Journals Articles Conferences News About CLP
Submit a paper Email Alert
Search
Search
Articles
Journals
News
Advanced Search
Top Searches
2D MaterialsTransformation opticsQuantum PhotonicsSmall lasersSemiconductor UV PhotonicsSupersymmetric microring laser arrays

Laser & Optoelectronics Progress, Volume. 58, Issue 4, 0415009(2021)

Classification Method of Crossing Vehicle Based on Improved Residual Network

Yuxin Li, Fan Yang*, Zhao Liu, and Yazhong Si
Author Affiliations
  • School of Electronic and Information Engineering, Hebei University of Technology, Tianjin 300401, China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (12)
    Equations (0)
    References (19)
    Cited By (1)
    Tools
    Paper Information
    Topics
    Figures & Tables(12)
    Flowchart of the proposed method

    Fig. 1. Flowchart of the proposed method

    Download full size
    Improve comparison. (a) Original residual block; (b) improved residual block

    Fig. 2. Improve comparison. (a) Original residual block; (b) improved residual block

    Download full size
    Normal convolution

    Fig. 3. Normal convolution

    Download full size
    Group convolution

    Fig. 4. Group convolution

    Download full size
    Attention model

    Fig. 5. Attention model

    Download full size
    Heat maps processed by different models. (a) Original map; (b) original model ResNet; (c) model with attention mechanism

    Fig. 6. Heat maps processed by different models. (a) Original map; (b) original model ResNet; (c) model with attention mechanism

    Download full size
    Partial images in Stanford Cars dataset

    Fig. 7. Partial images in Stanford Cars dataset

    Download full size
    Partial images in real crossing dataset

    Fig. 8. Partial images in real crossing dataset

    Download full size
    Accuracy of ablation experiment

    Fig. 9. Accuracy of ablation experiment

    Download full size
    Loss of ablation experiment

    Fig. 10. Loss of ablation experiment

    Download full size

    • Table 1. Accuracy of different models on Stanford Cars dataset

      View table

      Table 1. Accuracy of different models on Stanford Cars dataset

      ModelAccuracy /%
      Three-scale Attention[17]81.50
      B-CNN[18]86.50
      Kernel-Pooling[19]85.70
      FA-ResNet86.97

    • Table 2. Results of ablation experiment

      View table

      Table 2. Results of ablation experiment

      ExperimentNo.Group convolutionAttention modelFocal lossAccuracy /%
      180.44
      2√81.12
      3√88.43
      4√90.15
      5√√88.91
      6√√92.13
      7√√94.19
      8√√√94.96
    Tools

    Get Citation

    Copy Citation Text

    Yuxin Li, Fan Yang, Zhao Liu, Yazhong Si. Classification Method of Crossing Vehicle Based on Improved Residual Network[J]. Laser & Optoelectronics Progress, 2021, 58(4): 0415009

    Download Citation

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

    Category: Machine Vision

    Received: Sep. 2, 2020

    Accepted: Nov. 5, 2020

    Published Online: Feb. 22, 2021

    The Author Email: Fan Yang (commanderjy@163.com)

    DOI:10.3788/LOP202158.0415009

    Topics

    laser devices and laser physics
    Lasers and Laser Optics
    Laser physics
    laser manufacturing
    Instrumentation, Measurement and Metrology