Journals Articles Conferences News About CLP
Submit a paper Email Alert
Search
Search
Articles
Journals
News
Advanced Search
Top Searches
2D MaterialsTransformation opticsQuantum PhotonicsSmall lasersSemiconductor UV PhotonicsSupersymmetric microring laser arrays

Laser & Optoelectronics Progress, Volume. 60, Issue 22, 2210009(2023)

YOLOv5-Based Lightweight Algorithm for Detecting Bottle-Cap Packaging Defects

Lei Zhao1,2,3、*, Likuan Jiao1,2,3、**, Ran Zhai1,2,3, Bin Li1,2,3, and Meiye Xu1,2,3
Author Affiliations
  • 1Tianjin Key Laboratory for Advanced Mechatronic System Design and Intelligent Control, Tianjin University of Technology, Tianjin 300384, China
  • 2National Demonstration Center for Experimental Mechanical and Electrical Engineering Education (Tianjin University of Technology), Tianjin 300384, China
  • 3School of Mechanical Engineering, Tianjin University of Technology, Tianjin 300384, China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (17)
    Equations (11)
    References (26)
    Cited By (3)
    Tools
    Paper Information
    Topics
    Figures & Tables(17)
    Pictures of bottle cap defects. (a) Break point; (b) broken edge; (c) deformation; (d) damage; (e) data error; (f) spinning

    Fig. 1. Pictures of bottle cap defects. (a) Break point; (b) broken edge; (c) deformation; (d) damage; (e) data error; (f) spinning

    Download full size
    Pictures of defects in complex situation

    Fig. 2. Pictures of defects in complex situation

    Download full size
    Structure of the original YOLOv5s model

    Fig. 3. Structure of the original YOLOv5s model

    Download full size
    CIoU loss

    Fig. 4. CIoU loss

    Download full size
    Improved deep learning model SEGC-YOLO

    Fig. 5. Improved deep learning model SEGC-YOLO

    Download full size
    Lightweight Backbone module

    Fig. 6. Lightweight Backbone module

    Download full size
    Channel Shuffle

    Fig. 7. Channel Shuffle

    Download full size
    Structure of the ECA mechanism

    Fig. 8. Structure of the ECA mechanism

    Download full size
    Structures of GhostConv-1, GhostConv-2, and C3-Ghost

    Fig. 9. Structures of GhostConv-1, GhostConv-2, and C3-Ghost

    Download full size
    Diagram of Ghost operation

    Fig. 10. Diagram of Ghost operation

    Download full size
    Structure of CARAFE upsampling operator

    Fig. 11. Structure of CARAFE upsampling operator

    Download full size
    Analog inspection equipment

    Fig. 12. Analog inspection equipment

    Download full size
    mAP and loss curves of YOLOv5s and SEGC-YOLO. (a) mAP@0.5; (b) mAP@0.5∶0.95; (c) loss

    Fig. 13. mAP and loss curves of YOLOv5s and SEGC-YOLO. (a) mAP@0.5; (b) mAP@0.5∶0.95; (c) loss

    Download full size
    Comparison of defect detection using different algorithms

    Fig. 14. Comparison of defect detection using different algorithms

    Download full size

    • Table 1. Data of ablation experiments

      View table

      Table 1. Data of ablation experiments

      No.BackboneAdamECAGhostCARAFEFLOPs /109Parameters /106Model file /MBmAP@0.5 /%mAP@0.5∶0.95 /%
      116.37.0714.582.948.5
      2√6.93.316.980.646.5
      3√√6.93.316.982.748.3
      4√√√6.93.317.081.848.6
      5√√√√4.41.914.282.748.6
      6√√√√√4.92.044.484.149.0
      7√√√√4.92.044.483.448.3

    • Table 2. Comparative data on multiple attention mechanisms

      View table

      Table 2. Comparative data on multiple attention mechanisms

      Attention mechanismFLOPs /109Parameters /106Model file /MBmAP@0.5 /%mAP@0.5∶0.95 /%
      ECA［17］4.92.044.484.149.0
      SE［20］5.42.785.883.949.0
      CBAM［21］6.02.735.882.147.7
      CA［22］6.43.086.583.648.7
      ShuffleAttention［23］4.92.044.481.748.6
      NAM［24］4.92.044.582.948.0

    • Table 3. Comparative experimental data of multiple algorithms

      View table

      Table 3. Comparative experimental data of multiple algorithms

      AlgorithmFLOPs /109Parameters /106Model file /MBmAP@0.5 /%mAP@0.5∶0.95 /%Inference time /ms
      YOLOv5s16.37.0714.582.948.516.3
      SEGC-YOLO4.92.044.484.149.015.8
      YOLOv3-tiny13.08.6817.582.346.18.3
      YOLOv3155.361.55123.683.548.434.3
      YOLOv7105.237.2274.884.149.041.4
    Tools

    Get Citation

    Copy Citation Text

    Lei Zhao, Likuan Jiao, Ran Zhai, Bin Li, Meiye Xu. YOLOv5-Based Lightweight Algorithm for Detecting Bottle-Cap Packaging Defects[J]. Laser & Optoelectronics Progress, 2023, 60(22): 2210009

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Set citation alerts for article
    Save article for my favorites
    Paper Information

    Category: Image Processing

    Received: May. 6, 2023

    Accepted: Jun. 25, 2023

    Published Online: Nov. 6, 2023

    The Author Email: Zhao Lei (leizhaotjut@163.com), Jiao Likuan (1913194980@qq.com)

    DOI:10.3788/LOP231231

    Topics

    laser devices and laser physics
    Lasers and Laser Optics
    Laser physics
    laser manufacturing
    Instrumentation, Measurement and Metrology