• Acta Optica Sinica
  • Vol. 40, Issue 3, 0310001 (2020)
Zhehan Zhang1、2, Wei Fang1、*, Lili Du1, Yanli Qiao1, Dongying Zhang1, and Guoshen Ding1、2
Author Affiliations
  • 1Key Laboratory of Optical Calibration and Characterization, Anhui Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Hefei, Anhui 230031, China
  • 2University of Science and Technology of China, Hefei, Anhui 230026, China
  • show less
    DOI: 10.3788/AOS202040.0310001 Cite this Article Set citation alerts
    Zhehan Zhang, Wei Fang, Lili Du, Yanli Qiao, Dongying Zhang, Guoshen Ding. Semantic Segmentation of Remote Sensing Image Based on Encoder-Decoder Convolutional Neural Network[J]. Acta Optica Sinica, 2020, 40(3): 0310001 Copy Citation Text show less
    Architecture comparison. (a) U-Net; (b) SegNet; (c) SegProNet
    Fig. 1. Architecture comparison. (a) U-Net; (b) SegNet; (c) SegProNet
    Remote sensing image semantic segmentation network structure of SegProNet
    Fig. 2. Remote sensing image semantic segmentation network structure of SegProNet
    Maxpooling indices and upsampling
    Fig. 3. Maxpooling indices and upsampling
    Display of data sets. (a) Training set image; (b) corresponding label visualization image
    Fig. 4. Display of data sets. (a) Training set image; (b) corresponding label visualization image
    Some training set images
    Fig. 5. Some training set images
    Comparison of experimental results. (a) Original image; (b) original label visualization result; (c) U-Net segmentation result; (d) SegNet segmentation result; (e) segmentation result of SegProNet+ReLU; (f) segmentation result of SegProNet+ELU
    Fig. 6. Comparison of experimental results. (a) Original image; (b) original label visualization result; (c) U-Net segmentation result; (d) SegNet segmentation result; (e) segmentation result of SegProNet+ReLU; (f) segmentation result of SegProNet+ELU
    Comparison of experimental details. (a) Original label visualization result; (b) U-Net segmentation result; (c) SegNet segmentation result; (d) segmentation result of SegProNet+ReLU; (e) segmentation result of SegProNet+ELU
    Fig. 7. Comparison of experimental details. (a) Original label visualization result; (b) U-Net segmentation result; (c) SegNet segmentation result; (d) segmentation result of SegProNet+ReLU; (e) segmentation result of SegProNet+ELU
    Each network loss and accuracy curves. (a) U-Net; (b) SegNet; (c) SegProNet; (d) SegProNet+ELU
    Fig. 8. Each network loss and accuracy curves. (a) U-Net; (b) SegNet; (c) SegProNet; (d) SegProNet+ELU
    Feature categoryNo.Label color (R,G,B)
    Background0(0,0,0)
    Vegetation1(50,205,50)
    Building2(245, 254, 0)
    Water3(0, 255, 255)
    Road4(255, 92, 75)
    Table 1. Information of labels
    Network categoryTraining time /hPrediction time /h
    U-Net16.30.26
    SegNet27.60.61
    SegProNet23.20.52
    SegProNet+ELU22.60.51
    Table 2. Comparison of training and prediction time of each network
    CategoryEvaluationVegetationBuildingWaterRoad
    U-NetPrecision0.75650.52320.73680.6904
    Recall0.63410.78470.82370.7571
    IoU0.52660.45750.63640.5652
    SegNetPrecision0.83400.72540.85550.7576
    Recall0.81640.83380.85410.8722
    IoU0.70230.63390.74640.6819
    SegProNetPrecision0.85330.75080.88370.7815
    Recall0.82390.85870.86530.8736
    IoU0.72160.66820.77680.7021
    SegProNet+ELUPrecision0.85310.88610.87920.8664
    Recall0.85340.75280.89140.8164
    IoU0.74410.68630.79420.7251
    Table 3. Evaluation indicators of each method
    Zhehan Zhang, Wei Fang, Lili Du, Yanli Qiao, Dongying Zhang, Guoshen Ding. Semantic Segmentation of Remote Sensing Image Based on Encoder-Decoder Convolutional Neural Network[J]. Acta Optica Sinica, 2020, 40(3): 0310001
    Download Citation