• Laser & Optoelectronics Progress
  • Vol. 57, Issue 12, 121002 (2020)
Siyao Li, Yuhong Liu, and Rongfen Zhang*
Author Affiliations
  • College of Big Data and Information Engineering, Guizhou University, Guiyang, Guizhou 550002, China
  • show less
    DOI: 10.3788/LOP57.121002 Cite this Article Set citation alerts
    Siyao Li, Yuhong Liu, Rongfen Zhang. Fine-Grained Image Classification Based on Multi-Scale Feature Fusion[J]. Laser & Optoelectronics Progress, 2020, 57(12): 121002 Copy Citation Text show less
    Detection effect for small object regions
    Fig. 1. Detection effect for small object regions
    Design of network architecture
    Fig. 2. Design of network architecture
    Process of multi-scale small object detection
    Fig. 3. Process of multi-scale small object detection
    Visualization of image processing
    Fig. 4. Visualization of image processing
    DatasetResnet101Resnet101+FPN(featurepyramidnetworks)Resnet101+FPN+objectfilter
    CUB-200-201181.483.485.7
    Stanford Dogs77.581.383.5
    Table 1. Classification accuracy of different components%
    DatasetTwo areasThree areasFour areas
    CUB-200-201183.685.786.0
    Stanford Dogs80.983.584.0
    Table 2. Classification accuracy for different numbers of small target areas%
    ReferenceMethodAccuracy /%
    Ref. [10]Pyramid matching80.4
    Ref. [8]DVAN81.5
    Ref. [9]RACNN87.3
    This paparMulti-scale feature fusion83.5
    Table 3. Comparison of classification accuracy of different algorithms for Stanford Dogs dataset
    ReferenceMethodAccuracy /%
    Ref. [11]Part-RCNN81.6
    Ref. [14]B-CNN84.1
    Ref. [7]PDFR (picking deepfilter responses)84.5
    Ref. [9]RACNN85.3
    This paparMulti-scale feature fusion85.7
    Table 4. Comparison of classification accuracy of different algorithms for CUB-200-2011 dataset
    Siyao Li, Yuhong Liu, Rongfen Zhang. Fine-Grained Image Classification Based on Multi-Scale Feature Fusion[J]. Laser & Optoelectronics Progress, 2020, 57(12): 121002
    Download Citation