New Plugin: Latex OCR!

Share & showcase
1

Obsidian Latex OCR allows you to generate Latex equations from images in your clipboard or vault!

demo

  • This project requires installing an accompanying python package; see the plugin’s installation instructions for more information.
  • Currently this plugin is in beta; please let me know if you run into any issues, or would like to see additional features!
6 Likes
2

HI! For the past few days, I’ve been getting only part of the LaTeX code from the whole expression that is in the image.

Like this:

b_{1}(x)=a r g m i n\frac{
\partial^{3+}\mathrm{b^{3}}=(
({a}+{b})^{3}={a}^
{\mathrm{a}}^{3+}{\mathrm{b}

It’s like something interrupts it at a certain point. What could be the problem? I use HuggingFase model.

3

I’m seeing the same issue; I think the problem is with HuggingFace as using the inference API on the website isn’t returning the full text either. Apparently there were some issues earlier today, this might still be a result of that: https://twitter.com/huggingface/status/1756401140495679871. Using the model locally still seems to work fine as an alternative, but I’ll keep investigating.

1 Like
4

I am also having this issue, and am using the HuggingFace model.

1 Like
6

Thank you for that wonderful plugin, it does save me a lot of time.

Is there a way to shut down the locally running inference server ? It is consuming too much memory to keep it around on my machine.

7

My current workaround is to kill the process that uses the port specified in the plugin settings with kill $(lsof -ti :$PORT) on Linux (replace $PORT with the actual port).