Research on Hard-tipped Calligraphy Classification Based on Deep Learning Method
Xinxing Qiang, Minhua Wu, Liming Luo · 2019
Cultivating students' ability to write hard-tipped Calligraphy is a very important part of primary and secondary school education. The fair and professional evaluation of hard-tipped calligraphy can effectively promote students to correctly write Chinese characters. All along, the hard-tipped calligraphy written by the students is subjectively scored by the teacher. The difference of teachers' professional level leads to that they cannot give objective professional evaluation of hard-tipped calligraphy just like the experts in the field. Therefore, the method to use intelligent technology to achieve automatic machine evaluation is one of the research hotspots of current intelligent applications. This paper proposes the automatic classification of hard-tipped calligraphy copybooks using the convolutional neural network method based on caffe deep learning framework. Experiments show that the accuracy of single-word neural network modeling evaluation can reach 82.3%, and this method has good classification effect and strong robustness. The research work of this paper also has reference significance for the automatic evaluation of Chinese characters.