023), 1–9. doi:10.1109/TIM.2023.3272377

    work page doi:10.1109/tim.2023.3272377 2023

  45. [45]

    Luying Zhang, Yuchen Bian, Peng Jiang, Yang Huang, and Ying Liu. 2024. Improv- ing Pipeline Magnetic Flux Leakage (MFL) Detection Performance With Mixed Attention Mechanisms (AMs) and Deep Residual Shrinkage Networks (DRSNs). IEEE Sensors Journal24, 4 (2024), 5162–5171. doi:10.1109/JSEN.2023.3347510

    work page doi:10.1109/jsen.2023.3347510 2024

  46. [46]

    He Zhao, Jinhai Liu, Jianhua Tang, Xiangkai Shen, Senxiang Lu, and Qiannan Wang. 2023. A MFL Mechanism-Based Self-Supervised Method for Defect Detec- tion With Limited Labeled Samples.IEEE Transactions on Instrumentation and Measurement72 (2023), 1–10. doi:10.1109/TIM.2022.3212041

    work page doi:10.1109/tim.2022.3212041 2023

  47. [47]

    He Zhao, Jinhai Liu, Huaguang Zhang, Qiannan Wang, Xiangkai Shen, and Xin Zhang. 2025. A Masked Multi-View Bidirectional Network for Class Ex- tremely Imbalanced Object Detection Under Magnetic Flux Leakage Signals. IEEE Transactions on Automation Science and Engineering22 (2025), 905–915. doi:10.1109/TASE.2024.3355984 10 PipeMFL-240K: A Large-scale Datas...

    work page doi:10.1109/tase.2024.3355984 2025

  48. [48]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp

    Yian Zhao, Wenyu Lv, Shangliang Xu, Jinman Wei, Guanzhong Wang, Qingqing Dang, Yi Liu, and Jie Chen. 2024. DETRs Beat YOLOs on Real-time Object Detec- tion. In2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 16965–16974. doi:10.1109/CVPR52733.2024.01605 11 Arxiv Preprint, February 04, 2026, 2602.07044 Tianyi Qu et al. A PipeMFL-...

    work page doi:10.1109/cvpr52733.2024.01605 2024