Journals Articles Conferences News About CLP
Submit a paper Email Alert
Search
Search
Articles
Journals
News
Advanced Search
Top Searches
2D MaterialsTransformation opticsQuantum PhotonicsSmall lasersSemiconductor UV PhotonicsSupersymmetric microring laser arrays

Acta Physica Sinica, Volume. 69, Issue 8, 084203-1(2020)

Review of pedestrian tracking: Algorithms and applications

Zi-Qiang Cao, Bin Sai, and Xin Lu*
Author Affiliations
  • College of Systems Engineering, National University of Defense Technology, Changsha 410073, China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (9)
    Equations (49)
    References (113)
    Tools
    Paper Information
    Topics
    Figures & Tables(9)
    Markov decision process flow chart[31]

    Fig. 1. Markov decision process flow chart[31]

    Download full size
    View in Article
    CNN basic structure diagramt

    Fig. 2. CNN basic structure diagramt

    Download full size
    View in Article
    RCNN algorithm flowchart[45]

    Fig. 3. RCNN algorithm flowchart[45]

    Download full size
    View in Article
    SPP-Net structure diagramt[48]

    Fig. 4. SPP-Net structure diagramt[48]

    Download full size
    View in Article
    RPN structure diagramt[47]

    Fig. 5. RPN structure diagramt[47]

    Download full size
    View in Article
    CornerNet structure diagramt[59]

    Fig. 6. CornerNet structure diagramt[59]

    Download full size
    View in Article

    • Table 1.

      Prediction and update process of Kalman filtering

      卡尔曼滤波的预测阶段和更新阶段

      View table
      View in Article

      Table 1.

      Prediction and update process of Kalman filtering

      卡尔曼滤波的预测阶段和更新阶段

      预测阶段更新阶段
      $\hat { x}_k^ - \!=\! { A}\hat { x}_{k-1}^ - \!+\! { B}{ U_{k - 1} }$${ { K}_k} \!=\! { P}_k^ - { { H}^{\rm T} }{({ {HP} }_k^ - { { H}^{\rm T} } \!+\! { R})^{ - 1} }$
      ${ P}_k^ - \!=\! { A}{ { P}_{k - 1} }{ { A}^T} \!+\! { Q}$${ {\hat { x} }_k} \!=\! \hat { x}_k^ - \!+\! { { K}_k}({ { y}_k} \!-\! { H}\hat { x}_k^ - )$
      ${{ P}_k} = ({ I} - {{ K}_k}{ H}){ P}$

    • Table 2.

      Parameters and meanings in the Kalman filter formula

      卡尔曼滤波公式中的参数及含义

      View table
      View in Article

      Table 2.

      Parameters and meanings in the Kalman filter formula

      卡尔曼滤波公式中的参数及含义

      参数含义
      $\hat { x}_k^-$目标在 $k$时刻的先验状态估计值, 包括目标的位置、速度等参数, 一般是 $n$维向量
      ${ {\hat { x}}_k}$目标在 $k$时刻的后验状态估计值, 是对 $\hat { x}_k^-$应用卡尔曼滤波更新后的值
      ${{\hat { x}}_{k - 1}}$目标在 $k-1$时刻的后验状态估计值
      ${ A}$状态转移矩阵, 一般是 $n \times n$阶的方阵
      ${ B}$控制矩阵, 一般为0
      ${ U}_{k-1}$外部控制量, 一般也为0
      ${ P}_k^-$$k$时刻的先验误差协方差矩阵, 需要事先给定一个初始值, 以后的值可以由卡尔曼滤波递归得到
      ${ P}_k$$k$时刻的后验误差协方差矩阵, 是对 ${ P}_k^-$的修正
      ${ K}_k$卡尔曼增益
      ${ y}_k$测量值, 一般只能测量目标的位置, 是 $m$维向量
      ${ Q}$系统噪声协方差矩阵, 是一个需要调节的参数, 一般假定它是一个固定的值, 在实验中需要通过不断 调节 $Q$值, 来寻找滤波器的最优值
      ${ R}$观测噪声协方差矩阵, 和测量仪器有关, 在实验中要不断尝试来确定最优的 ${ R}$值
      ${ H}$观测矩阵, 是 $m \times n$阶矩阵, 用于将 $m$维的测量值 ${ y}_k$转换为与预测值 ${{\hat { x}}_k}$相同的 $n$维向量

    • Table 3.

      Performance comparison between different algorithms

      不同算法之间的性能对比

      View table
      View in Article

      Table 3.

      Performance comparison between different algorithms

      不同算法之间的性能对比

      算法MOTA↑MOTP↑MT↑ML↓IDS↓数据集类别
      卡尔曼滤波[64]85.00%————MIT Traffic video dataset传统跟踪算法
      多假设跟踪算法[21]29.10%71.70%12.10%53.30%476MOT Benchmark传统跟踪算法
      粒子滤波算法[27]——80.80%0.70%10CAVIAR dateset传统跟踪算法
      基于马尔科夫决策的 多目标跟踪算法[31]30.30%71.30%13.00%38.40%680MOT Benchmark传统跟踪算法
      相关滤波算法[65]83.40%73.50%———Urban Tracker dataset传统跟踪算法
      基于Faster-RCNN的跟踪算法[66]38.50%72.60%8.70%37.40%586MOT 15 Benchmark深度学习跟踪算法
      基于YOLOV3的跟踪算法[67]60.50%79.30%30.20%19.60%1129MOT 16 Benchmark深度学习跟踪算法
    Tools

    Get Citation

    Copy Citation Text

    Zi-Qiang Cao, Bin Sai, Xin Lu. Review of pedestrian tracking: Algorithms and applications[J]. Acta Physica Sinica, 2020, 69(8): 084203-1

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Set citation alerts for article
    Save article for my favorites
    Paper Information

    Received: Nov. 11, 2019

    Accepted: --

    Published Online: Nov. 24, 2020

    The Author Email:

    DOI:10.7498/aps.69.20191721

    Topics

    Nuclear physics
    Particles and fields
    Particle and nuclear astrophysics and cosmology