Spectroscopy and Spectral Analysis, Volume. 42, Issue 10, 3275(2022)

Improved YOLOv4 Remote Sensing Image Detection Method of Ground Objects Along Railway

Yang-ping WANG*, Shu-mei HAN1; *;, Jing-yu YANG1; 2;, Jian-wu DANG1; 2;, and Zhan-ping ZHANG1;
Author Affiliations
  • 1. School of Electronic and Information Engineering, Lanzhou Jiaotong University, Lanzhou 730070, China
  • show less
    Figures & Tables(9)
    Improved DenseNet network model
    Improved DenseNet for YOLOv4 feature extraction network
    SE-CSP module proposed in this article
    ICBAM module
    Improved YOLOv4 network model
    Test results of ablation experiments (a): Original remote sensing images; (b): Detection using YOLOv4; (c): Improved DenseNet detection for original YOLOv4 feature extraction; (d): Improved DenseNet and SE modules for original YOLOv4 detection; (e): Improved DenseNet, SE and CBAM modules for original YOLOv4 detection
    Experimental results of different target detection algorithms and improved algorithms(a): Original remote sensing images; (b): Detection using YOLOv3 algorithm; (c): Detection using YOLOv3-UAV[10];(d): Detection using YOLOv3-ship[11]; (e): Detection using original YOLOv4; (f): Detection using the algorithm proposed in this paper
    • Table 1. Analysis of ablation experiments

      View table
      View in Article

      Table 1. Analysis of ablation experiments

      MethodPrecision/%Recall/%mAP/%Model Size/MBTimes/s
      YOLOv480.6679.0281.65250.430.058
      DenseNet81.3580.5382.37233.180.049
      DenseNet and SE84.2783.1283.76231.650.052
      DenseNet, SE and ICBAM83.5982.8183.36229.060.052
    • Table 2. Comparison of different target detection algorithms

      View table
      View in Article

      Table 2. Comparison of different target detection algorithms

      MethodPrecision
      /%
      Recall
      /%
      mAP
      /%
      F1Model
      Size
      MB
      YOLOv368.5267.6078.5268.06247
      YOLOv3-UAV76.3773.2080.3874.75247.36
      YOLOv3-Ship77.6276.8080.8677.21246.27
      YOLOv480.6679.0281.6579.83250.43
      本算法83.5982.8183.7683.20229.06
    Tools

    Get Citation

    Copy Citation Text

    Yang-ping WANG, Shu-mei HAN, Jing-yu YANG, Jian-wu DANG, Zhan-ping ZHANG. Improved YOLOv4 Remote Sensing Image Detection Method of Ground Objects Along Railway[J]. Spectroscopy and Spectral Analysis, 2022, 42(10): 3275

    Download Citation

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

    Category: Research Articles

    Received: Aug. 23, 2021

    Accepted: Mar. 1, 2022

    Published Online: Nov. 23, 2022

    The Author Email: Yang-ping WANG (1328396793@qq.com)

    DOI:10.3964/j.issn.1000-0593(2022)10-3275-08

    Topics