Identifing of Alphanumerical Codes in Promotional products by Using of Deep Neural Network

Abstract

Using the codes in promotional products was often considered a waste of time. For this reason most codes are not used and promotions do not show sufficient effectiveness. The theme of the project was to identify the promotional code on the product using artificial neural networks and deep learning methods. Bu projede, karakterlerin tanımlanması için, Keras ve Tensorflow kütüphaneleri kullanılarak Convolutional Neural Network (CNN) yapısında bir yapay sinir ağı kullanılmıştır. The project was created in the direction of the Optical Character Recognition application (OCR) requirement that emerged in a software for a customer company in adesso Turkey. The client company’s requirement is that a program to recognize characters with a special font. A mobile application has been implemented in the iOS environment to increase the efficiency and ease of implementation of the project. The OCR (Optical Character Recognition) library created in the project has been converted to Objective-C. Then the Objective-C library used in iOS program to use the Python model. In the system where the number of samples used in the training set is 7091, the model accuracy for 1200 test data is 99.7%.