• Laser & Optoelectronics Progress
  • Vol. 59, Issue 18, 1810010 (2022)
Yuehua Yu, Haibo Zhang, Xin Li, Jiaojiao Kou, Kang Li, Guohua Geng, and Mingquan Zhou*
Author Affiliations
  • College of Information Science and Technology, Northwest University, Xi’an 710127, Shaanxi . China
  • show less
    DOI: 10.3788/LOP202259.1810010 Cite this Article Set citation alerts
    Yuehua Yu, Haibo Zhang, Xin Li, Jiaojiao Kou, Kang Li, Guohua Geng, Mingquan Zhou. Data Enhanced Depth Classification Model for Terracotta Warriors Fragments[J]. Laser & Optoelectronics Progress, 2022, 59(18): 1810010 Copy Citation Text show less
    Schematic diagram of Terracotta Warriors fragment data enhancement model structure
    Fig. 1. Schematic diagram of Terracotta Warriors fragment data enhancement model structure
    Comparison between new samples and original samples in some parts
    Fig. 2. Comparison between new samples and original samples in some parts
    Structure diagram of CBAM module
    Fig. 3. Structure diagram of CBAM module
    Channel attention module
    Fig. 4. Channel attention module
    Spatial attention module
    Fig. 5. Spatial attention module
    Residual block structure diagram of integrated CBAM
    Fig. 6. Residual block structure diagram of integrated CBAM
    Classification flow chart of Terracotta Warriors and horses fragments
    Fig. 7. Classification flow chart of Terracotta Warriors and horses fragments
    Fragment sample images
    Fig. 8. Fragment sample images
    Contrast of convergence
    Fig. 9. Contrast of convergence
    LayerOutput sizeSize of conv kernelOutput channelsStride
    Input layer256×2563
    conv1128×1287×7642
    max pool128×1283×3642
    conv2_x64×643×33×3×26464×21
    conv3_x32×323×33×3×2128128×21
    conv4_x16×163×33×3×2256256×21
    conv5_x8×83×33×3×2512512×21
    Fc_6d1×16
    Table 1. Network structure and parameters
    MethodAccuracy-Avg /%
    SIFT Feature78.42
    Shape Feature67.55
    SIFT+Shape Feature84.41
    Salient geometric Feature71.32
    ResNet18+CBAM+CutMix88.69
    Table 2. Comparison of effects of traditional classification methods and proposed classification method
    MethodAccuracy-Avg /%
    ResNet1884.76
    ResNet18+CBAM86.21
    ResNet18+CBAM+CutMix88.69
    Table 3. Comparison of effects of classification methods before and after optimization of ResNet18
    MethodAccuracy-Avg /%
    ResNet1889.25
    ResNet18+CBAM91.21
    ResNet18+CBAM+CutMix91.73
    Table 4. Comparison of classification results of ResNet18 in CIFAR-10 before and after optimization
    Yuehua Yu, Haibo Zhang, Xin Li, Jiaojiao Kou, Kang Li, Guohua Geng, Mingquan Zhou. Data Enhanced Depth Classification Model for Terracotta Warriors Fragments[J]. Laser & Optoelectronics Progress, 2022, 59(18): 1810010
    Download Citation