Inspiration
Oftentimes, our problem sets are given to us as PDF's. In order to submit our homeworks, which must contain problem statements, someone (usually a friend) spends a lot of time putting the problem statements in LaTeX.
What it does
Uses GoogleVision API as well as MathPix API to read the PDF and eventually output LaTeX code for the PDF.
How we built it
The GoogleVision API finds bounding boxes for our text. We then determine if the bounding box contains words or math. We take the math boxes and use the MathPix API to convert the math to LaTeX.
Challenges we ran into
Neither of us have any experience with either API and MathPix is very particular about the input bounding boxes. The GoogleAPI bounding boxes were difficult to accurately convert into the format needed for MathPix.
Accomplishments that we're proud of
We learned how to understand and utilize an API with no prior experience or understand (particularly API's with poor documentation). We are able to get each individual components to work despite compatibility errors.
We got a lot done for only having 2 people with very little experience on the team.
What we learned
Not all API's will work with each other. Just because an output could potentially work as an input doesn't mean the new API will accept it.
What's next for MITHacks
Hopefully, we can get this working. Possibly with a neural net that has the functionality of both of these API's. Or figure out a way to make the current API's work together.
Log in or sign up for Devpost to join the conversation.