https://github.com/wikunia/he2latex

Converting handwritten equations to LaTeX
https://github.com/wikunia/he2latex

convolutional-neural-networks ocr-recognition sequence-to-sequence

Last synced: 12 months ago
JSON representation

Converting handwritten equations to LaTeX

• Host: GitHub
• URL: https://github.com/wikunia/he2latex
• Owner: Wikunia
• Created: 2018-07-27T14:13:42.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2022-07-13T15:46:39.000Z (over 3 years ago)
• Last Synced: 2025-02-27T15:46:32.942Z (about 1 year ago)
• Topics: convolutional-neural-networks, ocr-recognition, sequence-to-sequence
• Language: Jupyter Notebook
• Homepage: http://opensourc.es/blog/he2latex
• Size: 43 MB
• Stars: 111
• Watchers: 10
• Forks: 36
• Open Issues: 5
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# HE2LaTeX

**Not supported anymore**

I can't support this project anymore with the necessary download files as they are too huge for my small server.

The main goal of this repository is to provide a framework of constructing LaTeX equations out of handwritten images.

Converting this:

![Handwritten equation](report/formula.png)

to:

```

$\frac{977}{312}\lambda^3-580 \neq 458$

```

A short test can be done using the jupyter notebook [short_test.ipynb](short_test.ipynb).

For downloading all data needed for training run:

```

python download.py

```

The sequence to sequence model can be trained using the jupyter notebook [Seq2Seq.ipynb](Seq2Seq.ipynb) where also the model used is shown.

If the model itself changed the file `Seq2SeqModel/Seq2SeqModel.py` has to be changed as well.