On Friday morning, I was desperately studying for my biology test. One of the upperclassmen had given us his notes for the class as a collection of handwritten images. Those notes saved me, but I wished I was able to search through them.

What it does

It takes a collection of images, extracts text and diagrams, and turns it into LaTeX code to make the result look very nice.

Nice features:

  • is able to create new sections based on the size of the writing alone
  • supports lists (\begin{itemize} in LaTeX), and it also supports nested listed. The program is able to detect indents.
  • supports diagrams via additional markup from user; small amount of markup inserts diagrams into the file

How I built it

I used Google Vision API and opencv.

Challenges I ran into

This was the first time I tried to do anything related to computer vision. Due to limitations (it seems) of the Google Vision API, my strategy for locating diagrams in the images changed very late into the project. Generally, it was quite difficult to do everything correctly.

Accomplishments that I'm proud of

What I learned

What's next for LaTeX-it

I would like to be able to render equations as well. I was planning on using the detexify open-source implementation to detect translate math characters into LaTeX, but I ran out of time.

Built With

Share this project: