Laser & Optoelectronics Progress, Volume. 57, Issue 16, 161505(2020)

Convolutional Channel Pruning and Weighting for Accurate Location Visual Tracking

Manqiang Che*, Shubin Li, and Jinpeng Ge
Author Affiliations
  • Unmanned Systems Technology Innovation Center, Guangzhou Haige Communications Group Incorporated Company, Guangzhou, Guangdong 510700, China
  • show less
    Figures & Tables(7)
    Remained channel features after adaptively channel pruning
    Correlational filters of channel weighting
    Results of ten algorithms on OTB-100 dataset. (a) Precision plot; (b) success rate plot
    Precision of ten algorithms on different attributes
    Comparison of partial tracking results of eight algorithms at different scenes
    • Table 1. Tracking results of different tracking strategies

      View table

      Table 1. Tracking results of different tracking strategies

      Parameterp4p4scpcpgcpgwOurs
      Average CLE /pixel23.612.812.512.212.412.1
      Average DP /%83.787.788.889.190.591.3
      Average OP /%70.274.676.477.277.978.2
      Speed /(frame·s-1)26.233.434.033.232.131.8
    • Table 2. Tracking results of different algorithms on VOT2018 dataset

      View table

      Table 2. Tracking results of different algorithms on VOT2018 dataset

      ParameterECOCFCFCFWCRLSARTProposed algorithm
      EAO0.2800.2860.3030.3230.319
      Accuracy0.4830.5090.4840.4930.491
      Robustness0.2760.2810.2670.2180.225
    Tools

    Get Citation

    Copy Citation Text

    Manqiang Che, Shubin Li, Jinpeng Ge. Convolutional Channel Pruning and Weighting for Accurate Location Visual Tracking[J]. Laser & Optoelectronics Progress, 2020, 57(16): 161505

    Download Citation

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

    Category: Machine Vision

    Received: Jan. 6, 2020

    Accepted: Jan. 16, 2020

    Published Online: Aug. 5, 2020

    The Author Email: Manqiang Che (1229462669@qq.com)

    DOI:10.3788/LOP57.161505

    Topics