Journals Articles Conferences News About CLP
Submit a paper Email Alert
Search
Search
Articles
Journals
News
Advanced Search
Top Searches
2D MaterialsTransformation opticsQuantum PhotonicsSmall lasersSemiconductor UV PhotonicsSupersymmetric microring laser arrays

Laser & Optoelectronics Progress, Volume. 57, Issue 12, 121011(2020)

Bilinear Residual Attention Networks for Fine-Grained Image Classification

Yang Wang and Libo Liu*
Author Affiliations
  • School of Information Engineering, Ningxia University, Yinchuan, Ningxia 750021, China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (11)
    Equations (0)
    References (27)
    Cited By (3)
    Tools
    Paper Information
    Topics
    Figures & Tables(11)
    Architecture of B-CNN model

    Fig. 1. Architecture of B-CNN model

    Download full size
    Schematic of bilinear feature combination

    Fig. 2. Schematic of bilinear feature combination

    Download full size
    Structure of residual unit

    Fig. 3. Structure of residual unit

    Download full size
    Channel attention module

    Fig. 4. Channel attention module

    Download full size
    Spatial attention module

    Fig. 5. Spatial attention module

    Download full size
    Residual attention structure of BRAN model

    Fig. 6. Residual attention structure of BRAN model

    Download full size
    Example of training data augmentation

    Fig. 7. Example of training data augmentation

    Download full size
    Visualization of different feature maps. (a) Original images; (b) B-CNN; (c) channel attention maps; (d) spatial attention maps

    Fig. 8. Visualization of different feature maps. (a) Original images; (b) B-CNN; (c) channel attention maps; (d) spatial attention maps

    Download full size

    • Table 1. Detailed statistics of three fine-grained image datasets

      View table

      Table 1. Detailed statistics of three fine-grained image datasets

      DatasetClassTrainTestTotal
      CUB-200-20112005994579411788
      Stanford Dogs12012000858020580
      Stanford Cars1968144804116185

    • Table 2. Ablation experiment and analysis of proposed method on CUB-200-2011 dataset

      View table

      Table 2. Ablation experiment and analysis of proposed method on CUB-200-2011 dataset

      ApproachBackboneAccuracy /%
      B-CNN(baseline)VGG-M+VGG-D84.1
      B-CNN(resnet×2)ResNet-34×285.0
      BRAN(cha. attention)ResNet-34×2 + channel attention86.2
      BRAN(spa. attention)ResNet-34×2 + spatial attention85.5
      BRAN(cha.& spa. attention)ResNet-34×2 + cha. & spa. attention87.2

    • Table 3. Comparison with weakly-supervised methods in terms of classification accuracy

      View table

      Table 3. Comparison with weakly-supervised methods in terms of classification accuracy

      ApproachBackboneAccuracy /%
      Birds[9]Dogs[10]Cars[11]
      Two-level attention[22]VGG1977.9--
      NAC[23]VGG1981.0168.61-
      B-CNN[6]VGG-M+VGG-D84.1-91.3
      ST-CNN[24]Inception-v2×384.1--
      DVAN[25]VGG-16×379.081.587.1
      RA-CNN[26]VGG-19×385.387.392.5
      MA-CNN[19]VGG-19×386.5-92.8
      MAMC[27]ResNet-10186.585.293.0
      BRANResNet-34×287.289.292.5
    Tools

    Get Citation

    Copy Citation Text

    Yang Wang, Libo Liu. Bilinear Residual Attention Networks for Fine-Grained Image Classification[J]. Laser & Optoelectronics Progress, 2020, 57(12): 121011

    Download Citation

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

    Category: Image Processing

    Received: Aug. 19, 2019

    Accepted: Nov. 2, 2019

    Published Online: Jun. 3, 2020

    The Author Email: Liu Libo (liulib@163.com)

    DOI:10.3788/LOP57.121011

    Topics

    laser devices and laser physics
    Lasers and Laser Optics
    Laser physics
    laser manufacturing
    Instrumentation, Measurement and Metrology