User interface modeling for convolutional neural network for complex character recognition

In this article, we design a user interface for a prototype desktop application using the capabilities of the author’s neural network for recognizing texts in Japanese written by one of the two Japanese alphabets – katakana or hiragana. During the design, the UML notation, a Use-Case Diagram, was used to build scenarios for using the program, and the BPMN notation was used to describe a program’s main algorithm. In the beginning of this article short versions of previous two articles were also given – the basics of proposed method for preprocessing of machine learning data and the main parameters of the proposed convolutional neural network model including its efficiency against reference model EfficientNetB0. In the work, the principles and the tool base for designing the interface of the software solution were defined, the scenarios for using the program, the algorithms of the program were designed, a prototype of the user interface was created.