Cover-based book type deep learning classification method

Abstract

The invention relates to the technical field of book management, in particular to a cover-based book type deep learning classification method. The method comprises the following steps: S1, collecting book cover images and making a training data set DTP; s2, inputting the training data set DTP into the deep learning model and training; and S3, classifying books by using the deep learning model obtained in the S2. In the step S2, the deep learning model comprises two deep learning networks, a full connection layer and a softmax classifier, the vision-based deep learning network uses a residual neural network model, and the text-based deep learning network uses a general sentence encoder model. The method has higher classification accuracy and higher applicability and stability, books can be quickly and effectively classified, the precision is high, and the speed is high. The method is mainly applied to book type classification.

Description

technical field [0001] The present invention relates to the technical field of book management, and more specifically, to a cover-based deep learning classification method for book types. Background technique [0002] Book covers are often the reader's first impression, and they often convey important information about the book's contents. With the continuous advancement of the complete digitization of books, book type classification based on cover is an extremely important content for many modern retrieval systems. The main difficulties in solving this problem are as follows: first, there are many types of books, many of which are not specifically defined; second, the cover of the book, such as color, style, text information, etc. Third, due to the influence of external factors such as country, culture, target reader group, etc., the cover design of books will also be different. With the increasing competitiveness of the book industry, in order to attract readers, book co...

AI Technical Summary

Problems solved by technology

The disadvantage of this invention is: this invention is only for textual information recognition of books, and does not directly classify books. After the text is recognized, it still needs to manually classify books, so this invention has great limitations in actual use.

Images