Laser & Optoelectronics Progress, Volume. 61, Issue 24, 2437006(2024)

Lightweight Underwater Target Detection Algorithm Based on Improved YOLOv8n

Guobo Xie, Lihui Liang, Zhiyi Lin*, Songze Lin, and Qing Su
Author Affiliations
  • School of Computer Science and Technology, Guangdong University of Technology, Guangzhou 510006, Guangdong , China
  • show less
    Figures & Tables(13)
    Structure of YOLOv8n algorithm
    Structure of improved YOLOv8n algorithm
    Structure of GAM
    Schematic diagram of SPDConv when scale is 2
    Structure of SPD_C2f_GAM
    Structural of CAREFE
    Structure of PR-DyHead
    ReLU and PReLU function images. (a) ReLU function; (b) PReLU function
    Statistic on number of labels in each class on the dataset
    Detection effect comparison of different algorithms. (a) Original images; (b) detection results by YOLOv8n algorithm; (c) detection results by improved YOLOv8n algorithm
    • Table 1. Ablation experimental results of YOLOv8n

      View table

      Table 1. Ablation experimental results of YOLOv8n

      Experiment groupMethodP /%R /%PmAP /%Parameters /MGFLOPs
      1YOLOv8n83.7178.2083.423.018.1
      2YOLOv8n +SPD_C2f_GAM84.2579.5184.755.0510.7
      3YOLOv8n +SPD_C2f_GAM+CARAFE84.5779.9185.365.1811.0
      4YOLOv8n +SPD_C2f_GAM+CARAFE+NWD-CIoU84.7380.5986.015.1811.0
      5YOLOv8n +SPD_C2f_GAM+CARAFE+NWD-CIoU+PR-DyHead85.0681.4386.625.6712.5
    • Table 2. Ablation experimental results of individual class

      View table

      Table 2. Ablation experimental results of individual class

      ClassP /%R /%PmAP /%
      YOLOv8nImproved YOLOv8nYOLOv8nImproved YOLOv8nYOLOv8nImproved YOLOv8n
      holothurian84.885.568.075.579.683.4
      echinus88.888.683.784.590.890.1
      scallop82.685.667.772.376.382.6
      starfish85.086.583.583.788.789.9
      fish74.179.263.972.866.979.9
      corals72.875.866.969.570.173.6
      diver87.491.189.890.694.094.2
      cuttlefish95.395.093.894.197.697.4
      turtle93.495.092.795.194.897.3
      jellyfish72.968.372.076.275.477.8
    • Table 3. Performance comparison of different models

      View table

      Table 3. Performance comparison of different models

      ModelP /%R /%PmAP /%Parameters /MGFLOPs
      Faster R-CNN80.2370.1278.6540.45371.0
      YOLOv3-tiny78.9669.4576.288.6913.0
      YOLOv5s84.2578.7584.577.0816.6
      RepViTS-YOLOX83.8277.5684.3112.4028.7
      YOLOv7-tiny81.8576.6382.326.0313.3
      YOLOv784.9181.2686.2537.24105.3
      YOLOv8n83.7178.2083.423.018.1
      YOLOv8s84.6878.2584.7711.1228.5
      IEMAyoloViT82.9271.9179.514.029.7
      Improved YOLOv8n85.0681.4386.625.6712.5
    Tools

    Get Citation

    Copy Citation Text

    Guobo Xie, Lihui Liang, Zhiyi Lin, Songze Lin, Qing Su. Lightweight Underwater Target Detection Algorithm Based on Improved YOLOv8n[J]. Laser & Optoelectronics Progress, 2024, 61(24): 2437006

    Download Citation

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

    Category: Digital Image Processing

    Received: Mar. 25, 2024

    Accepted: May. 20, 2024

    Published Online: Dec. 11, 2024

    The Author Email: Zhiyi Lin (lzy291@gdut.edu.cn)

    DOI:10.3788/LOP240955

    CSTR:32186.14.LOP240955

    Topics