• Laser & Optoelectronics Progress
  • Vol. 56, Issue 5, 051002 (2019)
Pan Ou, Zheng Zhang*, Kui Lu, and Zeyang Liu
Author Affiliations
  • School of Instrumentation Science and Opto-Electronic Engineering, Beihang University, Beijing 100191, China
  • show less
    DOI: 10.3788/LOP56.051002 Cite this Article Set citation alerts
    Pan Ou, Zheng Zhang, Kui Lu, Zeyang Liu. Object Detectionin of Remote Sensing Images Based on Convolutional Neural Networks[J]. Laser & Optoelectronics Progress, 2019, 56(5): 051002 Copy Citation Text show less
    Schematic of rotation transformation. (a) Rotation principle; (b) target rotation
    Fig. 1. Schematic of rotation transformation. (a) Rotation principle; (b) target rotation
    Structural diagram of spatial transformation network
    Fig. 2. Structural diagram of spatial transformation network
    Schematic of improved method
    Fig. 3. Schematic of improved method
    Statistical figure of angle distribution. (a) Histogram of angle distribution; (b) targets with different orientations
    Fig. 4. Statistical figure of angle distribution. (a) Histogram of angle distribution; (b) targets with different orientations
    Samples from training dataset
    Fig. 5. Samples from training dataset
    Negative samples
    Fig. 6. Negative samples
    Average precision versus sample number
    Fig. 7. Average precision versus sample number
    Detection results of plane
    Fig. 8. Detection results of plane
    Detection results of car
    Fig. 9. Detection results of car
    Detection methodDetection timeof image /sAP /%
    HOG+SVM1.0364.61
    Faster R-CNN0.3196.58
    Faster R-CNN+STN0.3197.31
    Table 1. Performance comparison among different detection methods
    Detection methodAP of planeAP of carAP of ship
    HOG+SVM64.6161.4060.97
    Faster R-CNN96.5888.3584.87
    Faster R-CNN+STN97.3189.7184.76
    Table 2. Comparison of detection results for different classes%
    Pan Ou, Zheng Zhang, Kui Lu, Zeyang Liu. Object Detectionin of Remote Sensing Images Based on Convolutional Neural Networks[J]. Laser & Optoelectronics Progress, 2019, 56(5): 051002
    Download Citation