Optics and Precision Engineering, Volume. 31, Issue 15, 2295(2023)

Review of deep learning-based algorithms for ship target detection from remote sensing images

Zexian HUANG1,2, Fanlu WU1, Yao FU1, Yu ZHANG1, and Xiaonan JIANG1、*
Author Affiliations
  • 1Changchun Institute of Optics, Fine Mechanics and Physics, Chinese Academy of Sciences, Changchun30033, China
  • 2University of Chinese Academy of Sciences, Beijing100049, China
  • show less
    Figures & Tables(23)
    Development process of target detection algorithms
    R-CNN detection process
    SPPNet network structure
    Fast R-CNN structure
    Faster R-CNN structure
    YOLO structure
    SSD structure
    CornerNet structure
    CenterNet structure
    FSAF modules
    FCOS architecture
    Common multi-scale detection methods
    Five-parameter representation of rotating bounding box
    Circular label smoothing
    Six-parameter representation of rotating bounding box
    Common methods for improving effectiveness of small target detection
    Common methods for streamlining models
    Model compression and quantification processes
    Large-area remote sensing image segmentation detection process
    Schematic diagram of IOU
    PR curves
    • Table 1. Comparison of classical algorithm detection results

      View table
      View in Article

      Table 1. Comparison of classical algorithm detection results

      数据集

      算法

      VOC2007

      (mAP/%)

      VOC2012

      (mAP/%)

      COCO

      (AP/%)

      R-CNN58.553.3
      SPPNet59.2
      Fast R-CNN66.965.719.7
      Faster R-CNN69.967.021.9
      YOLO63.457.9
      SSD74.372.428.8
      YOLOv278.673.421.6
      YOLOv328.2
      YOLOv443.5
      YOLOv555
      CornerNet42.1
      CenterNet28.1
      FASF44.6
      FCOS44.7
    • Table 2. Comparison of ship datasets

      View table
      View in Article

      Table 2. Comparison of ship datasets

      数据集

      图像

      类型

      舰船图像数目图像尺寸分辨率/m标注方式采用数据集文献
      DOTA可见光573

      800×800~

      4 000×4 000

      水平边界框、旋转边界框

      24][38][41][48

      50][55][59][68][92

      HSRC2016可见光1 070

      300×300~

      1 500×900

      0.4~2旋转边界框

      26][39][42][48

      52][55][58][64][93

      NWPU VHR-10可见光

      533×597~

      1 728×1 028

      0.5~2水平边界框43][44][50][94
      Kaggle可见光768×768二进制掩码27][45][56][71][83
      MASATI可见光3 113512×512水平边界框72][96
      HRRSD可见光3 8860.15~1.2水平边界框55][97
      DIOR可见光2 702800×8000.5~30水平边界框42][98
      FGSC-23可见光4 0520.4~2长宽比及方向99
      SSDDSAR1 160500×5001~15水平边界框

      29][31][32

      37][39][40][50

      81][84][87][100

      SAR-ship-DatasetSAR43 819256×2563,5,8,10水平边界框37][54][101
      AIR-SARShipSAR313 000×3 0001,3水平边界框69][102
      HRSIDSAR5 604800×8000.5~3水平边界框37][53][103
    Tools

    Get Citation

    Copy Citation Text

    Zexian HUANG, Fanlu WU, Yao FU, Yu ZHANG, Xiaonan JIANG. Review of deep learning-based algorithms for ship target detection from remote sensing images[J]. Optics and Precision Engineering, 2023, 31(15): 2295

    Download Citation

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

    Category: Information Sciences

    Received: Sep. 16, 2022

    Accepted: --

    Published Online: Sep. 5, 2023

    The Author Email: JIANG Xiaonan (jxn_ciomp@qq.com)

    DOI:10.37188/OPE.20233115.2295

    Topics