Laser & Optoelectronics Progress, Volume. 62, Issue 6, 0637010(2025)

SGS-YOLO: Method for Detecting Dress Code Violations by Airport Security Personnel

Zeping Deng*, Hui Liu, Jiliang Tu, Shenhui Ye, Naizhi Liao, and Guochao Lai
Author Affiliations
  • School of Information Engineering, Nanchang Hangkong University, Nanchang 330063, Jiangxi , China
  • show less
    Figures & Tables(15)
    The SGS-YOLO network architecture
    SimAM attention module structure
    GSConv module structure
    GSbottleneck module and VoV-GSCSP architecture
    Schematic of the angle loss
    Schematic of the distance loss
    Partial dataset samples
    mAP@0.5 comparison
    Comparison of the accuracy of different detection algorithms
    Comparison of detection results between YOLOv8n and SGS-YOLO. (a) Original images; (b) YOLOv8n detection results; (c) SGS-YOLO detection results
    • Table 1. Hyperparameter setting

      View table

      Table 1. Hyperparameter setting

      Parameter nameParameter value
      Learning0.01
      Momentum0.937
      Weight decay0.0005
      Batch size8
      Epoch300
    • Table 2. Comparison of different attention mechanisms

      View table

      Table 2. Comparison of different attention mechanisms

      ModelAttention mechanismmAP@0.5 /%Params /106FPS
      YOLOv8n88.43.0192
      CA91.33.0597
      SE89.63.01105
      CBAM90.13.07102
      SimAM92.73.0190
    • Table 3. Comparison of different loss functions

      View table

      Table 3. Comparison of different loss functions

      ModelLoss functionmAP@ 0.5 /%Params /106FPS
      YOLOv8nCIOU88.43.0192
      EIOU88.73.0192
      DIOU87.83.0197
      SIOU89.13.01101
    • Table 4. Ablation experiments

      View table

      Table 4. Ablation experiments

      SchemeModelmAP@0.5 /%Params /106FLOPs /109FPS
      1YOLOv8n88.43.018.192
      2YOLOv8n+SimAM92.73.018.190
      3YOLOv8n+GSConv88.62.787.694
      4YOLOv8n+SIOU89.13.018.1101
      5YOLOv8n+SimAM+SIOU94.23.018.189
      6YOLOv8n+GSConv+VoV-GSCSP89.22.727.495
      7YOLOv8n+GSConv+SimAM+ SIOU94.52.787.694
      8YOLOv8n+GSConv+VoV-GSCSP+SimAM+ SIOU94.72.727.494
    • Table 5. Comparison of the different detection algorithms

      View table

      Table 5. Comparison of the different detection algorithms

      AlgorithmP /%mAP@0.5 /%Params /106FLOPs /109FPS
      Faster R-CNN82.579.841.3273.417
      SSD6.273.626.362.746
      YOLOv5s89.386.97.2115.978
      YOLOv788.692.537.30105.373
      YOLOv8n-MS94.793.12.914.495
      YOLOv8s97.495.311.1028.781
      SGS-YOLO96.294.72.727.494
    Tools

    Get Citation

    Copy Citation Text

    Zeping Deng, Hui Liu, Jiliang Tu, Shenhui Ye, Naizhi Liao, Guochao Lai. SGS-YOLO: Method for Detecting Dress Code Violations by Airport Security Personnel[J]. Laser & Optoelectronics Progress, 2025, 62(6): 0637010

    Download Citation

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

    Category: Digital Image Processing

    Received: Jul. 25, 2024

    Accepted: Aug. 28, 2024

    Published Online: Mar. 13, 2025

    The Author Email:

    DOI:10.3788/LOP241729

    CSTR:32186.14.LOP241729

    Topics