Zhen Latex OCR

🚀 MixTeX

MixTeX is a project that provides image - to - text functionality, including a win client for local offline CPU inference and a model for image - based text extraction.

🚀 Quick Start

The win client local offline CPU inference exe has been released. You can download it on GitHub. Alternatively, you can directly download the zip file in the "Files" section, unzip it, and then run the exe file directly (no installation required).

Github: MixTeX Repository

💻 Usage Examples

Basic Usage

from transformers import AutoTokenizer, VisionEncoderDecoderModel, AutoImageProcessor
from PIL import Image
import requests

feature_extractor = AutoImageProcessor.from_pretrained("MixTex/ZhEn-Latex-OCR")
tokenizer = AutoTokenizer.from_pretrained("MixTex/ZhEn-Latex-OCR", max_len=296)
model = VisionEncoderDecoderModel.from_pretrained("MixTex/ZhEn-Latex-OCR")

imgen = Image.open(requests.get('https://cdn-uploads.huggingface.co/production/uploads/62dbaade36292040577d2d4f/eOAym7FZDsjic_8ptsC-H.png', stream=True).raw)
#imgzh = Image.open(requests.get('https://cdn-uploads.huggingface.co/production/uploads/62dbaade36292040577d2d4f/m-oVg8dsQbQZ1fDWbwKtO.png', stream=True).raw)
print(tokenizer.decode(model.generate(feature_extractor(imgen, return_tensors="pt").pixel_values)[0]).replace('\\[','\\begin{align*}').replace('\\]','\\end{align*}'))

You can also try the code in Colab: MixTeX Colab Notebook

📚 Documentation

Dataset example

The author will release a small portion of the generated synthetic dataset and the data collection methods.

Supplementary information

💡 Usage Tip

• Suggested image dimensions: approximately (400, 500), (H, W).
• Recommended output length: approximately 100 - 300 tokens.
• LaTeX environment:

\usepackage{amssymb}
\usepackage{amsmath}
\usepackage{stmaryrd}
\usepackage{color}

📄 License

This project is licensed under the Apache - 2.0 license.