• Laser & Optoelectronics Progress
  • Vol. 58, Issue 8, 0810022 (2021)
Yuchen Jiang* and Bin Zhu
Author Affiliations
  • State Key Laboratory of Pulsed Power Laser Technology, College of Electronic Countermeasures, National University of Defense Technology, Hefei, Anhui 230009, China
  • show less
    DOI: 10.3788/LOP202158.0810022 Cite this Article Set citation alerts
    Yuchen Jiang, Bin Zhu. Data Augmentation for Remote Sensing Image Based on Generative Adversarial Networks Under Condition of Few Samples[J]. Laser & Optoelectronics Progress, 2021, 58(8): 0810022 Copy Citation Text show less
    Basic structural frame of GAN
    Fig. 1. Basic structural frame of GAN
    Data augmentation process of remote sensing image
    Fig. 2. Data augmentation process of remote sensing image
    Structural diagram of discrimination model
    Fig. 3. Structural diagram of discrimination model
    Examples of generated samples. (a)(b) Chimney; (c)(d) basketball court; (e)(f) stadium
    Fig. 4. Examples of generated samples. (a)(b) Chimney; (c)(d) basketball court; (e)(f) stadium
    CategoryAirportBasketball courtBridgeChimneyOverpassStadium
    StyleGAN273.8070.38102.69125.3783.9856.17
    Ours67.3552.7095.9678.5155.3853.25
    CategoryDamExpressway-service-areaGolf fieldGround track fieldTrain station
    StyleGAN2123.0462.4267.0345.20113.78
    Ours103.0054.2257.9839.8793.00
    Table 1. Comparison of FID values of generated target on DIOR datasets
    CategoryOverpassPlayground
    StyleGAN288.52127.86
    Ours110.42168.30
    Table 2. Comparison of FID values of generated target on RSOD datasets
    MethodPercent of added data /Accuracy (mAP)
    Affine transformation0/46.06100/47.69200/48.45300/47.74400/49.14
    Ours0/46.065/48.1410/46.9715/45.6020/46.17
    Combined method0/46.06105/48.09205/49.48305/47.96405/49.51
    Table 3. Comparison of detection accuracy for adding different percent of enhancement data on DIOR dataset%
    MethodPercent of added data /Accuracy (mAP)
    Affine transformation0/70.78100/74.22200/79.28300/76.76400/77.84
    Ours0/70.785/75.3710/74.8315/76.5220/77.25
    Combined method0/70.78120/79.34220/80.13320/77.77420/78.35
    Table 4. Comparison of detection accuracy for adding different percent of enhancement data on RSOD dataset%
    Yuchen Jiang, Bin Zhu. Data Augmentation for Remote Sensing Image Based on Generative Adversarial Networks Under Condition of Few Samples[J]. Laser & Optoelectronics Progress, 2021, 58(8): 0810022
    Download Citation