Simulation Tools and Techniques. 11th International Conference, SIMUtools 2019, Chengdu, China, July 8–10, 2019, Proceedings

Research Article

Study the Preprocessing Effect on RNN Based Online Uyghur Handwriting Word Recognition

Download
86 downloads
  • @INPROCEEDINGS{10.1007/978-3-030-32216-8_43,
        author={Wujiahemaiti Simayi and Mayire Ibrayim and Askar Hamdulla},
        title={Study the Preprocessing Effect on RNN Based Online Uyghur Handwriting Word Recognition},
        proceedings={Simulation Tools and Techniques. 11th International Conference, SIMUtools 2019, Chengdu, China, July 8--10, 2019, Proceedings},
        proceedings_a={SIMUTOOLS},
        year={2019},
        month={10},
        keywords={Online handwriting recognition Preprocessing Input representation Recurrent neural networks Connectionist temporal classification Uyghur words},
        doi={10.1007/978-3-030-32216-8_43}
    }
    
  • Wujiahemaiti Simayi
    Mayire Ibrayim
    Askar Hamdulla
    Year: 2019
    Study the Preprocessing Effect on RNN Based Online Uyghur Handwriting Word Recognition
    SIMUTOOLS
    Springer
    DOI: 10.1007/978-3-030-32216-8_43
Wujiahemaiti Simayi1, Mayire Ibrayim1, Askar Hamdulla1,*
  • 1: Xinjiang University
*Contact email: askar@xju.edu.cn

Abstract

There is little work done on unconstrained handwritten Uyghur word recognition by implementing deep neural networks. This paper carries out a comparative study to see the preprocessing effect on training a neural network based online handwriting Uyghur word recognition system. Bidirectional recurrent neural network with connectionist temporal classification is implemented for unconstrained handwriting word recognition experiments on a dataset of 23400 Uyghur word samples. The results are directly obtained from model output without any lexicon or language model. Experiments showed that proper preprocessing can improve the training speed very effectively. The comparative study conducted in this paper can be good reference for later studies.