Handwritten mathematical expression recognition : ScanIt a web based A\application

With the increasing use of computational devices in our everyday lives, more and more tasks have transitioned to an easier methodology, using these devices. However, this is not the case for every single task. Some, like inputting mathematical expressions, have become even more tedious than they were to write out by hand. The goal of this project was to develop a web-based application that can help decrease the arduous nature of this task. The proposed solution implements a client-server architecture in conjunction with a strong convolutional neural network to produce an accessible and intuitive web-based application that accepts user images and returns the LaTeX notation detected. To implement this system, these tasks were conducted: - A literature review and research to determine the optimal model to be used for this use case. - Ported the model into a Flask app, which can be accessed via an API call. - Designed and Developed a user interface through HTML5, CSS and Bootstrap, that follows Schneiderman’s 8 golden rules of a good interface. - Developed a backend server via Node.JS and Express that seamlessly controls the logic and routing of the web application. - Made use of database system MongoDB to store schemas necessary for the application. - Used cloud SaaS provider Cloudinary to host user uploaded images. - Saves user submissions such that they can be accessible anytime, anywhere, and provides options to edit in the case of errors. - Conduct testing along the development process to ensure that any inconsistencies and bugs are eliminated as soon as they are discovered. Through these tasks, the goal of the project was achieved, and initial User Acceptance Testing has provided positive feedback. Bachelor of Engineering (Computer Science