Evaluation for Optical Character Recognition of Mobile Application

The library's ancient books are deteriorating, due to faded ink, and damaged paper. It is challenging for librarians to both manage and discover information. Therefore, a technique to overcome this is required, such as implementing text recognition by leveraging many IOT -based OCR apps in mobile phones, such as Text Fairy and Camscanner to find out the results of the effectiveness of OCR mobile apps. Accuracy calculations are needed, in this case, the use of CER and WER is implemented to determine the accuracy. From the results of CER and WER calculations, it was found that in book cover data, Camscanner comes out better than Text Fairy. However, on data that is entirely text, Text Fairy is better when compared to Camscanner. It can be concluded that Text Fairy is superior in recognizing text-only parts, whilst Camscanner is superior at recognizing parts that are not entirely text.