CoChat
Sign Up
Cited by 24Infrastructure Maintenance and MonitoringIndustrial Vision Systems and Defect DetectionAdvanced Neural Network Applications

The DOI matches its Crossref record; 33 of the 33 references it cites were found in a registry or an index.

Detection of Missing Bolts for Engineering Structures in Natural Environment Using Machine Vision and Deep Learning

Open access
Yang, Zhenglin, Zhao, Yadian, Xu, Chao·2023·Sensors
DOI 10.3390/s23125655PMC PMC10304286
Open this paper in CoChat — free accountOpen PDF

Abstract

The development of an accurate and efficient method for detecting missing bolts in engineering structures is crucial. To this end, a missing bolt detection method that leveraged machine vision and deep learning was developed. First, a comprehensive dataset of bolt images captured under natural conditions was constructed, which improved the generality and recognition accuracy of the trained bolt target detection model. Second, three deep learning network models, namely, YOLOv4, YOLOv5s, and YOLOXs, were compared, and YOLOv5s was selected as the bolt target detection model. With YOLOv5s as the target recognition model, the bolt head and bolt nut had average precisions of 0.93 and 0.903, respectively. Third, a missing bolt detection method based on perspective transformation and IoU was presented and validated under laboratory conditions. Finally, the proposed method was applied to an actual footbridge structure to test its feasibility and effectiveness in real engineering scenarios. The experimental results showed that the proposed method could accurately identify bolt targets with a confidence level of over 80% and detect missing bolts under different image distances, perspective angles, light intensities, and image resolutions. Moreover, the experimental results on a footbridge demonstrated that the proposed method could reliably detect the missing bolt even at a shooting distance of 1 m. The proposed method provided a low-cost, efficient, and automated technical solution for the safety management of bolted connection components in engineering structures.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.3390/s23125655

Cites

33 of 33 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 Four categories in bolt target detection and the corresponding labeled images.
Figure 1 Four categories in bolt target detection and the corresponding labeled images.CC BY · PubMed Central
Figure 2 Number of categories of statistics.
Figure 2 Number of categories of statistics.CC BY · PubMed Central
Figure 3 YOLOv4 network structure.
Figure 3 YOLOv4 network structure.CC BY · PubMed Central
Figure 4 YOLOv5s network structure.
Figure 4 YOLOv5s network structure.CC BY · PubMed Central
Figure 5 YOLOXs network structure.
Figure 5 YOLOXs network structure.CC BY · PubMed Central
Figure 6 Perspective transformation with 30° tilt angle for bolt image reconstruction. (a–d) 30° tilt in the horizontal and vertical directions and transformed images, (e) reference image.
Figure 6 Perspective transformation with 30° tilt angle for bolt image reconstruction. (a–d) 30° tilt in the horizontal and vertical directions and transformed images, (e) reference image.CC BY · PubMed Central
Figure 7 Intersection over union (IoU).
Figure 7 Intersection over union (IoU).CC BY · PubMed Central
Figure 8 Flowchart of the developed framework.
Figure 8 Flowchart of the developed framework.CC BY · PubMed Central
Figure 9 Results of comparing YOLOv4, YOLOv5s, and YOLOXs algorithms for bolt detection.
Figure 9 Results of comparing YOLOv4, YOLOv5s, and YOLOXs algorithms for bolt detection.CC BY · PubMed Central
Figure 10 Performance of YOLOv4, YOLOv5s, and YOLOXs.
Figure 10 Performance of YOLOv4, YOLOv5s, and YOLOXs.CC BY · PubMed Central
Figure 11 Experimental structure.
Figure 11 Experimental structure.CC BY · PubMed Central
Figure 12 Detection results from different perspectives horizontally: (a) 0°, (b) 10°, (c) 20°, and (d) 30°.
Figure 12 Detection results from different perspectives horizontally: (a) 0°, (b) 10°, (c) 20°, and (d) 30°.CC BY · PubMed Central
Figure 13 Detection results from different perspectives vertically: (a) 0°, (b) 10°, (c) 20°, and (d) 30°.
Figure 13 Detection results from different perspectives vertically: (a) 0°, (b) 10°, (c) 20°, and (d) 30°.CC BY · PubMed Central
Figure 14 The detection results with different distance settings: (a) near, (b) medium, and (c) far.
Figure 14 The detection results with different distance settings: (a) near, (b) medium, and (c) far.CC BY · PubMed Central
Figure 15 Detection results with different lighting conditions: (a) dark with flash on, (b) dim, (c) normal, and (d) bright.
Figure 15 Detection results with different lighting conditions: (a) dark with flash on, (b) dim, (c) normal, and (d) bright.CC BY · PubMed Central
Figure 16 Results with different image qualities: (a) lens blur 0, (b) lens blur 10, (c) lens blur 20, and (d) lens blur 25.
Figure 16 Results with different image qualities: (a) lens blur 0, (b) lens blur 10, (c) lens blur 20, and (d) lens blur 25.CC BY · PubMed Central
Figure 17 Reference and target images: (a) reference image, (b) target image from 30° vertical perspective, (c) target image from 30° horizontal angle, and (d) target image at 45 cm distance.
Figure 17 Reference and target images: (a) reference image, (b) target image from 30° vertical perspective, (c) target image from 30° horizontal angle, and (d) target image at 45 cm distance.CC BY · PubMed Central
Figure 18 Results of missing detection with different conditions: (a) target image from 30° vertical angle, (b) target image from 30° horizontal angle, and (c) target image at 45 cm distance.
Figure 18 Results of missing detection with different conditions: (a) target image from 30° vertical angle, (b) target image from 30° horizontal angle, and (c) target image at 45 cm distance.CC BY · PubMed Central
Figure 19 Pedestrian bridge for testing. (a) Bridge location (from Amap), (b) Real view of the bridge.
Figure 19 Pedestrian bridge for testing. (a) Bridge location (from Amap), (b) Real view of the bridge.CC BY · PubMed Central
Figure 20 Image capture of bolted joints by UAV. (a) UAV approaching bolted connection and (b) target bolts.
Figure 20 Image capture of bolted joints by UAV. (a) UAV approaching bolted connection and (b) target bolts.CC BY · PubMed Central
Figure 21 Bolt identification results: (a) small perspective (perspective angle ≤ 5°) angle bolt recognition and (b) perspective bolt identification.
Figure 21 Bolt identification results: (a) small perspective (perspective angle ≤ 5°) angle bolt recognition and (b) perspective bolt identification.CC BY · PubMed Central
Figure 22 Detection of target images with perspective and missing detection results. (a) target images with perspective, (b) missing detection results.
Figure 22 Detection of target images with perspective and missing detection results. (a) target images with perspective, (b) missing detection results.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 33 references

