Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Mar 5;24(5):1674.
doi: 10.3390/s24051674.

An Infrared Image Defect Detection Method for Steel Based on Regularized YOLO

Yongqiang Zou  1 , Yugang Fan  1
Affiliations

An Infrared Image Defect Detection Method for Steel Based on Regularized YOLO

Yongqiang Zou et al. Sensors (Basel). .

Abstract

Steel surfaces often display intricate texture patterns that can resemble defects, posing a challenge in accurately identifying actual defects. Therefore, it is crucial to develop a highly robust defect detection model. This study proposes a defect detection method for steel infrared images based on a Regularized YOLO framework. Firstly, the Coordinate Attention (CA) is embedded within the C2F framework, utilizing a lightweight attention module to enhance the feature extraction capability of the backbone network. Secondly, the neck part design incorporates the Bi-directional Feature Pyramid Network (BiFPN) for weighted fusion of multi-scale feature maps. This creates a model called BiFPN-Concat, which enhances feature fusion capability. Finally, the loss function of the model is regularized to improve the generalization performance of the model. The experimental results indicate that the model has only 3.03 M parameters, yet achieves a mAP@0.5 of 80.77% on the NEU-DET dataset and 99.38% on the ECTI dataset. This represents an improvement of 2.3% and 1.6% over the baseline model, respectively. This method is well-suited for industrial detection applications involving non-destructive testing of steel using infrared imagery.

Keywords: YOLOv8; cross-entropy; defect detection; infrared image; regularization.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflicts of interest.

Figures

Figure 2
Figure 2
The module structure of Coordinate Attention.
Figure 3
Figure 3
The module structure of CA-C2F.
Figure 1
Figure 1
The network structure of infrared image defect detection model.
Figure 4
Figure 4
Pulse eddy current testing device.
Figure 5
Figure 5
Some steel defect samples in ECTI.
Figure 6
Figure 6
Model performance at different degrees of regularization.
Figure 7
Figure 7
The mAP@0.5 of Regularized YOLO compared with YOLOv8n.
Figure 8
Figure 8
The PR curve of Regularized YOLO on NEU-DET.
Figure 9
Figure 9
Detection plot of partial ECTI.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. Xian T., Wei H., De X. A Survey of Surface Defect Detection Methods Based on Deep Learning. Acta Autom. Sin. 2021;47:1017–1034.
    1. Suresh B.R., Fundakowski R.A., Levitt T.S., Overland J.E. A Real-time Automated Visual Inspection System for Hot Steel Slabs. IEEE Trans. Pattern Anal. Mach. Intell. 1983;PAMI-5:563–572. doi: 10.1109/TPAMI.1983.4767445. - DOI - PubMed
    1. Meng G., Liaolin H., Jiangtao Z. Surface Defect Detection Method of Ceramic Bowl Based on Kirsch and Canny Operator. Acta Opt. Sin. 2016;36:27–33.
    1. Nieniewski M. Morphological Detection and Extraction of Rail Surface Defects. IEEE Trans. Instrum. Meas. 2020;69:6870–6879. doi: 10.1109/TIM.2020.2975454. - DOI
    1. Shiyang Z. Ph.D. Thesis. Huazhong University of Science & Technology; Wuhan, China: 2017. Research on Method for Image of Surface Detect of Steel Sheet Based on Visual Saliency and Sparse Representation.