• Laser & Optoelectronics Progress
  • Vol. 59, Issue 16, 1628008 (2022)
Qikai Zhou1, Wei Zhang1, Dongjin Li2, and Fu Niu1、*
Author Affiliations
  • 1Academy of Systems Engineering of Academy of Military Science of Chinese PLA, Beijing 100071, China
  • 2Beijing Institute of Control and Electronic Technology, Beijing 100038, China
  • show less
    DOI: 10.3788/LOP202259.1628008 Cite this Article Set citation alerts
    Qikai Zhou, Wei Zhang, Dongjin Li, Fu Niu. Ship Classification and Detection Method for Optical Remote Sensing Images Based on Improved YOLOv5s[J]. Laser & Optoelectronics Progress, 2022, 59(16): 1628008 Copy Citation Text show less
    EB-YOLOv5s structure
    Fig. 1. EB-YOLOv5s structure
    Efficient channel attention module
    Fig. 2. Efficient channel attention module
    Feature pyramid network (FPN).(a) Traditional feature pyramid network; (b) path aggregation network; (c) multi-stage weighted fusion pyramid network
    Fig. 3. Feature pyramid network (FPN).(a) Traditional feature pyramid network; (b) path aggregation network; (c) multi-stage weighted fusion pyramid network
    Samples in dataset
    Fig. 4. Samples in dataset
    Detection effect of different methods
    Fig. 5. Detection effect of different methods
    KmAPAP1(Aircraft_carrier)AP2(Warship)AP3(Civilian_ship)AP4(Submarine)
    185.895.095.868.384.0
    387.499.294.469.186.8
    584.494.695.166.481.4
    Table 1. Experimental results under different K
    Experiment No.Number of modulesmAPAP1(Aircraft_carrier)AP2(Warship)AP3(Civilian_ship)AP4(Submarine)
    1286.192.895.870.984.7
    2387.395.295.759.381.5
    34(Neck)81.890.094.264.878.0
    44(Backbone)87.499.294.469.186.8
    5587.498.295.968.886.6
    Table 2. Experimental results under different number and location of modules
    MethodmAP /%t /sWeight /MB
    YOLOv384.60.2870235
    SSD89.71.62292.1
    YOLOv5s83.90.186113.7
    EB-YOLOv5s89.20.207062.7
    Table 3. Performance comparison of different detection methods
    MethodmAPAP1(Aircraft_carrier)AP2(Warship)AP3(Civilian_ship)AP4(Submarine)
    YOLOv5s83.997.394.962.880.8
    YOLOv5s+SE85.997.896.065.384.7
    YOLOv5s+ECA87.499.294.469.186.8
    YOLOv5s+BIFPN86.594.094.366.990.8
    EB-YOLOv5s89.298.995.871.690.3
    Table 4. Comparison for average accuracy
    Methodt /sWeight /MBLayerParameters
    YOLOv5s0.186113.72247062001
    YOLOv5s+SE0.221614.12567237281
    YOLOv5s+ECA0.199813.72407062013
    YOLOv5s+BIFPN0.199615.72368128517
    EB-YOLOv5s0.206762.72528128529
    Table 5. Comparison for detection speed and network complexity
    Qikai Zhou, Wei Zhang, Dongjin Li, Fu Niu. Ship Classification and Detection Method for Optical Remote Sensing Images Based on Improved YOLOv5s[J]. Laser & Optoelectronics Progress, 2022, 59(16): 1628008
    Download Citation