Skip to content
helloztingPublic

About

No description, website, or topics provided.

Resources

Stars

2 stars

Watchers

2 watching

Forks

Latest commit

 

History

14 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MagiceGeo
Training-Free Text-Guided Geometric Diagram Generation

MagicGeo is a training-free framework designed for generating geometric diagrams from textual descriptions. Creating precise diagrams directly from text can be challenging, but MagicGeo simplifies the process by formulating diagram generation as a coordinate optimization problem, integrating large language models for text translation and a formal language solver to ensure geometric correctness. MagicGeo’s coordinate-aware generation effectively handles spatial relationships in text-to-diagram tasks, enabling accurate and efficient geometric diagram synthesis.

Framework of MagicGeo

Figure 1. Framework of MagicGeo

Performance

example

Figure 2. example

Framework of MagiceGeo Framework of MagiceGeo

Table of Contents

Install

pip install -r requirements.txt

Usage

As long as the required dependencies are installed, using MagiceGeo to generate, compile, render, and save TikZ drawings is straightforward.

get your own api

    client = OpenAI(
        api_key="***",
        base_url="***",
    )

Clone this project

    git clone https://link-to-project

Go to project directory

    cd geo

Startup server

    python text_to_geometric.py

Datasets

  • circle: json/circle.json
  • triangle: json/triangle.json
  • quadrangle: json/quadrangle.json

Citation

If MagiceGeo has been beneficial for your research or applications, we kindly request you to acknowledge its use by citing it as follows:

@article{wang2025magicgeo,
  title={MagicGeo: Training-Free Text-Guided Geometric Diagram Generation},
  author={Wang, Junxiao and Zhang, Ting and Yu, Heng and Wang, Jingdong and Huang, Hua},
  journal={arXiv preprint arXiv:2502.13855},
  year={2025}
}

Acknowledgments

The implementation of our model is largely based on DeepSeek.

Maintainers

@wjx

About

No description, website, or topics provided.

Resources

Stars

2 stars

Watchers

2 watching

Forks

Releases

Packages

Contributors

Languages