• Optical Instruments
  • Vol. 46, Issue 5, 1 (2024)
Dong LIU, Rongfu ZHANG*, Junxiang QIN, Junzhe GONG, and Zhibin CAO
Author Affiliations
  • School of Optical-Electrical and Computer Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China
  • show less
    DOI: 10.3969/j.issn.1005-5630.202308160108 Cite this Article
    Dong LIU, Rongfu ZHANG, Junxiang QIN, Junzhe GONG, Zhibin CAO. Architectural style classification algorithm fusing CNN and Transformer[J]. Optical Instruments, 2024, 46(5): 1 Copy Citation Text show less
    Edwardian architecture with different stylistic elements
    Fig. 1. Edwardian architecture with different stylistic elements
    Structure of architectural style classification network(FCT-Net)
    Fig. 2. Structure of architectural style classification network(FCT-Net)
    Structure diagram of CT-Block
    Fig. 3. Structure diagram of CT-Block
    Confusion matrix of partial results on Architectural Style Dataset
    Fig. 4. Confusion matrix of partial results on Architectural Style Dataset
    模型准确率/%
    40%类别100%类别
    注:黑体为同类别中最大准确率
    DCNN[6]72.4266.60
    MonuNet[24]71.2061.93
    ResNet-5080.1967.41
    Inception-v367.1560.06
    ViT70.0157.14
    Swin-Transformer75.3665.28
    Visformer76.3370.49
    FCT-Net(ours)83.0979.83
    Table 1. Comparison of accuracy of different models on Architectural Style Dataset
    模型准确率/%
    注:黑体为最大准确率。
    MobileNet-V256.63
    Swin-Transformer52.05
    Mobile-former60.39
    Conformer63.50
    Visformer61.54
    FCT-Net(ours)68.41
    Table 2. Comparison of accuracy of different models on WikiChurches
    模型准确率/%
    Architectural Style DatasetWikiChurches
    注:黑体为最大准确率。
    ResNet-5067.4162.36
    MobileNet-V266.6756.63
    ViT57.1449.10
    Swin-Transformer65.2852.05
    FCT-Net(ours)79.8368.41
    Table 3. Comparison of accuracy of different types of models on public datasets
    模型准确率/%
    Net175.91
    Net263.69
    FCT-Net79.83
    Table 4. Comparison of accuracy of CT-Block modules on the Architectural Style Dataset
    Dong LIU, Rongfu ZHANG, Junxiang QIN, Junzhe GONG, Zhibin CAO. Architectural style classification algorithm fusing CNN and Transformer[J]. Optical Instruments, 2024, 46(5): 1
    Download Citation