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
. 2019 May 23;12(10):1681.
doi: 10.3390/ma12101681.

Research on Recognition Technology of Aluminum Profile Surface Defects Based on Deep Learning

Ruofeng Wei  1 , Yunbo Bi  2   3
Affiliations

Research on Recognition Technology of Aluminum Profile Surface Defects Based on Deep Learning

Ruofeng Wei et al. Materials (Basel). .

Abstract

Aluminum profile surface defects can greatly affect the performance, safety, and reliability of products. Traditional human-based visual inspection has low accuracy and is time consuming, and machine vision-based methods depend on hand-crafted features that need to be carefully designed and lack robustness. To recognize the multiple types of defects with various size on aluminum profiles, a multiscale defect-detection network based on deep learning is proposed. Then, the network is trained and evaluated using aluminum profile surface defects images. Results show 84.6%, 48.5%, 96.9%, 97.9%, 96.9%, 42.5%, 47.2%, 100%, 100%, and 43.3% average precision (AP) for the 10 defect categories, respectively, with a mean AP of 75.8%, which illustrate the effectiveness of the network in aluminum profile surface defects detection. In addition, saliency maps also show the feasibility of the proposed network.

Keywords: aluminum profile surface defects; average precision (AP); deep learning; multiscale defect-detection network; saliency maps.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Figure 1
Figure 1
Images of aluminum profile surface defects.
Figure 2
Figure 2
Diagram of the number of images with defects in each category.
Figure 3
Figure 3
Data augmentation.
Figure 4
Figure 4
Architecture of the multiscale defect-detection network.
Figure 5
Figure 5
Structure of the bottleneck.
Figure 6
Figure 6
Architecture of feature fusion in the multiscale defect-detection network.
Figure 7
Figure 7
Structure of the improved RPN.
Figure 8
Figure 8
(a) Fifteen kinds of anchors; (b) anchors set on the input image.
Figure 9
Figure 9
(a) The schematic diagram of the regression layer; (b) ROI pooling.
Figure 10
Figure 10
(a) The total_loss curve during the process of network training; and (b) the roi_cls_loss, roi_loc_loss, rpn_cls_loss, and rpn_loc_loss curves during the training.
Figure 11
Figure 11
Average precisions (APs) for ten types of defects of Faster R-CNN and the multiscale defect-detection network.
Figure 12
Figure 12
(a) Images with a single type of defect; (b) images with multiple types of defects.
Figure 12
Figure 12
(a) Images with a single type of defect; (b) images with multiple types of defects.
Figure 13
Figure 13
(a) Aluminum profile defect images; (b) saliency maps.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Liu Z., Li L., Yi J., Li S., Wang Z., Wang G. Influence of Heat Treatment Conditions on Bending Characteristics of 6063 Aluminum Alloy Sheets. Trans. Nonferrous Met. Soc. China. 2017;27:1498–1506. doi: 10.1016/S1003-6326(17)60170-5. - DOI
    1. Bingöl S., Bozacı A. Experimental and Numerical Study on the Strength of Aluminum Extrusion Welding. Materials. 2015;8:4389–4399. doi: 10.3390/ma8074389. - DOI - PMC - PubMed
    1. Liu Z., Li L., Li S., Yi J., Wang G. Simulation Analysis of Porthole Die Extrusion Process and Die Structure Modifications for an Aluminum Profile with High Length−Width Ratio and Small Cavity. Materials. 2018;11:1517. doi: 10.3390/ma11091517. - DOI - PMC - PubMed
    1. Tony W.A. Automated inspection of metal products not quite ready for prime time. Iron Steelmak. 1992;19:14–19.
    1. Kong X., Li J. Image Registration-Based Bolt Loosening Detection of Steel Joints. Sensors. 2018;18:1000. doi: 10.3390/s18041000. - DOI - PMC - PubMed

LinkOut - more resources