High Power Laser and Particle Beams, Volume. 36, Issue 8, 089001(2024)

Siamese single-object tracking algorithm based on multiple attention mechanisms and response fusion

Wenliang Feng1...2, Fanbao Meng1, Chuan Yu1 and Anqing You1 |Show fewer author(s)
Author Affiliations
  • 1Institute of Applied Electronics, CAEP, Mianyang 621900, China
  • 2Graduate School of China Academy of Engineering Physics, Mianyang 621900, China
  • show less
    Figures & Tables(11)
    Structure of Siamese single-object tracking algorithm based on multiple attention mechanism with response fusion
    Object feature fusion after conv3 downsampling and after conv5
    Structural diagram of the multiple attention mechanism
    Loss function curve during algorithm training
    Algorithm trace flowchart
    Precision and success rate of this algorithm versus other algorithms in the OTB-100 dataset
    Comparison of the success rate of this algorithm with other algorithms for 11 different attribute cases in the OTB-100 dataset
    Tracking results of 6 video sequences for this algorithm vs other algorithms
    Comparison of success rates under ablation experiments
    • Table 1. New backbone features to extract network information

      View table
      View in Article

      Table 1. New backbone features to extract network information

      definition layerlayerconvolution kernelstride/channelobject template sizesearch area size
      input/3127×127255×255
      conv1conv1-BN3×31/64125×125253×253
      conv2-BN3×31/128123×123251×251
      conv3-BN-ReLu1×11/64123×123251×251
      MaxPool2×22/6461×61125×125
      conv2conv4-BN3×31/12859×59123×123
      conv5-BN1×11/6459×59123×123
      conv6-BN-ReLu3×31/12857×57121×121
      MaxPool2×22/12828×2860×60
      conv3conv7-BN3×31/25626×2658×58
      conv8-BN1×11/12826×2658×58
      conv9-BN-ReLu3×31/25624×2456×56
      MaxPool2×22/25612×1228×28
      conv4conv10-BN3×31/51210×1026×26
      conv11-BN1×11/25610×1026×26
      conv12-BN3×31/5128×824×24
      conv13-BN-ReLu1×11/2568×824×24
      conv5conv143×31/2566×622×22
    • Table 2. Tracking speed comparison

      View table
      View in Article

      Table 2. Tracking speed comparison

      arithmeticaverage running speed/(frame/s)
      ours60
      SiamFC86
      MEEM6
      SRDCF4
      SAMF7
      DSST25
      CSK362
    Tools

    Get Citation

    Copy Citation Text

    Wenliang Feng, Fanbao Meng, Chuan Yu, Anqing You. Siamese single-object tracking algorithm based on multiple attention mechanisms and response fusion[J]. High Power Laser and Particle Beams, 2024, 36(8): 089001

    Download Citation

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

    Category: Advanced Interdisciplinary Science

    Received: Apr. 18, 2024

    Accepted: Jun. 30, 2024

    Published Online: Aug. 8, 2024

    The Author Email:

    DOI:10.11884/HPLPB202436.240130

    Topics