• Journal of Applied Optics
  • Vol. 43, Issue 6, 1088 (2022)
Guoyin REN, Xiaoqi LYU*, and Yuhao LI
Author Affiliations
  • School of Mechanical Engineering, Inner Mongolia University of Science and Technology, Baotou 014010, China
  • show less
    DOI: 10.5768/JAO202243.0604007 Cite this Article
    Guoyin REN, Xiaoqi LYU, Yuhao LI. Night vision dense crowd counting based on mid-term fusion of thermal imaging features[J]. Journal of Applied Optics, 2022, 43(6): 1088 Copy Citation Text show less
    Structure diagram of Rgb-T-net heatmap fusion perceptron
    Fig. 1. Structure diagram of Rgb-T-net heatmap fusion perceptron
    Flow chart of overall system design
    Fig. 2. Flow chart of overall system design
    Visualization of curve parameters in model training process
    Fig. 3. Visualization of curve parameters in model training process
    Visualization results from ShanghaiTechPartA, ShanghaiTechPartB, UCF-QNRF and UCF_CC_50 datasets (from left to right: input images, real annotations, Bayesian results and results from proposed method)
    Fig. 4. Visualization results from ShanghaiTechPartA, ShanghaiTechPartB, UCF-QNRF and UCF_CC_50 datasets (from left to right: input images, real annotations, Bayesian results and results from proposed method)
    Crowd detection results on heatmap datasets
    Fig. 5. Crowd detection results on heatmap datasets
    Comparison of average precision recall curves
    Fig. 6. Comparison of average precision recall curves
    数据集分辨率数据类型数量最大最小平均总计模态
    Rgb-T-CC640×480Rgb+T4060824568138,389Rgb-T
    Table 1. Parameter information for Rgb-T-CC dataset
    模型(Rgb-T-CC数据集)MAEMSE
    UCNet[18]33.9656.31
    HDFNet[19]22.3633.93
    BBSNet[20]19.5632.48
    MCNN[15]21.8937.44
    CSRNet[16]20.435.26
    Bayesian Loss[17]18.732.67
    MCNN+IADM[15]19.7730.34
    CSRNet+IADM[15]17.9430.91
    Bayesian Loss+IADM[15]15.6128.18
    本文Rgb-T-net18.1632.14
    Table 2. Comparison of different state-of-the-art methods on Rgb-T-CC dataset
    融合方式(Rgb-T-CC数据集)MAEMSE
    AGK22.4638.97
    AGK+Rgb-T-net(早期融合)18.0131.49
    AGK+Rgb-T-net(中期融合)18.1632.14
    AGK+Rgb-T-net(晚期融合)19.3534.71
    Table 3. Comparison of different fusion methods on Rgb-T-CC dataset
    Guoyin REN, Xiaoqi LYU, Yuhao LI. Night vision dense crowd counting based on mid-term fusion of thermal imaging features[J]. Journal of Applied Optics, 2022, 43(6): 1088
    Download Citation