Laser & Optoelectronics Progress, Volume. 58, Issue 2, 0210022(2021)

Prohibited Item Identification Algorithm Based onLightweight Segmentation Network

Shaoqing Yao1 and Zhigang Su1,2、*
Author Affiliations
  • 1Tianjin Key Laboratory for Advanced Signal Processing, Civil Aviation University of China, Tianjin 300300, China
  • 2Sino-European Institute of Aviation Engineering, Civil Aviation University of China, Tianjin 300300, China
  • show less
    Figures & Tables(9)
    Prohibited item identification model
    Results of data augmentation
    Dilated convolution module. (a) Without feature multiplexing operation; (b) with 3-layer convolution; (c) with 2-layer convolution
    Asymmetric convolution module. (a) Downsampling; (b) upsampling
    Example of images in dataset
    Segmentation results of contrast experiment
    Segmentation results of ablation experiment
    • Table 1. Results of contrast experiment

      View table

      Table 1. Results of contrast experiment

      AlgorithmmIoU /10-2Time /msFPS
      FCN8s62.151208.3
      PSPNet70.561526.5
      Deeplabv371.151626.1
      An's model74.512763.6
      ENet51.217128.7
      Proposed algorithm73.183627.1
    • Table 2. Results of ablation experiment

      View table

      Table 2. Results of ablation experiment

      AlgorithmmIoU /10-2Time /msFPS
      Group 163.286116.3
      Group 269.758411.8
      Group 368.305817
      Group 467.373329.5
      Group 567.844025
      Proposed algorithm73.183627.1
    Tools

    Get Citation

    Copy Citation Text

    Shaoqing Yao, Zhigang Su. Prohibited Item Identification Algorithm Based onLightweight Segmentation Network[J]. Laser & Optoelectronics Progress, 2021, 58(2): 0210022

    Download Citation

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

    Category: Image Processing

    Received: Jun. 24, 2020

    Accepted: Jul. 20, 2020

    Published Online: Jan. 8, 2021

    The Author Email: Su Zhigang (ssrsu@vip.sina.com)

    DOI:10.3788/LOP202158.0210022

    Topics