Opto-Electronic Engineering, Volume. 51, Issue 5, 240044(2024)

Improvement of GBS-YOLOv7t for steel surface defect detection

Liming Liang... Pengwei Long*, Baohe Lu and Renjie Li |Show fewer author(s)
Author Affiliations
  • School of Electrical Engineering and Automation, Jiangxi University of Science and Technology, Ganzhou, Jiangxi 341000, China
  • show less
    Figures & Tables(15)
    Network structure of GBS-YOLOV7t
    GAC-FPN network structure
    GSConv network structure
    BRA network structure
    Calculation method of SIoU loss function
    Images of various defects on steel surface
    Comparison of AP values of GAC-FPN, PANet and AFPN networks for detecting various types of defects
    Comparison of detection results of the improved algorithm
    Comparison of detection effect between the proposed algorithm and other algorithms
    • Table 1. Results of GAS-FPN ablation experiment

      View table
      View in Article

      Table 1. Results of GAS-FPN ablation experiment

      ABmAP/%Params/MFLOPs/GFPS
      69.27.1114.1106.38
      70.77.4414.694.34
      71.96.5613.2111.11
    • Table 2. Comparison experiment of GAS-FPN

      View table
      View in Article

      Table 2. Comparison experiment of GAS-FPN

      ModelmAP/%Params/MFPS
      PANet68.76.02108.12
      AFPN69.27.11106.38
      GAC-FPN71.96.56111.11
    • Table 3. Experimental results of BRA position

      View table
      View in Article

      Table 3. Experimental results of BRA position

      LocationmAP/%Params/MFPS
      Baseline68.76.02108.69
      Stage369.06.0979.37
      Stage469.96.20104.17
      Stage569.37.08111.11
    • Table 4. BRA comparison experiments

      View table
      View in Article

      Table 4. BRA comparison experiments

      ModelmAP/%Params/MFLOPs/G
      Baseline68.76.0213.1
      SE69.311.5730.8
      TA68.96.0213.2
      CA68.36.0313.5
      BRA (Ours)69.96.2013.2
    • Table 5. Results of ablation experiment

      View table
      View in Article

      Table 5. Results of ablation experiment

      M1M2M3mAP/%Params/MP/%R/%
      68.76.0261.472.7
      71.96.5663.870.2
      72.56.5665.373.5
      72.96.8369.970.5
    • Table 6. Comparison of the experimental results

      View table
      View in Article

      Table 6. Comparison of the experimental results

      ModelmAP/%Params/MFPS
      Faster R-CNN65.772.017.8
      SSD61.024.441.0
      YOLOv367.061.531.5
      YOLOv451.052.545.0
      YOLOv5s70.17.07102.0
      YOLOX-s71.88.046.0
      YOLOv770.037.236.1
      YOLOv7-tiny68.76.02108.1
      FCOS68.843.212.0
      RetinaNet69.518.315.1
      PC-YOLOv7[13]71.25.9761.0
      文献[22]73.09.54
      文献[23]74.123.975
      GBS-YOLOv7t72.96.83104.1
    Tools

    Get Citation

    Copy Citation Text

    Liming Liang, Pengwei Long, Baohe Lu, Renjie Li. Improvement of GBS-YOLOv7t for steel surface defect detection[J]. Opto-Electronic Engineering, 2024, 51(5): 240044

    Download Citation

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

    Category: Article

    Received: Feb. 29, 2024

    Accepted: Mar. 22, 2024

    Published Online: Jul. 31, 2024

    The Author Email: Long Pengwei (龙鹏威)

    DOI:10.12086/oee.2024.240044

    Topics