Spectroscopy and Spectral Analysis, Volume. 42, Issue 6, 1965(2022)

Identification of Two-Dimensional Material Nanosheets Based on Deep Neural Network and Hyperspectral Microscopy Images

Ren-miao PENG1,*... Peng-peng XU2,2;, Yi-mo ZHAO2,2;, Li-jun BAO1,1; and Cheng LI2,2; *; |Show fewer author(s)
Author Affiliations
  • 11. School of Electronic Science and Engineering, Xiamen University, Xiamen 361005, China
  • 22. College of Physical Science and Technology, Xiamen University, Xiamen 361005, China
  • show less
    Figures & Tables(14)
    Characterization of MoS2 sample(a), (d): MoS2 optical images; (b), (e): AFM images; (c), (f): Thicknesses obtained by AFM analysis
    Partial labels of MoS2 nanosheets marked by different thickness ranges
    Processed optical images and corresponding label images
    Diagram of Network structure
    Structure of residual convolution network
    Structure of pyramid pooling model
    Schematic diagram of low-level feature map reuse
    Loss curve and accuracy curve during 2D-Net network training
    Confusion matrix of test results
    The prediction results of the four network structures
    Output results of feature map visualization(a): MoS2 optical images and F2 feature maps; (b): MoS2 optical images and F5 feature maps
    Transfer learning for exfoliated grapheme(a): Predicted results; (b): Confusion matrix of testing
    • Table 1. Network parameters in the encoding and decoding stages

      View table
      View in Article

      Table 1. Network parameters in the encoding and decoding stages

      编码过程解码过程
      B1Conv(3×3, 64)×2[F1]Conv(3×3, 64)×2
      B2Conv(3×3, 128)×2[F2, 1/2]Conv(3×3, 128)×2
      B3Conv(3×3, 256)×3[F3, 1/4]Conv(3×3, 256)×3
      B4Conv(3×3, 512)×3[F4, 1/8]Conv(3×3, 512)×3
      B5Conv(3×3, 512)×3[F5, 1/16]Conv(3×3, 512)×3
    • Table 2. Test results of four network structures

      View table
      View in Article

      Table 2. Test results of four network structures

      PAMPAMIoU
      FCN894.48%87.11%65.63%
      SegNet95.28%88.51%72.34%
      U-Net95.66%88.91%73.29%
      2D-Net97.38%90.38%75.86%
    Tools

    Get Citation

    Copy Citation Text

    Ren-miao PENG, Peng-peng XU, Yi-mo ZHAO, Li-jun BAO, Cheng LI. Identification of Two-Dimensional Material Nanosheets Based on Deep Neural Network and Hyperspectral Microscopy Images[J]. Spectroscopy and Spectral Analysis, 2022, 42(6): 1965

    Download Citation

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

    Category: Research Articles

    Received: May. 23, 2021

    Accepted: --

    Published Online: Nov. 14, 2022

    The Author Email: PENG Ren-miao (prm649286898@126.com)

    DOI:10.3964/j.issn.1000-0593(2022)06-1965-09

    Topics