README.md

metadata

title: Image Data Extractor
emoji: 🧾
colorFrom: blue
colorTo: indigo
sdk: gradio
sdk_version: 6.20.0
python_version: '3.12'
app_file: app.py
pinned: false
models:
  - numind/NuExtract3
short_description: Extract structured JSON from documents with NuExtract3.

Image Data Extractor

Image Data Extractor is a Gradio app for extracting structured JSON from document images with NuExtract3.

Upload a receipt, invoice, or other document image, optionally add accompanying text, and choose a JSON template for schema-guided structured extraction. You can also generate a grounded template from the uploaded image before running extraction.

The inference function uses Hugging Face ZeroGPU dynamic GPU allocation. Select ZeroGPU in the Space hardware settings before launching the app.

Run locally

Install the project dependencies and launch the Hugging Face Spaces entry point:

uv sync
uv run app.py

The Hugging Face Space starts from app.py. Model loading and the ZeroGPU pipelines are defined in model.py.