Chinese Journal of Liquid Crystals and Displays, Volume. 39, Issue 1, 111(2024)

Improved YOLOv5 lightweight binocular vision UAV obstacle avoidance algorithm based on Ghost module

Yifan JIA1,2, Tianyi CAO3, and Yue BAI1、*
Author Affiliations
  • 1Changchun Institute of Optics,Fine Mechanics and Physics,Chinese Academy of Sciences,Changchun 130033,China
  • 2University of Chinese Academy of Sciences,Beijing 100049,China
  • 3SWJTU-Leeds Joint School,Chengdu 610097,China
  • show less
    Figures & Tables(16)
    Ghost module
    Ghost-BottleNeck
    Network structure of YOLOv5s-Ghost
    Feature point extraction
    Extraction results of ORB feature point
    Feature point matching
    Feature point extraction and matching constrained by object detection
    Obstacle avoidance solution for UAVs
    Training loss curve for YOLOv5s-Ghost network
    Mean average precision(mAP)on the test set
    Calibration images
    UAV and its components
    Testing of the UAV obstacle avoidance system
    • Table 1. Comparison of performance indicators for different detection algorithms

      View table
      View in Article

      Table 1. Comparison of performance indicators for different detection algorithms

      模型mAP/%平均检测时间/ms参数量/M浮点运算量/G
      YOLOv5s77.250.97.216.5
      YOLOv5s-Ghost76.880.75.411.5
    • Table 2. Intrinsic parameters of left and right camera

      View table
      View in Article

      Table 2. Intrinsic parameters of left and right camera

      相机焦距光心位置坐标畸变系数量偏斜系数
      左相机843.886 64(676.817 09,522.237 67)(0.043 37,-0.054 12,0.002 05,-0.001 36,0.000 00)0.000 00
      右相机843.189 91(667.508 83,556.936 52)(-0.002 81,0.007 27,-0.005 64,0.001 13,0.000 00)0.000 00
    • Table 3. Distance measurement results

      View table
      View in Article

      Table 3. Distance measurement results

      编号a/mb/me/%
      11.351.414.26
      21.651.565.77
      32.482.412.90
      42.903.054.92
      55.645.531.99
      68.358.033.99
      711.6712.284.97
      812.8311.888.36
      918.6417.426.70
      1020.1522.5310.56
      1123.6221.0212.37
      1225.4722.4513.40
    Tools

    Get Citation

    Copy Citation Text

    Yifan JIA, Tianyi CAO, Yue BAI. Improved YOLOv5 lightweight binocular vision UAV obstacle avoidance algorithm based on Ghost module[J]. Chinese Journal of Liquid Crystals and Displays, 2024, 39(1): 111

    Download Citation

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

    Category: Research Articles

    Received: Feb. 21, 2023

    Accepted: --

    Published Online: Mar. 27, 2024

    The Author Email: Yue BAI (baiy@ciomp.?ac.cn)

    DOI:10.37188/CJLCD.2023-0069

    Topics