• Laser & Optoelectronics Progress
  • Vol. 58, Issue 8, 0810011 (2021)
Chunyan Yu, Yan Xu*, Lisha Gou, and Zhefeng Nan
Author Affiliations
  • School of Electronic and Information Engineering, Lanzhou Jiaotong University, Lanzhou, Gansu 730070, China
  • show less
    DOI: 10.3788/LOP202158.0810011 Cite this Article Set citation alerts
    Chunyan Yu, Yan Xu, Lisha Gou, Zhefeng Nan. Crowd Counting Based on Single-Column Deep Spatiotemporal Convolutional Neural Network[J]. Laser & Optoelectronics Progress, 2021, 58(8): 0810011 Copy Citation Text show less
    CNN based on single-column deep spatiotemporal counting
    Fig. 1. CNN based on single-column deep spatiotemporal counting
    Structure of the improved FCN
    Fig. 2. Structure of the improved FCN
    Structure of the ST counting network
    Fig. 3. Structure of the ST counting network
    Counting results of our model on the UCSD data set
    Fig. 4. Counting results of our model on the UCSD data set
    Experimental results of our model on the Mall data set. (a) Density map; (b) counting result
    Fig. 5. Experimental results of our model on the Mall data set. (a) Density map; (b) counting result
    Experimental results of our model on the self-built data set. (a) Density map; (b) counting result
    Fig. 6. Experimental results of our model on the self-built data set. (a) Density map; (b) counting result
    Accuracies of different models
    Fig. 7. Accuracies of different models
    Training loss curves of the network before and after the improvement
    Fig. 8. Training loss curves of the network before and after the improvement
    ModelsfMAEfMSE
    ConvLSTM[3]1.301.79
    Bidirectional ConvLSTM[3]1.131.43
    Ref. [19]2.247.97
    Ref. [20]2.257.82
    Ref. [10]1.543.02
    Ref. [21]2.076.86
    Ours1.051.59
    Table 1. Performance indexes of different models on the UCSD data set
    ModelfMAEfMSE
    ConvLSTM[3]2.248.50
    Bidirectional ConvLSTM[3]2.107.60
    Ref. [20]3.5919.00
    Ref.[21]3.4317.70
    Ours1.957.50
    Table 2. Performance indexes of different models on the Mall data set
    ModelfMAEfMSE
    ConvLSTM[3]4.515.91
    MCNN[22]3.814.92
    Ours without ST4.325.21
    Ours3.515.10
    Table 3. Performance indexes of different models on self-built data set
    Data setNo dilated No STdilated+No ST
    fMAEfMSEfMAEfMSE
    UCSD1.714.251.524.13
    Mall2.899.012.138.51
    Self-built4.746.654.325.21
    Table 4. Confirmation experiment results1 of different data sets
    Data setNo dilated No STNo dilated+ STdilated+ST
    fMAEfMSEfMAEfMSEfMAEfMSE
    UCSD1.714.251.413.521.051.59
    Mall2.899.012.018.431.957.50
    Self-built4.746.654.335.233.515.10
    Table 5. Confirmation experiment results 2 of different data sets
    Chunyan Yu, Yan Xu, Lisha Gou, Zhefeng Nan. Crowd Counting Based on Single-Column Deep Spatiotemporal Convolutional Neural Network[J]. Laser & Optoelectronics Progress, 2021, 58(8): 0810011
    Download Citation