Optics and Precision Engineering, Volume. 30, Issue 1, 117(2022)

Transfer learning techniques for semantic segmentation of machine vision inspection and identification based on label-reserved Softmax algorithms

Guixiong LIU* and Jian HUANG
Author Affiliations
  • School of Mechanical and Automotive Engineering, South China University of Technology, Guangzhou510640, China
  • show less
    Figures & Tables(11)
    Fine-tuning transfer learning framework based on label-reserved improved Softmax algorithms
    Model of label-reserved improved Softmax algorithms
    Head architecture of label-reserved Mask R-CNN
    Comparison of different migration learning methods
    Semantic segmentation results of chassis assemblies
    Basic functions of MVAQ2 chassis standard parts assembly quality inspection and identification software
    MVAQ2 chassis standard parts assembly quality inspection and identification devices
    Chassis learning process of MVAQ2 devices
    • Table 1. Ttrain and LCE of pre-trained Mask R-CNN model fine-tuning transfer learning initial various weights WCNN

      View table
      View in Article

      Table 1. Ttrain and LCE of pre-trained Mask R-CNN model fine-tuning transfer learning initial various weights WCNN

      源数据集Dsource目标数据集Dtarget模型结构调整模型初始权值WCNNLCETtrain/min
      ImageNet7iFixit18输出层Wmain4.892258
      COCO8iFixit18输出层WmainWseg1.542115
      iFixit18iFixit18WmainWseg1.29785
      iFixit18iFixit18WmainWsegWcla0.01958
    • Table 2. Parameter comparison of label-reserved and general Softmax algorithms

      View table
      View in Article

      Table 2. Parameter comparison of label-reserved and general Softmax algorithms

      参 数典型Softmax算法17标签预留改进Softmax算法
      输入特征向量nsource×1维Xnsource+nreserved)×1维X'
      分类器权值

      nsourcensource×1维

      Wcla={W1,,Wnsource}

      nsource+nreserved)个(nsource+nreserved)×1维

      W'cla={W1,,Wnsource+nreserved}

      输出概率向量nsource×1维Pnsource+nreserved)×1维P'
      标签集C'cla=CsourceC'cla=CsourceCreversed
    • Table 3. Training time ofdifferent transfer learning methods

      View table
      View in Article

      Table 3. Training time ofdifferent transfer learning methods

      方 法源数据集Dsource目标数据集Dtarget模型结构调整Ttrain/min
      Fine-tuning17不可扩展机箱可扩展机箱输出层42.8
      特征提取Fine-tuning20不可扩展机箱可扩展机箱输出层36.5
      本文方法不可扩展机箱可扩展机箱30.1
    Tools

    Get Citation

    Copy Citation Text

    Guixiong LIU, Jian HUANG. Transfer learning techniques for semantic segmentation of machine vision inspection and identification based on label-reserved Softmax algorithms[J]. Optics and Precision Engineering, 2022, 30(1): 117

    Download Citation

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

    Category: Information Sciences

    Received: May. 16, 2021

    Accepted: --

    Published Online: Jan. 20, 2022

    The Author Email: LIU Guixiong (megxliu@scut.edu.cn)

    DOI:10.37188/OPE.20223001.0117

    Topics