Optics and Precision Engineering, Volume. 31, Issue 21, 3145(2023)

Intra-inter channel attention for few-shot classification

Liping YANG1,*... Tianyang ZHANG1, Yuyang WANG1 and Xiaohua GU2 |Show fewer author(s)
Author Affiliations
  • 1Key Laboratory of Optoelectronic Technique & System of Ministry of Education, Chongqing University, Chongqing400044, China
  • 2School of Electrical Engineering, Chongqing University of Science & Technology, Chongqing401331, China
  • show less
    Figures & Tables(12)
    Prototypical Network
    Intra-inter channel attention prototypical network for FSC
    Schematic diagram of intra-inter channel attention
    Schematic diagram of intra-class channel attention
    schematic diagram of inter-class channel attention
    Channel distance splitting and replication f(φ)=f(φ1)⋃f(φ2)
    • Table 1. Structural parameters of fully connected network

      View table
      View in Article

      Table 1. Structural parameters of fully connected network

      层名称输出尺寸
      输入Npre×C
      全连接层Npre×C/2
      批归一化Npre×C/2
      ReLUNpre×C/2
      全连接层Npre×C
      SigmoidNpre×C
      转换层Npre×C
    • Table 2. Parameter setting of convolutional neural network

      View table
      View in Article

      Table 2. Parameter setting of convolutional neural network

      模块名称
      卷积层

      c1:64,c2:64,c3:64,c4:64

      k1:3,k2:3,k3:3,k4:3

      批归一化c1:64,c2:64,c3:64,c4:64
      最大池化

      s1:2,2,s2:2,2,

      s3:2,2,s4:(2,2)

    • Table 3. Parameter configuration of few-shot classification task in meta-train stage

      View table
      View in Article

      Table 3. Parameter configuration of few-shot classification task in meta-train stage

      N-way K-shot元训练阶段
      wayshotquery
      5-way 5-shot20515
      5-way 1-shot30115
    • Table 4. Experimental results for ICAFSC updating ICAM parameters in different training stages on MiniImagenet

      View table
      View in Article

      Table 4. Experimental results for ICAFSC updating ICAM parameters in different training stages on MiniImagenet

      模 型准确率/%
      5-way 1-shot5-way 5-shot
      ProtoNet49.62±0.7867.96±0.62
      ICAFSC-pre51.55±0.7569.11±0.65
      ICAFSC-tra49.76±0.7768.14±0.60
      ICAFSC-pre+tra50.70±0.7568.87±0.63
    • Table 5. Experimental results for ICAFSC under different classification task configurations in pre-training phase on MiniImagenet

      View table
      View in Article

      Table 5. Experimental results for ICAFSC under different classification task configurations in pre-training phase on MiniImagenet

      模 型预训练任务准确率/%
      WayShotQuery5-way 1-shot5-way 5-shot
      ProtoNet---49.62±0.7867.96±0.62
      ICAFSC51001551.55±0.7569.11±0.65
      ICAFSC101001550.30±0.7768.79±0.60
      ICAFSC201001549.97±0.7968.47±0.66
    • Table 6. Classification scores of different methods with Conv4-64 as backbone on MiniImagenet dataset

      View table
      View in Article

      Table 6. Classification scores of different methods with Conv4-64 as backbone on MiniImagenet dataset

      模 型准确率/%
      5-way 1-shot5-way 5-shot
      ProtoNet849.62±0.7867.96±0.62
      MatchNet1443.40±0.7851.09±0.71
      MAML1648.07±1.7563.15±0.91
      P-RelationNet1749.54±0.8268.34±0.70
      BOIL1949.61±0.1666.45±0.37
      D2D22051.20±0.6068.80±0.10
      HyperProto2150.21±0.3166.48±0.71
      DAM2253.80±0.1969.31±0.16
      CSS2350.85±0.8468.08±0.73
      HyperShot2352.42±0.4668.78±0.29
      OVE2448.00±0.2467.14±0.23
      IMP2549.60±0.8068.10±0.80
      ICAFSC51.55±0.7569.11±0.65
    Tools

    Get Citation

    Copy Citation Text

    Liping YANG, Tianyang ZHANG, Yuyang WANG, Xiaohua GU. Intra-inter channel attention for few-shot classification[J]. Optics and Precision Engineering, 2023, 31(21): 3145

    Download Citation

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

    Category:

    Received: Apr. 6, 2023

    Accepted: --

    Published Online: Jan. 5, 2024

    The Author Email: YANG Liping (yanglp@cqu.edu.cn)

    DOI:10.37188/OPE.20233121.3145

    Topics