• Laser & Optoelectronics Progress
  • Vol. 57, Issue 8, 081002 (2020)
Wanrong Huang, Kai He*, Kun Liu, and Shengnan Gao
Author Affiliations
  • School of Electrical and Information Engineering, Tianjin University, Tianjin 300072, China
  • show less
    DOI: 10.3788/LOP57.081002 Cite this Article Set citation alerts
    Wanrong Huang, Kai He, Kun Liu, Shengnan Gao. Handwritten Chinese Character Recognition Based on Attention Mechanism[J]. Laser & Optoelectronics Progress, 2020, 57(8): 081002 Copy Citation Text show less
    AT model network structure
    Fig. 1. AT model network structure
    AT-CNN model structure
    Fig. 2. AT-CNN model structure
    Examples of HWDB dataset
    Fig. 3. Examples of HWDB dataset
    AT-CNN model training result. (a) Accuracy graph; (b) loss graph
    Fig. 4. AT-CNN model training result. (a) Accuracy graph; (b) loss graph
    Handwritten Chinese text
    Fig. 5. Handwritten Chinese text
    Sequence of layerFeature map sizeNumber of feature mapsFilter sizeStep size
    C164×64323×31
    MP232×32322×22
    C332×32643×31
    MP416×16642×22
    C516×161283×31
    MP68×81282×22
    C78×82563×31
    C88×82563×31
    MP94×42562×22
    FC1×11024
    Output1×13755
    Table 1. Structure and parameters of the AT-CNN model
    Recognition methodRAM/MBAccuracy /%
    MQDF[20]89.55
    CCPR-2010 champion: HKU[21]339. 1089. 99
    MQDF+CNN[20]92.03
    ICDAR-2011 champion: IDSIAnn-2[22]71. 3592. 18
    ICDAR-2013 champion: Fujitsu[3]2460. 0094. 77
    HCCR-Ensemble-GoogLeNet[11]277. 2596. 74
    CNN108.3093.01
    AT-CNN115.3095.05
    Table 2. Comparison of the recognition accuracy of different methods on HWDB dataset
    Wanrong Huang, Kai He, Kun Liu, Shengnan Gao. Handwritten Chinese Character Recognition Based on Attention Mechanism[J]. Laser & Optoelectronics Progress, 2020, 57(8): 081002
    Download Citation