Key points are not available for this paper at this time.
Modern mobile applications offer a standard user interface for a wide variety of tasks. An automatically generated adaptive user interface based on existing applications of a known category can be used to solve the problem of redundancy of applications and their data, and will also reduce the complexity of interface development. The user interface appearance provides the initial data for determining the application's category, searching for the desired function and creating the user interface. This paper discusses an approach to automating the analysis of the mobile devices' user interface to find features of mobile applications. Following the research, a dataset comprising 12 thousand images of the mobile application interface was curated for training the model. The paper also describes a convolutional neural network model for determining the mobile application's category based on the image of the user interface. The conclusion reveals the outcome of assessing the model's accuracy in predicting the category of mobile applications across both the training and testing datasets.
Vidmanov et al. (Thu,) studied this question.