Development of a Web-based Optical Character Recognition System

Abstract

Optical Character Recognition Technology, OCR in short, is one of the
most crucial field of research in improving and forwarding machine learning and AI
technology. The proposed system is a Python language developed web application
to deliver one of the simpler and general use of such technology to solve daily
problems, which is to convert texts and character in image into editable text insert
into document files. The system integrates authentication system with login and
register to track, record and store their uploaded data into the system’s database.
The main function is an OCR function that accepts image files from users to convert
them into document files where the users can then download it from the server,
powered by Tesseract, an open source and free command line program with an OCR
engine, and can be commanded in Python with the Pytesseract library. Users can
send feedback and help with the learning and training of the AI by uploading
multiple files as training sample in the feedback function. The system allows editing
of profile credentials through the user profile page. The end result of the project is a
web app that focus primarily on the OCR function, where users can directly and
indirectly help with the improving and learning of the AI. Although still having
some flaws including the download function not working well with client devices,
further improvements and recommendations have been suggested to better make the
system works in future.