Opto-Electronic Engineering, Volume. 51, Issue 7, 240099(2024)

Remote sensing image detection algorithm integrating visual center mechanism and parallel patch perception

Liming Liang... Kangquan Chen, Chengbin Wang, Yao Feng and Pengwei Long |Show fewer author(s)
Author Affiliations
  • School of Electrical Engineering and Automation, Jiangxi University of Science and Technology, Ganzhou, Jiangxi 341000, China
  • show less
    Figures & Tables(14)
    Remote sensing image detection model integrating visual center mechanism and parallel patch perception
    Explicit visual center mechanism
    Parallel multi-branch feature extraction module
    Large selective kernel module
    Multi-scale feature fusion module
    Remote sensing target detection results of different algorithms
    • Table 1. Parameter setting

      View table
      View in Article

      Table 1. Parameter setting

      参数参数值
      输入图像分辨率640×640
      初始学习率0.01
      动量参数0.937
      权重衰减系数0.0005
      训练轮次300
      批量大小16
    • Table 2. Experiments on contrasting attentional differences

      View table
      View in Article

      Table 2. Experiments on contrasting attentional differences

      注意力参数量/MFPSmAP@0.5/%
      CBAM11.39296.1
      SE11.19093.4
      CA11.19296.2
      EMA11.19395.6
      LSK11.59296.5
    • Table 3. Effectiveness of decomposing a large kernel into two sequences of depth-wise separable kernels

      View table
      View in Article

      Table 3. Effectiveness of decomposing a large kernel into two sequences of depth-wise separable kernels

      (k1, d1)(k2, d2)RFFPSmAP@0.5/%
      (3, 1)(5, 2)1110492.0
      (5, 1)(7, 3)2310595.0
      (7, 1)(9, 4)398894.7
    • Table 4. Comparison of experiments between MFFM and ELAN

      View table
      View in Article

      Table 4. Comparison of experiments between MFFM and ELAN

      模块参数量/MFPSmAP@0.5/%
      ELAN6.08894.6
      MFFM4.812694.2
    • Table 5. Ablation experimental data

      View table
      View in Article

      Table 5. Ablation experimental data

      模型准确率P/%召回率R/%平均准确率AP/%平均准确率均值 mAP@0.5/%
      飞机油桶立交桥操场
      M190.393.197.997.885.097.794.6
      M292.691.297.798.588.598.495.8
      M392.095.297.898.891.499.596.9
      M491.895.597.798.693.098.897.0
    • Table 6. Comparison of detection data from different algorithms

      View table
      View in Article

      Table 6. Comparison of detection data from different algorithms

      模型参数量/MFPS平均准确率AP/%平均准确率均值 mAP@0.5%
      飞机油桶立交桥操场
      Faster R-CNN72.01071.098.085.0100.088.5
      SSD24.44379.098.073.0100.087.5
      YOLOv3-tiny12.110494.296.476.998.591.5
      YOLOv4-tiny6.15070.797.361.799.182.4
      YOLOv5s9.19097.497.887.499.395.5
      YOLOv5m25.05697.096.889.499.295.6
      YOLOv7-tiny6.08897.997.885.097.794.6
      YOLOv8s11.19797.697.282.899.494.3
      YOLOv8m25.85397.298.184.399.594.8
      ours11.58597.798.693.098.897.0
    • Table 7. Comparison of detection results on NWPU VHR-10 dataset

      View table
      View in Article

      Table 7. Comparison of detection results on NWPU VHR-10 dataset

      模型准确率P/%召回率R/%参数量/MFPSmAP@0.5/%
      YOLOv7-tiny88.788.46.08390.7
      Ours92.587.611.57993.7
    • Table 8. Comparison of detection results on DOTA dataset

      View table
      View in Article

      Table 8. Comparison of detection results on DOTA dataset

      模型准确率P/%召回率R/%参数量/MFPSmAP@0.5/%
      YOLOv7-tiny78.270.46.08274.7
      Ours80.071.211.57776.0
    Tools

    Get Citation

    Copy Citation Text

    Liming Liang, Kangquan Chen, Chengbin Wang, Yao Feng, Pengwei Long. Remote sensing image detection algorithm integrating visual center mechanism and parallel patch perception[J]. Opto-Electronic Engineering, 2024, 51(7): 240099

    Download Citation

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

    Category: Article

    Received: May. 1, 2024

    Accepted: Jul. 10, 2024

    Published Online: Nov. 12, 2024

    The Author Email:

    DOI:10.12086/oee.2024.240099

    Topics