Chinese Journal of Liquid Crystals and Displays, Volume. 38, Issue 7, 955(2023)

Improved lightweight human pose estimation algorithm

Ming-he WANG1, Wang-ming XU1,2、*, and Hao-kun JIANG1
Author Affiliations
  • 1School of Information Science and Engineering,Wuhan University of Science and Technology,Wuhan 430081,China
  • 2Engineering Research Center for Metallurgical Automation and Detecting Technology of Ministry of Education,Wuhan University of Science and Technology,Wuhan 430081,China
  • show less
    Figures & Tables(7)
    Various human postures
    Network structure of LCSA-Net
    Network structure of LCSA-YOLO-Pose
    Effect comparison of the original YOLO-Pose algorithm and the proposed algorithm for human pose estimation
    • Table 1. Performance comparison of lightweight human object detection algorithms

      View table
      View in Article

      Table 1. Performance comparison of lightweight human object detection algorithms

      ModelModel Size/MFLOPs/GInput SizemAP/%AP50/%Speed/ms
      YOLOv3-tiny8.6612.9640×64043.477.69.3
      CBAM-YOLOv38.6812.9640×64043.878.79.3
      LCSA-YOLOv38.6812.9640×64051.883.69.4
      YOLOv5s7.0115.8640×64057.784.816.2
      CBAM-YOLOv5s7.0315.8640×64064.289.416.3
      LCSA-YOLOv5s7.0215.8640×64064.989.016.3
    • Table 2. Performance comparison of LCSA-Net used for HigherHRNet model

      View table
      View in Article

      Table 2. Performance comparison of LCSA-Net used for HigherHRNet model

      ModelModel size/MInput SizemAP/%AP50/%AP75/%APM/%APL/%
      HigherHRNet28.6512×51266.686.272.661.175.2
      LCSA-HrHRNet28.6512×51267.086.372.861.775.5
    • Table 3. Performance comparison of YOLO-Pose models under different improved strategies

      View table
      View in Article

      Table 3. Performance comparison of YOLO-Pose models under different improved strategies

      LCSA-NetDAWModel size/MmAP/%AP50/%AR/%
      12.5450.880.357.9
      12.5652.580.859.4
      12.5451.781.158.7
      12.5652.881.859.6
    Tools

    Get Citation

    Copy Citation Text

    Ming-he WANG, Wang-ming XU, Hao-kun JIANG. Improved lightweight human pose estimation algorithm[J]. Chinese Journal of Liquid Crystals and Displays, 2023, 38(7): 955

    Download Citation

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

    Category: Research Articles

    Received: Oct. 5, 2022

    Accepted: --

    Published Online: Jul. 31, 2023

    The Author Email: Wang-ming XU (xuwangming@wust.edu.cn)

    DOI:10.37188/CJLCD.2022-0323

    Topics