Acta Photonica Sinica, Volume. 53, Issue 1, 0130002(2024)

Infrared Detection of Gas Leaks Incorporating Structural Reparametric Transformations

Hong ZHUANG... Yinhui ZHANG, Zifen HE* and Huizhu CAO |Show fewer author(s)
Author Affiliations
  • Mechanical and Electrical Engineering,Kunming University of Science and Technology,Kunming 650500,China
  • show less
    Figures & Tables(22)
    GRNet model general architecture
    Change diagram of real frame of infrared the gas leakage
    Visualization of gas leakage candidate box clustering analysis
    Aspect ratio visualization results of candidate frames for gas leak infrared detection dataset
    Mosaic data enhancement
    Gamma transform rendering
    Image enhancement preprocessing schematic
    Fitting diagram of real box and prediction box
    CIoU positioning loss
    Structure of feature extraction network reconstructed by RepVGG module
    RepVGG module structure diagram
    The final test result of the GRNet network model
    Visualisation of ammonia leak concentration distribution
    Visualisation of the overall interface of the infrared gas leak detection system
    • Table 1. The size of the initial anchor frame of the three detection layers before and after clustering

      View table
      View in Article

      Table 1. The size of the initial anchor frame of the three detection layers before and after clustering

      Detection layerBefore clusteringAfter clustering
      Detection layer 1(10,13),(16,30),(33,23)(11,10),(29,12),(34,29)
      Detection layer 2(30,61),(62,45),(59,119)(52,61),(62,18),(64,38)
      Detection layer 3(116,90),(156,198),(373,326)(91,38),(115,22),(201,45)
    • Table 2. Hyperparameter configuration

      View table
      View in Article

      Table 2. Hyperparameter configuration

      Hyperparameter nameHyperparameter values
      Batch size16
      Learn rate0.01
      Epoch400
      Momentum0.937
      PolicyCosine annealing strategy
      Weight decay0.000 5
    • Table 3. Comparison of the verification performance of different localization loss functions for leaky targets

      View table
      View in Article

      Table 3. Comparison of the verification performance of different localization loss functions for leaky targets

      ModelParams/MBModel size/MBTime/msmAP/%
      YOLOv5s-GIoU7.0514.403.8092.20
      YOLOv5s-CIoU7.0514.403.6092.80
    • Table 4. Comparison of the network performance before and after image pre-processing

      View table
      View in Article

      Table 4. Comparison of the network performance before and after image pre-processing

      ModelParams/MBModel size/MBTime/msmAP/%
      YOLOv5s-GIoU7.0514.403.8092.20
      YOLOv5s-CIoU7.0514.403.6092.80
      YOLOv5s-CIoU-MG7.0514.403.6093.40
    • Table 5. Comparison of network performance before and after clustering

      View table
      View in Article

      Table 5. Comparison of network performance before and after clustering

      ModelParams/MBModelsize/MBmAP/%
      YOLOv5s-GIoU7.0514.4092.20
      YOLOv5s-GIoU-Km7.0514.4093.20
      YOLOv5s-CIoU-Km7.0514.4093.70
      YOLOv5s-CIoU-MG-Km7.0514.4094.00
    • Table 6. Different modules embedded network performance verification comparison

      View table
      View in Article

      Table 6. Different modules embedded network performance verification comparison

      ModelParams/MBModel size/MBTime/msmAP/%
      Conv-YOLOv5s7.0514.403.8092.20
      DWConv-YOLOv5s4.088.503.1093.90
      ODConv-YOLOv5s5.5011.303.7094.20
      GRNet5.4711.303.4094.90
    • Table 7. Comparison of the validation accuracy among various network models

      View table
      View in Article

      Table 7. Comparison of the validation accuracy among various network models

      ModelGFLOPsParams/MBModel size/MBTime/msmAP/%
      YOLOv3154.761.50123.4011.7092.70
      YOLOv3-tiny12.908.7017.403.4037.40
      YOLOv5s16.307.0514.403.8092.20
      YOLOx26.648.9471.908.4389.78
      GRNet15.605.4711.303.4094.90
    • Table 8. Embedded platform deployment test results

      View table
      View in Article

      Table 8. Embedded platform deployment test results

      ModelSpeed/(frame·s-1mAP/%
      YOLOv30.7692.70
      YOLOv5s2.4392.20
      GRNet3.0394.90
    Tools

    Get Citation

    Copy Citation Text

    Hong ZHUANG, Yinhui ZHANG, Zifen HE, Huizhu CAO. Infrared Detection of Gas Leaks Incorporating Structural Reparametric Transformations[J]. Acta Photonica Sinica, 2024, 53(1): 0130002

    Download Citation

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

    Category:

    Received: May. 26, 2023

    Accepted: Sep. 4, 2023

    Published Online: Feb. 1, 2024

    The Author Email: HE Zifen (zyhhzf1998@163.com)

    DOI:10.3788/gzxb20245301.0130002

    Topics