ImTeX is a web application that converts images containing mathematical equations, tables, and other structured content into LaTeX code.
- intelligent structure detection: automatically detects equations, tables, lists, and other document structures
- document type optimization: specify document types (equations, tables, resumes, etc.) for improved conversion accuracy
- high-quality LaTeX output: generates well-formatted LaTeX code with appropriate packages
- detailed conversion insights: provides information about detected structures and conversion confidence
- frontend: next.js, typescript, tailwindCSS
- image processing: cloudinary for image storage and management
- OCR: mistral OCR 2503 for high-accuracy text and structure recognition
- AI: google gemini API for advanced document structure analysis
- LaTeX conversion: custom processing pipeline with specialized handling for different document types and pandoc integration
- node.js (v18 or higher)
- npm or yarn
- pandoc (for markdown to LaTeX conversion)
- mistral AI API key
- google gemini API key
- cloudinary
- clone the repository
git clone https://github.com/shamikhan005/ImTeX.git
cd ImTeX- install dependencies
npm install
# or
yarn install- create a
.envfile in the root directory with the following variables:
MISTRAL_API_KEY=your_mistral_api_key
GEMINI_API_KEY=your_gemini_api_key
CLOUDINARY_CLOUD_NAME=your_cloudinary_cloud_name
CLOUDINARY_API_KEY=your_cloudinary_api_key
CLOUDINARY_API_SECRET=your_cloudinary_api_secret
- start the development server
npm run dev
# or
yarn dev- Open http://localhost:3000 in your browser
- upload an image containing equations, tables, or other structured content using drag-and-drop or file browser
- select the appropriate document type for better conversion results (Auto-detect, Mathematical Equations, Table/Structured Data, Resume/CV, or General Document)
- click "Convert to LaTeX" to process the image
- view the generated LaTeX code, copy it to your clipboard, or download it as a .tex file
- use the "Show Details" button to see information about detected document structures and conversion confidence