📄 AI Document Digitizer (Qwen3-VL)

An intelligent document processing API & UI powered by the Qwen3-VL multimodal model. It extracts structured data from scanned forms, invoices, and documents, ignoring boilerplate text and focusing on business values.

🚀 Features

• Strict Data Extraction: Ignores legal text and instructions, extracts only values.
• Smart Formatting: Converts tables and forms into structured JSON.
• Visual Verification: Interactive UI highlights extracted fields on the image.
• Multiple Exports: Download results as JSON, CSV, or Excel.
• Dual Mode: Works as a Web UI (Gradio) and a REST API (FastAPI) simultaneously.

🛠️ Installation

1. Clone the repository:

   git clone https://github.com/ituvtu/qwen-doc-parser.git
   cd qwen-doc-parser

2. Сreate a virtual environment and install dependencies:

   python -m venv .venv
   # Windows:
   .venv\\Scripts\\activate
   # Mac/Linux:
   source .venv/bin/activate
   pip install -r requirements.txt

3. Set up environment variables: Copy .env.example to .env and add your Hugging Face Token:

   HF_TOKEN=your_token_here

4. ▶️ Usage

   Run with Docker (Recommended)

   Option A: Using .env file (Best for security)

   docker build -t qwen-doc-parser .
   docker run -p 7860:7860 --env-file .env qwen-doc-parser

   Option B: Passing token directly

   docker run -p 7860:7860 -e HF_TOKEN=hf_YourTokenHere qwen-doc-parser

   Run Locally

   uvicorn app.main:app --host 0.0.0.0 --port 7860 --reload

   Open your browser at http://localhost:7860.

📡 API Example

You can use the API to extract data programmatically:

curl -X POST "http://localhost:7860/api/v1/extract" \\
     -H "accept: application/json" \\
     -H "Content-Type: multipart/form-data" \\
     -F "file=@/path/to/invoice.jpg"

🧪 Quick Test Script

The project includes a Python script to verify the API functionality immediately.

1. Open test_api.py and update the IMAGE_PATH variable to point to your test image.
2. Run the script:

   python test_api.py