• Laser & Optoelectronics Progress
  • Vol. 58, Issue 14, 1415004 (2021)
Bin Li, Cheng Wang*, Jing Wu, Jichao Liu, Lijia Tong, and Zhenping Guo
Author Affiliations
  • Fundamentals Department, Air Force Engineering University, Xi’an, Shaanxi 710038, China
  • show less
    DOI: 10.3788/LOP202158.1415004 Cite this Article Set citation alerts
    Bin Li, Cheng Wang, Jing Wu, Jichao Liu, Lijia Tong, Zhenping Guo. Surface Defect Detection of Aeroengine Components Based on Improved YOLOv4 Algorithm[J]. Laser & Optoelectronics Progress, 2021, 58(14): 1415004 Copy Citation Text show less
    YOLOv4 network structure
    Fig. 1. YOLOv4 network structure
    Improved YOLOv4 network structure
    Fig. 2. Improved YOLOv4 network structure
    Type of defect. (a) Crack; (b) gap; (c) pit; (d) scratch
    Fig. 3. Type of defect. (a) Crack; (b) gap; (c) pit; (d) scratch
    Data enhancement diagram. (a) Original picture; (b) horizontal flip; (c) exposure adjustment; (d) Mosaic data enhancement
    Fig. 4. Data enhancement diagram. (a) Original picture; (b) horizontal flip; (c) exposure adjustment; (d) Mosaic data enhancement
    Data labeling diagram. (a) Original picture; (b) picture annotation example; (c) xml tag file
    Fig. 5. Data labeling diagram. (a) Original picture; (b) picture annotation example; (c) xml tag file
    Loss function curve
    Fig. 6. Loss function curve
    Various defect detection results. (a) (e) Crack; (b) (f) gap; (c) (g) pit; (d) (h) scratch
    Fig. 7. Various defect detection results. (a) (e) Crack; (b) (f) gap; (c) (g) pit; (d) (h) scratch
    Comparison of the AP of different algorithms under the original YOLOv4 network
    Fig. 8. Comparison of the AP of different algorithms under the original YOLOv4 network
    Comparison of the Fβ of different algorithms under the original YOLOv4 network
    Fig. 9. Comparison of the Fβ of different algorithms under the original YOLOv4 network
    Comparison of the AP of different networks under improved parameter adjustment algorithm
    Fig. 10. Comparison of the AP of different networks under improved parameter adjustment algorithm
    Comparison of the Fβ of different networks under improved parameter adjustment algorithm
    Fig. 11. Comparison of the Fβ of different networks under improved parameter adjustment algorithm
    Network structureAlgorithm modelmAP /%Time /s
    Original YOLOv4K-means78.120.121
    Original YOLOv4Improved algorithm80.190.121
    Improved YOLOv4K-means81.830.123
    Improved YOLOv4Improved algorithm82.670.124
    Table 1. Improve the performance comparison of YOLOv4
    Network structuremAP /%Time /sFβ /%
    CrackGapPitScratch
    Faster R-CNN+VGG1657.130.465344.6036.2731.5931.66
    Faster R-CNN+ResNet-5064.810.802544.0939.8734.8430.81
    YOLOv375.450.126285.1384.7267.8864.70
    YOLOv478.120.121076.6883.9265.4564.13
    Improved YOLOv482.670.124082.9792.2371.9967.05
    Table 2. Comparison of detection performance of different algorithms
    Bin Li, Cheng Wang, Jing Wu, Jichao Liu, Lijia Tong, Zhenping Guo. Surface Defect Detection of Aeroengine Components Based on Improved YOLOv4 Algorithm[J]. Laser & Optoelectronics Progress, 2021, 58(14): 1415004
    Download Citation