• High Power Laser and Particle Beams
  • Vol. 34, Issue 11, 112002 (2022)
Yanhua Shao, Yupei Feng, Xiaoqiang Zhang, and Hongyu Chu*
Author Affiliations
  • School of Information Engineering, Southwest University of Science and Technology, Mianyang 621010, China
  • show less
    DOI: 10.11884/HPLPB202234.220023 Cite this Article
    Yanhua Shao, Yupei Feng, Xiaoqiang Zhang, Hongyu Chu. Using deep learning for surface defects identification of optical components[J]. High Power Laser and Particle Beams, 2022, 34(11): 112002 Copy Citation Text show less
    LeNet-5 network structure
    Fig. 1. LeNet-5 network structure
    [in Chinese]
    Fig. 1. [in Chinese]
    Three-channel input
    Fig. 2. Three-channel input
    [in Chinese]
    Fig. 2. [in Chinese]
    ICFNet basic structure
    Fig. 3. ICFNet basic structure
    [in Chinese]
    Fig. 3. [in Chinese]
    Some sample images from ICF-90 dataset
    Fig. 4. Some sample images from ICF-90 dataset
    Training results with different number of convolution layers
    Fig. 5. Training results with different number of convolution layers
    input channelsclassifieraccuracy/%
    SVM[2]92.2
    1SVM (Linear) SVM (RBF) LeNet-5 76.6 60.0 73.3
    ICFNet90.0
    3SVM (Linear)76.6
    SVM (RBF)63.3
    LeNet-586.7
    ICFNet96.7(+4.5)
    Table 1. Comparison of classification accuracy of different methods
    Yanhua Shao, Yupei Feng, Xiaoqiang Zhang, Hongyu Chu. Using deep learning for surface defects identification of optical components[J]. High Power Laser and Particle Beams, 2022, 34(11): 112002
    Download Citation