Research And Design Of Library Shelf Order Intelligent Recognition System Based On Deep Learning

With the development of the information age,in order to meet the increasing knowledge needs of readers,the service concept of libraries has changed from "literature-oriented" to "people-oriented",and the borrowing mode has also changed from "closed-shelf borrowing" to "open-shelf borrowing".These changes provide great convenience for readers' borrowing and other needs,but due to the irregularity of readers' borrowing and returning books or the inadequate management of Library managers,the phenomenon of book disorder often occurs,which brings great inconvenience to the circulation of books and readers' borrowing.At first,the library solved the problem of disorderly shelves mainly by strengthening the quality education of readers and increasing the number of librarians visiting libraries,but the efficiency is very low.After the introduction of Radio Frequency Identification(RFID),managers can scan bookshelves with hand-held devices to judge whether there are books in disorder,and then the human eye can find books in disorder.Although the efficiency of book in disorder is improved,there are still some shortcomings such as time-consuming and laborious.Deep learning is very successful in the field of image recognition.It can simulate human eye recognition image in many scenarios.Therefore,based on RFID technology,this paper designs an intelligent bookshelf order recognition system based on in-depth learning,which can greatly improve the efficiency of disordered bookshelf collation,and has very important practical significance.This system scans bookshelves according to RFID to determine whether there are disordered books and records information.If there are disorderly shelves,take photos of the shelves,and then use in-depth learning to identify all the books in the pictures,and match the identified call numbers with the recorded disorderly shelves,so as to achieve the accurate positioning of disorderly shelves.The main work of this paper is as follows:(1)Introducing the related technology of designing library frame order intelligent recognition system,comparing and optimizing the algorithms,choosing the technology used in this paper;(2)The overall design of Library frame order intelligent recognition system and the design of main functional modules;(3)The design of image recognition module of the system,under the framework of TensorFlow,advances the classical LeNet5 model.This paper improves the convolution neural network model and trains the model.(4)Taking Guizhou University of Finance and Economics Library as an example,a data set of call number characters is established by using image segmentation technology.