Opto-Electronic Engineering, Volume. 50, Issue 7, 230136(2023)

Infrared-visible person re-identification based on multi feature aggregation

Haijun Zheng1, Bin Ge1、*, Chenxing Xia1,2, and Cheng Wu1
Author Affiliations
  • 1College of Computer Science and Engineering, Anhui University of Science and Technology, Huainan, Anhui 232001, China
  • 2Institute of Energy, Hefei Comprehensive National Science Center, Hefei, Anhui 230031, China
  • show less
    Figures & Tables(11)
    MFANet structure diagram
    Adjacency feature aggregation module
    Multi scale aggregation module
    Multi scale feature aggregation module
    Inter-class intra-class distance and feature distribution diagram
    Heat map at different receptive fields of view
    Visual sorting results on SYSU-MM01
    • Table 1. Comparison results on SYSU-MM01 datasets

      View table
      View in Article

      Table 1. Comparison results on SYSU-MM01 datasets

      MethodSetting
      All searchIndoor search
      rank-1rank-10rank-20mAPmINPrank-1rank-10rank-20mAPmINP
      One-stream[14]12.0449.6866.7413.67-16.9463.5582.1022.95-
      Two-stream[14]11.6547.9965.5012.85-15.6061.1881.0221.49-
      Zero-Padding[14]14.8054.1271.3315.59-20.5868.3885.7926.92-
      HCML[18]14.3253.1669.1716.16-24.5273.2586.7330.08-
      BDTR[17]27.3266.9681.0727.32-31.9277.1889.2841.86-
      D2RL[4]28.9070.6082.4029.20------
      AlignGAN[19]42.0385.2593.7341.48-45.8690.1795.3955.18-
      AGW[16]47.5884.4592.1147.6935.3054.2991.1495.9963.0259.23
      Xmodal[20]49.9289.7995.9650.73------
      DDAG[8]53.6189.1795.3052.0239.6258.3791.9297.4265.4462.61
      CM-NAS[21]62.0492.9297.3160.00-67.0397.0299.3472.97-
      CAJ[9]68.2395.5998.4965.3253.6174.0197.7999.6778.5276.79
      MPANet[6]70.0795.3998.3967.07-76.3597.5699.4880.16-
      PIC[23]57.5--55.1-60.4--67.7-
      DART[24]68.7296.3998.9666.2953.2672.5297.8499.4678.1774.94
      SPOT[10]65.3492.7397.0462.2548.8669.4296.2299.1274.6370.48
      DML[7]58.4091.2095.8056.10-62.4095.2098.7069.50-
      PMT[12]67.5395.3698.6464.9851.8671.6696.7399.2576.5272.74
      SFANet[25]65.7492.9897.0560.83-71.6096.6099.4580.05-
      SIDA[26]68.3695.9198.5664.19-73.2897.3599.5277.49-
      MTMFE[27]69.4796.4299.1166.41-72.5696.9899.2076.58-
      Ours71.7796.1598.768.4355.2178.2498.2399.4981.978.44
    • Table 2. Comparison results on RegDB datasets

      View table
      View in Article

      Table 2. Comparison results on RegDB datasets

      MethodSetting
      Visible to infraredInfrared to visible
      rank-1rank-10rank-20mAPmINPrank-1rank-10rank-20mAPmINP
      Zero-Padding[14]17.7534.2144.3518.90-16.6334.6844.2517.82-
      HCML[18]24.4447.5356.7820.08-21.7045.0255.5822.24-
      BDTR[17]33.5658.6167.4332.76-32.9258.4668.4331.96-
      D2RL[4]43.4066.1076.3044.10-----
      AGW[16]70.0586.2191.5566.3750.1970.4987.2191.8465.9051.24
      Xmodal[20]62.2183.1391.7260.18------
      DDAG[8]69.3485.7789.9863.1949.2464.7783.8588.9058.5448.62
      CM-NAS[21]84.5495.1897.8580.32-82.5694.5297.3778.31-
      MCLNet[22]80.3192.7096.0373.07-75.9390.9394.5969.49-
      CAJ[9]84.7295.1797.3878.7065.3384.0994.7997.1177.2561.56
      PIC[23]83.6--79.6-79.5--77.4-
      DART[24]78.23--67.0448.3675.04--64.3843.32
      SPOT[10]80.3593.4896.4472.4656.1979.3792.7996.0172.2656.06
      DML[7]77.60--84.30-77.00--83.60-
      PMT[12]84.83--76.55-84.16--75.13-
      SFANet[25]76.3191.0294.2768.00-70.1585.2489.2763.77-
      SIDA[26]81.73-96.5575.07-79.71-95.4772.60-
      MTMFE[27]85.0494.3897.2282.52-81.1192.3596.1979.59-
      Ours85.3895.3997.5479.4965.7284.5895.2797.2378.0262.22
    • Table 3. Ablation study of four different settings on the SYSU-MM01 dataset

      View table
      View in Article

      Table 3. Ablation study of four different settings on the SYSU-MM01 dataset

      SettingsAll searchIndoor search
      AFAMMSAMrank-1rank-10rank-20mAPmINPrank-1rank-10rank-20mAPmINP
      68.2395.5998.4965.3251.9074.0197.7999.6778.5274.78
      69.3095.6998.4165.9552.1475.2797.8499.4879.5175.79
      70.8995.8898.5267.6154.377.6997.4399.2581.0977.48
      71.7796.1598.7068.4355.2178.2498.2399.4981.9078.44
    • Table 4. Multi scale feature aggregation module receptive field analysis

      View table
      View in Article

      Table 4. Multi scale feature aggregation module receptive field analysis

      SettingsReceptive field
      All searchIndoor search
      rank-1rank-10rank-20mAPmINPrank-1rank-10rank-20mAPmINP
      1,3,5,769.4195.8298.5466.2752.9675.3497.9699.6679.9876.53
      1,2,3,470.1795.5198.5167.1454.176.4498.0999.680.6577.22
      1,3,570.595.7798.5467.1153.6876.6697.8199.5180.5476.99
      1,2,370.5395.8698.6767.2153.7776.5997.8899.3780.6877.22
      1,371.7796.1598.7068.4355.2178.2498.2399.4981.9078.44
    Tools

    Get Citation

    Copy Citation Text

    Haijun Zheng, Bin Ge, Chenxing Xia, Cheng Wu. Infrared-visible person re-identification based on multi feature aggregation[J]. Opto-Electronic Engineering, 2023, 50(7): 230136

    Download Citation

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

    Category: Article

    Received: Jun. 15, 2023

    Accepted: Aug. 11, 2023

    Published Online: Sep. 25, 2023

    The Author Email:

    DOI:10.12086/oee.2023.230136

    Topics