OCR-Math

This project depends on Pix2Text for formula recognitions and TesseractOCR for text-only recognitions.

An OCR application to extract text and/or mathematical expressions from images, which can be exported in .docx format for easy copy paste process.

They can also be copied one by one and paste into MS Words as Equation/Text.

Environments:

• Ubuntu 24.04.1 LTS
• Python 3.12
• Node v23.10.0

How to Setup:

1. Clone this repository

git clone https://github.com/GangYiKhor/math-ocr.git
cd math-ocr

2. Linux prerequisites

sudo apt-get update
sudo apt-get install gcc
sudo apt-get install protobuf-compiler libprotoc-dev

3. Install Python and packages

sudo add-apt-repository ppa:deadsnakes/ppa
sudo apt install python3.12 python3.12-dev
sudo apt install pipx
pipx ensurepath
pipx install virtualenv
# May need to restart for virtualenv to be in PATH

virtualenv -p /usr/bin/python3.12 .venv
source .venv/bin/activate

# Install ONNX
export CMAKE_ARGS="-DONNX_USE_PROTOBUF_SHARED_LIBS=ON"
pip install onnx

# Optional, only install CPU version of PyTorch
pip install -r requirements-pytorch-cpu-linux.txt

pip install -r requirements.txt

4. Install NVM, Node and modules

curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.2/install.sh | bash
export NVM_DIR="$HOME/.nvm"
[ -s "$NVM_DIR/nvm.sh" ] && \. "$NVM_DIR/nvm.sh"
[ -s "$NVM_DIR/bash_completion" ] && \. "$NVM_DIR/bash_completion"

nvm install
nvm use
npm --prefix ./frontend install

5. Install Tesseract and languages

# Refer to https://tesseract-ocr.github.io/tessdoc/Installation.html for Windows
sudo apt install tesseract-ocr

# Download Malay language
wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/refs/heads/main/msa.traineddata
sudo mv msa.traineddata /usr/share/tesseract-ocr/5/tessdata/

6. Create .env File

cp .env.template .env
# Update env if needed

7. Create user (Optional if WITH_AUTH is false)

python backend/createsuperuser.py

8. User management (Optional if WITH_AUTH is false)

Change pasword

python backend/changepassword.py

Activate user accounts

python backend/activateuser.py

9. Run development server

• Two servers required
• You may also run debug in VSCode for both servers
• Server will be started at http://localhost:8000

# Python server
fastapi dev backend/main.py

# Vue server
npm --prefix ./frontend run dev

How to Run Production Server

Run Makefile script

make run/prod

=== OR ===

1. Build frontend

npm --prefix ./frontend run build

2. Run FastAPI server

python startserver.py