• Laser & Optoelectronics Progress
  • Vol. 56, Issue 4, 041501 (2019)
Jie Zhang1,2, Hongdong Zhao1,2, Yuhai Li2,*, Miao Yan1, and Zetong Zhao1
Author Affiliations
  • 1 School of Electronic and Information Engineering, Hebei University of Technology, Tianjin 300401, China
  • 2 Science and Technology Electro-Optical Information Security Control Laboratory, Tianjin 300308, China
  • show less
    DOI: 10.3788/LOP56.041501 Cite this Article Set citation alerts
    Jie Zhang, Hongdong Zhao, Yuhai Li, Miao Yan, Zetong Zhao. Classifier for Recognition of Fine-Grained Vehicle Models under Complex Background[J]. Laser & Optoelectronics Progress, 2019, 56(4): 041501 Copy Citation Text show less
    Flow chart of Softmax-SVM classifier
    Fig. 1. Flow chart of Softmax-SVM classifier
    Structural diagram of Softmax-SVM classifier based on DCNN
    Fig. 2. Structural diagram of Softmax-SVM classifier based on DCNN
    Partial samples of 27 types of vehicle models
    Fig. 3. Partial samples of 27 types of vehicle models
    DCNN performance versus training under different loss functions. (a) Training accuracy; (b) test accuracy; (c) training loss; (d) test loss
    Fig. 4. DCNN performance versus training under different loss functions. (a) Training accuracy; (b) test accuracy; (c) training loss; (d) test loss
    Loss functionAccuracy /%
    Training setTest set
    Square-loss80.076.58
    Cross-entropy-loss99.895.51
    Exp-loss56.057.28
    Table 1. Recognition accuracies of DCNN training for 350 times under different loss functions
    Loss functionLoss/arb. units
    Training setTest set
    Square-loss0.37290.3886
    Cross-entropy-loss0.08100.2144
    Exp-loss0.63540.6448
    Table 2. Losses of DCNN training for 350 times under different loss functions
    Feature extractionClassifierAccuracy /%
    HOGSVM87.59
    SURFBag for word46.00
    DCNNSoftmax95.51
    DCNNSoftmax-SVM97.78
    Table 3. Test accuracies of different classifier models
    FeatureextractionClassifierTime /s
    Training setTest set
    HOGSVM195.2122.127
    SURFBag279.5348.869
    DCNNSoftmax312.3862.441
    DCNNSoftmax-SVM239.6900.759
    Table 4. Time for training and recognition of all test samples by different classifiers
    LayerAccuracy /%LossTime /s
    FC297.031.00240.4609
    ReLU697.401.00200.7639
    FC197.781.00180.7591
    ReLU595.541.00274.5576
    Table 5. Softmax-SVM performance in extracting features of different layers of DCNN to train SVM
    Jie Zhang, Hongdong Zhao, Yuhai Li, Miao Yan, Zetong Zhao. Classifier for Recognition of Fine-Grained Vehicle Models under Complex Background[J]. Laser & Optoelectronics Progress, 2019, 56(4): 041501
    Download Citation