"In math, most students usually handwrite their own equations. Its very convenient, as with just a pen and paper, you can solve almost all math problems. However, what if you wanted to show it to someone? What if you wanted to send it across the web? What if you were supposed to present it to your class? Many students may just take a picture of their handwritten work, but it may sometimes be messy, especially if the student wrote down the math too fast. Some students like to type out their own math, be it with LaTeX or some rich text editor. Although it is good to look at and oftentimes the teacher's preferred option, it is time consuming and tedious, having to write down the equations twice. 

Hence, a viable option might be to use an OCR programme designed for math. OCR(Optical character recognition) is the process of transforming an image of a text into text that can be edited and read by a computer. This process is used in a lot of functions to allow computers to “read”, some examples including Preview on MacOS and Google Lens on android phones. Implementation of these algorithms into commonly used apps has been revolutionary, and made very convenient and efficient for processing from all fields of work. Not only has OCR been used to automatically classify cancer based on pathology reports, with a word accuracy of up to 98% (Zuccon et al, 2012), it has also been used to read music on printed music sheets to convert them into a digital, playable form (Singh et. al, 2012). Given this, there are already some solutions that help to convert written maths into machine-readable maths.

One powerful and popular solution would be google lens. Not only can it recognise text from images, it can also do this recognition in real time, giving the user more options to use it. It is also very fast, returning results in a few seconds. This makes it convenient for the user when the app is used for large amounts of workings, which can happen in some maths.

However, this app falls short in several areas. For example, when copying maths, the user is often directed directly to a google search, and it is hard for the user to copy the mathematics that was detected. The maths formulas copied are also not in a format which would be natural for someone to write down, like LaTeX. It instead uses signs like carots(instead of superscript) to represent exponentiation, slashes to represent division, and so on. Although this is not unconventional, it can hinder presentation, as people take longer to read the maths which is represented in a more unfamiliar style.

There are also applications like MathPix which can translate handwritten maths into LaTeX code quickly and efficiently. It only has a 6.88% error rate for handwritten characters(Devi and Chinmayee, n.d.). It can translate handwritten maths into LaTeX form in a short period, and gives multiple options to export the maths that was read by the application. However, MathPix has most of its features hidden behind a paywall, making it inaccessible for most students, who would not pay a subscription just to use this service.

Our app solves many of the problems highlighted above, and we fulfilled a majority of the design criteria we set out to achieve. Our app has a very high accuracy in transcribing math up to the secondary level, possibly even being able to transcribe Junior College or university level math. It is quite fast, being able to transcribe equations in most cases under 10 seconds. This makes it time efficient for the user. Users can upload their photos from many different channels, including from google drive and directly from the webcam. Once it is transcribe, we also provide many options for the user for them to extract their equations, from the LaTeX code itself to PDFs and PNGs resulting from the compilation of the code. It is free and does not have download requirements from the user, except maybe a web browser. This makes it runnable without hassle, straightaway."

Developed by: Tian Cheng and Samuel