unimernet 0.0.4

UniMERNet: A Universal Network for Real-World Mathematical Expression Recognition

UniMERNet: A Universal Network for Real-World Mathematical Expression Recognition

This is the official repository for UniMERNet, a math recogition model that can be used for image to LaTeX conversion for a wide range of senarios.

Project page: https://gitlab.pjlab.org.cn/fdc/mllm/unimernet

Installation

conda create -n unimernet python=3.10

conda activate unimernet

pip install unimernet

For Mac

brew install freetype imagemagick
export MAGICK_HOME=/opt/homebrew/opt/imagemagick

Quickstart

python demo.py

or

jupyter-lab ./demo.ipynb

Training

To train or finetune UniMERNet model, run

torchrun --nproc-per-node 4 --master_port 29500 train.py --cfg-path configs/unimernet_train.yaml

or

bash scripts/train.sh

Evaluation

To evalate the model, run

python test.py --cfg configs/unimernet_eval.yaml

Performance Comparison (BLEU) with SOTA Methods.

Visualization Result with Different Methods.

Citation

If you find our models / code / papers useful in your research, please consider giving ⭐ and citations 📝, thx :)

Acknowledgement

• VIGC. This repository is built upon VIGC!
• Texify.
• Latex-OCR. The original open source Latex OCR project.
• Donut.
• Nougat.

Contact us

If you have any questions, comments or suggestions, please do not hesitate to contact us at [e-mail].

License

Apache License 2.0