Chinese Journal of Liquid Crystals and Displays, Volume. 39, Issue 8, 1079(2024)

Video object tracking algorithm based on dual-branch online optimization and feature fusion

Xinpeng LI1, Peng WANG1、*, Xiaoyan LI1, Mengyu SUN1, Zuntian CHEN2, and Hui GAO1
Author Affiliations
  • 1School of Electronics and Information Engineering,Xi'an Technological University,Xi'an 710021,China
  • 2Xi'an Institute of Electromechanical Information Technology,Xi'an 710065,China
  • show less
    Figures & Tables(12)
    Network architecture of D3S algorithm
    Network architecture of the algorithm proposed in this article
    Structure of encoder module
    Training loss of the algorithm proposed in this article
    Experimental results of UAV123 dataset
    Experimental results of occlusion and similar object attribute in UAV123 dataset
    Visualized results of four sequences on VOT2018 dataset
    • Table 1. Experimental results on VOT2018 dataset

      View table
      View in Article

      Table 1. Experimental results on VOT2018 dataset

      TrackerEAOAccuracyRobustness
      Ours0.5180.6600.140
      D3S0.4890.6400.150
      SiamRPN++0.4140.6000.234
      SiamRPN0.3830.5880.276
      ATOM0.4010.5900.204
      SiamMask0.3800.6090.276
    • Table 2. Precision results of visual attribute experiment on VOT2018 dataset

      View table
      View in Article

      Table 2. Precision results of visual attribute experiment on VOT2018 dataset

      Tracker运动模糊光照变化快速运动遮挡尺度变化无标签
      Ours0.655 00.621 40.642 00.508 10.662 30.679 9
      D3S0.641 20.649 00.618 30.444 10.660 50.673 5
      ATOM0.598 40.590 40.599 70.503 70.604 50.581 6
      SiamMask0.620 30.615 80.620 00.467 80.628 00.607 9
      SiamRPN0.591 40.564 80.588 80.501 60.574 40.589 3
      SiamRPN++0.620 10.604 20.602 80.507 50.575 80.603 5
    • Table 3. Robustness results of visual attribute experiment on VOT2018 dataset

      View table
      View in Article

      Table 3. Robustness results of visual attribute experiment on VOT2018 dataset

      Tracker运动模糊光照变化快速运动遮挡尺度变化无标签
      Ours15.00.05.015.03.06.0
      D3S16.00.06.016.03.011.0
      ATOM18.80.811.415.65.69.4
      SiamMask29.00.017.018.06.09.0
      SiamRPN29.02.019.017.010.011.0
      SiamRPN++24.01.014.017.08.010.0
    • Table 4. Ablation experiments results on VOT2018 dataset

      View table
      View in Article

      Table 4. Ablation experiments results on VOT2018 dataset

      消融实验AccuracyRobustnessEAO
      baseline0.6400.1500.489
      baseline+BF0.6550.1640.498
      baseline+FF0.6490.1550.504
      baseline+BF+FF0.6570.1600.506
      baseline+FF+TU0.6510.1450.509
      baseline+BF+FF+TU0.6600.1400.518
    • Table 5. Real time experiments results on VOT2018 dataset

      View table
      View in Article

      Table 5. Real time experiments results on VOT2018 dataset

      实时性实验FPS
      D3S22.9
      Ours17.6
    Tools

    Get Citation

    Copy Citation Text

    Xinpeng LI, Peng WANG, Xiaoyan LI, Mengyu SUN, Zuntian CHEN, Hui GAO. Video object tracking algorithm based on dual-branch online optimization and feature fusion[J]. Chinese Journal of Liquid Crystals and Displays, 2024, 39(8): 1079

    Download Citation

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

    Category: Object Tracking and Recognition

    Received: Jul. 25, 2023

    Accepted: --

    Published Online: Sep. 27, 2024

    The Author Email: Peng WANG (wp_xatu@163.com)

    DOI:10.37188/CJLCD.2023-0256

    Topics