OTTOMAN CHARACTER RECOGNITION ON PRINTED DOCUMENTS USING DEEP LEARNING

Abstract

In this study, a deep learning-based method is developed for character detection and recognition in printed Ottoman documents. The character detection and recognition problem are considered as an object detection problem and for this purpose, an Ottoman character recognition model is developed based on the YOLO model, which is one of the most successful methods in object detection. In addition, in this study, a dataset consisting of Ottoman document images is created in which each character in the document images is marked. Data augmentation techniques are applied to improve the accuracy of character recognition and the robustness of the method. The Ottoman character recognition network was then trained using this dataset. The trained network model was tested with the test images in the dataset. The performance evaluation of the model was performed by calculating the average precision metric, which is frequently used in the literature. The average precision value was calculated for 34 character classes in the dataset and the results were interpreted in terms of the pros and cons of the method. The results show that the proposed method can detect and recognize characters in printed Ottoman documents with great accuracy, with a weighted average precision of 98.71%.