Laser & Optoelectronics Progress, Volume. 59, Issue 22, 2220001(2022)

Multi-Granularity and Cross-Modality Pedestrian Re-Identification Algorithm

Wei Xiong1,2、*, Ling Yue1, Lei Zhou1, Kai Zhang1, and Lirong Li1
Author Affiliations
  • 1School of Electrical & Electronic Engineering, Hubei University of Technology, Wuhan 430068, Hubei , China
  • 2Dept. of Computer Science & Engineering, University of South Carolina, Columbia 29201, SC, USA
  • show less
    Figures & Tables(6)
    Block diagram of proposed algorithm
    Non_local attention module
    SYSU-MM01 dataset
    • Table 1. Ablation experiment

      View table

      Table 1. Ablation experiment

      MethodAll SearchIndoor Search
      Rank-1mAPRank-1mAP
      Baseline57.3455.1363.1169.20
      Baseline+Non_local61.1657.0367.4872.22
      Baseline+Non_local+Branch 162.8358.1068.8973.47
    • Table 2. Comparison results on SYSU-MM01 dataset

      View table

      Table 2. Comparison results on SYSU-MM01 dataset

      MethodAll SearchIndoor Search
      Rank-1mAPmINPRank-1mAPmINP
      One-stream18(ICCV2017)12.0413.6716.9422.95
      Two-stream18(ICCV2017)11.6512.8515.6021.49
      Zero-padding18(ICCV2017)14.8015.9520.5826.92
      HCML6(AAAI2018)14.3216.1624.5230.08
      cmGAN8(IJCAI2018)26.9731.4931.6342.19
      BDTR19(TIFS2019)27.3227.3231.9241.86
      D2RL20(CVPR2019)28.9029.20
      MAC21(TIP2020)33.2636.2236.4337.03
      AlignGAN22(ICCV2019)42.4040.7045.9054.30
      Xmodal10(AAAI2020)49.9250.73
      DDAG23(ECCV2020)54.7553.0261.2067.98
      AGW13(TPAMI2021)47.5047.6535.3054.1762.9759.23
      DF2AM14(Arxiv2021)56.9355.1066.3971.52
      DGTL17(ISPL2021)57.3455.1363.1169.20
      Proposed method62.8358.1045.6268.8973.4768.06
    • Table 3. Comparison results on RegDB dataset

      View table

      Table 3. Comparison results on RegDB dataset

      MethodVisible to InfraredInfrared to Visible
      Rank-1mAPmINPRank-1mAPmINP
      HCML6(AAAI2018)24.4420.0821.7022.24
      Zero-padding18(ICCV2017)17.7518.9016.6317.82
      BDTR19(TIFS2019)33.5632.7632.9231.96
      D2RL20(CVPR2019)43.4044.10
      MAC21(TIP2020)36.4337.0336.2036.63
      AlignGAN22(ICCV2019)57.9053.6056.3053.40
      Xmodal10(AAAI2020)62.2160.18
      DDAG23(ECCV2020)69.3463.4668.0661.80
      AGW13(TPAMI2021)70.0566.3750.19
      DF2AM14(Arxiv2021)73.0667.8170.4963.85
      DGTL17(ISPL2021)83.9273.7881.5971.65
      Proposed method87.7876.2258.4884.9773.81
    Tools

    Get Citation

    Copy Citation Text

    Wei Xiong, Ling Yue, Lei Zhou, Kai Zhang, Lirong Li. Multi-Granularity and Cross-Modality Pedestrian Re-Identification Algorithm[J]. Laser & Optoelectronics Progress, 2022, 59(22): 2220001

    Download Citation

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

    Category: Optics in Computing

    Received: Aug. 23, 2021

    Accepted: Sep. 28, 2021

    Published Online: Oct. 13, 2022

    The Author Email: Xiong Wei (xw@mail.hbut.edu.cn)

    DOI:10.3788/LOP202259.2220001

    Topics