Book Detection System At Bogor Library Using Teachable Machine

Abstract

This research aims to develop a book detection system at the Bogor Library using Teachable Machine technology, focusing on improving efficiency in automatically searching and identifying books. The system is designed to replace the barcode-based search method, which is considered less flexible, especially since users often experience difficulties in returning books to their original places after reading. Through the application of machine learning, users can detect books based on their cover images with high accuracy, without needing to adjust the barcode’s position. This research involves collecting book data by capturing images from various angles to train the machine learning model. The developed model was tested under various conditions, with results showing detection accuracy above 80%, meeting the research targets. The application was developed using Flutter, with an interface designed to facilitate users in accessing book scanning and search features. The test results show that the system can detect books with high accuracy and provide information about the location of books in the library, such as shelf numbers and floors. This system is expected to improve the efficiency of book management in the library and assist users in finding and returning books to the correct location.