Research works printed but not linked: 6

  1. Ramana, L., Choi, W., and Cha, Y.-J. (2017). Sensors and Instrumentation, Springer.
  2. Liu, Y., Huo, L.-s., and Song, G.-b. (2018, January 27–31). Automatic detection on the bolt loose based on digital image processing. Proceedings of the 2018 Structures Congress, Incheon, Republic of Korea.
  3. Ge, Z., Liu, S., Wang, F., Li, Z., and Sun, J. (2021). Yolox: Exceeding yolo series in 2021. arXiv.
  4. Redmon, J., and Farhadi, A. (2018). Yolov3: An incremental improvement. arXiv.
  5. Bochkovskiy, A., Wang, C.-Y., and Liao, H.-Y.M. (2020). Yolov4: Optimal speed and accuracy of object detection. arXiv.
  6. Misra, D. (2019). Mish: A self regularized non-monotonic neural activation function. arXiv.

Cited by

Sorted by citations

Related

See all

Metadata

Authors
Yang, Z., Zhao, Y., & Xu, C.
Affiliations
Northwestern Polytechnical University (CN)
Corresponding author
Xu, Chao
ORCID
Venue
Sensors
Publisher
MDPI AG
Published
2023-06-16
Volume
23
Issue
12
Pages
5655
ISSN
1424-8220 (ISSN-L)
DOI
10.3390/s23125655
PMC
PMC10304286
OpenAlex
W4380993029
Cited by
22
References
33
Citations by year
2026: 7 · 2025: 9 · 2024: 6
FWCI
2.57
Type
journal article
Language
en
Open access
Licence
Locations
  • Sensors (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
  • Suzhou Science and Technology Funding (SWY2021003, S202210699639)
  • Provincial College Students’ innovation and entrepreneurship training program (SWY2021003, S202210699639)
Topics
  • Physical Sciences › Engineering › Civil and Structural Engineering › Infrastructure Maintenance and Monitoring
  • Physical Sciences › Engineering › Industrial and Manufacturing Engineering › Industrial Vision Systems and Defect Detection
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
Keywords
bolted joints, engineering structures, pedestrian bridge, YOLOv5, object detection, homography, intersection over union, structural safety monitoring, image dataset construction
MeSH terms
Deep Learning, Engineering, Environment, Laboratories, Light

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review