• Laser & Optoelectronics Progress
  • Vol. 59, Issue 16, 1628003 (2022)
Tingting Tian1、2、3 and Jun Yang1、2、3、*
Author Affiliations
  • 1Faculty of Geomatics, Lanzhou Jiaotong University, Lanzhou 730070, Gansu , China
  • 2National-Local Joint Engineering Research Center of Technologies and Applications for National Geographic State Monitoring, Lanzhou 730070, Gansu , China
  • 3Gansu Provincial Engineering Laboratory for National Geographic State Monitoring, Lanzhou 730070, Gansu , China
  • show less
    DOI: 10.3788/LOP202259.1628003 Cite this Article Set citation alerts
    Tingting Tian, Jun Yang. Object Detection For Remote Sensing Image Based on Multiscale Feature Fusion Network[J]. Laser & Optoelectronics Progress, 2022, 59(16): 1628003 Copy Citation Text show less
    Architecture of proposed object detection model
    Fig. 1. Architecture of proposed object detection model
    Comparison between convolution and dilated convolution. (a) General convolution; (b) dilated convolution
    Fig. 2. Comparison between convolution and dilated convolution. (a) General convolution; (b) dilated convolution
    Feature fusion network. (a) FPN; (B) PANet
    Fig. 3. Feature fusion network. (a) FPN; (B) PANet
    Multi-dimensional attention module
    Fig. 4. Multi-dimensional attention module
    Detection results on RSOD dataset. (a) Aircraft; (b) oiltank; (c) overpass; (d) playground
    Fig. 5. Detection results on RSOD dataset. (a) Aircraft; (b) oiltank; (c) overpass; (d) playground
    Detection results on DOTA dataset. (a) PL; (b) GTF and SBF; (c) BR; (d) BD; (e) LV and SV; (f) RA; (g) HA; (h) ST; (i) SH; (j) BC and TC; (k) SP; (l) HC
    Fig. 6. Detection results on DOTA dataset. (a) PL; (b) GTF and SBF; (c) BR; (d) BD; (e) LV and SV; (f) RA; (g) HA; (h) ST; (i) SH; (j) BC and TC; (k) SP; (l) HC
    Examples of miss detection and error detection. (a) Miss detection; (b) error detection
    Fig. 7. Examples of miss detection and error detection. (a) Miss detection; (b) error detection
    DatasetAircraftOiltankOverpassPlaygroundmAP
    RSOD81.0290.7710010092.95
    Table 1. Detection results on RSOD dataset
    DatasetPLBDBRGTFSVLVSHTCBCSTSBFRAHASPHCmAP
    DOTA89.6884.3952.1272.7164.4967.0777.4590.1183.9886.0164.0363.3374.4567.7462.8773.39
    Table 2. Detection results on DOTA dataset
    TypeAircraftOiltankOverpassPlaygroundmAP
    FPN78.8690.4899.73100.0092.26
    FPN++81.0290.77100.00100.0092.95
    Table 3. Experimental comparison of different FPN on RSOD dataset
    TypePLBDBRGTFSVLVSHTCBCSTSBFRAHASPHCmAP
    FPN88.8184.3350.9872.4562.7165.3575.3489.9681.1684.0753.2663.3372.2466.2960.1171.36
    FPN++89.6884.3952.1272.7164.4967.0777.4590.1183.9886.0155.0363.3374.4567.7462.8772.86
    Table 4. Experimental comparison of different FPN on DOTA dataset
    AlgorithmaircraftoiltankoverpassplaygroundmAP
    R-FCN71.4890.2376.8497.7084.07
    Deformable R-FCN71.5090.2681.4899.5385.70
    Faster R-CNN71.9090.90100.00100.0090.70
    RFN79.1090.50100.0099.7092.30
    MDCF2Det81.0290.77100.00100.0092.95
    Table 5. Accuracy comparison of different algorithms on RSOD dataset
    AlgorithmPLBDBRGTFSVLVSHTCBCSTSBFRAHASPHCmAP
    YOLO v276.9033.8722.7334.8838.7332.0252.3761.6548.5433.9129.2736.8336.4438.2611.6139.20
    R-FCN79.3344.2636.5853.5339.3834.1547.2945.6647.7465.8437.9244.2350.6450.6434.9047.24
    CenterFPANet88.7471.5248.9552.0648.5573.3761.1490.5357.6384.0666.6462.7173.3357.6342.7665.29
    FPN88.7075.1052.6059.2069.4078.8084.5090.6081.3082.6052.5062.1076.6066.3060.1072.00
    ICN90.0077.7053.4073.3073.5065.0078.2090.8079.1084.8057.2062.1173.4570.2258.0872.45
    FMSSD89.1181.5148.2267.9469.2373.5676.8790.7182.6773.3352.6567.5272.3780.5760.1572.43
    MDCF2Det89.6884.3952.1272.7164.4967.0777.4590.1183.9886.0155.0363.3374.4567.7462.8772.86
    Table 6. Accuracy comparison of different algorithms on DOTA dataset
    Tingting Tian, Jun Yang. Object Detection For Remote Sensing Image Based on Multiscale Feature Fusion Network[J]. Laser & Optoelectronics Progress, 2022, 59(16): 1628003
    Download Citation