GemmaOCR-APP 🔎

A polished, local-first OCR web app powered by Gemma-3 Vision and Streamlit.

GemmaOCR-APP lets you upload an image, run OCR with a multimodal LLM, and receive well-structured Markdown output (headings, lists, code blocks, etc.)—all in a clean interface.

---

✨ Features

• Local OCR pipeline with Ollama + gemma3:12b
• Simple Streamlit UI with:
  • image upload (png, jpg, jpeg)
  • one-click text extraction
  • clear/reset action
• Structured output, not plain text dump
• Runs on your machine (no external OCR SaaS required)

---

🧱 Tech Stack

• Python
• Streamlit (UI)
• Ollama Python client (model inference)
• Gemma-3 Vision model (gemma3:12b)
• Pillow (image handling)

---

📂 Project Structure

GemmaOCR-APP/
├── app.py        # Streamlit application
├── README.md     # Project documentation
└── LICENSE       # License file

Note: app.py references ./assets/gemma3.png for the header icon. Add this file if it is missing in your local copy.

---

✅ Prerequisites

Before running the app, ensure:

1. Python 3.9+ is installed
2. Ollama is installed and running
3. The Gemma model is pulled locally

ollama pull gemma3:12b

---

🚀 Quick Start

1) Clone the repository

git clone https://github.com/<your-username>/GemmaOCR-APP.git
cd GemmaOCR-APP

2) Create a virtual environment

python -m venv .venv
source .venv/bin/activate    # macOS/Linux
# .venv\Scripts\activate     # Windows PowerShell

3) Install dependencies

pip install streamlit ollama pillow

4) Run the app

streamlit run app.py

Then open the local URL shown by Streamlit (usually http://localhost:8501).

---

🖼️ How It Works

1. Upload an image from the sidebar.
2. Click Extract Text 🔍.
3. The app sends your image and OCR prompt to gemma3:12b via Ollama.
4. The model response is rendered as Markdown in the main panel.
5. Click Clear 🗑️ to reset current output.

---

🔐 Security & Credentials

This project is designed to run locally and does not require usernames, passwords, or API keys by default.

For security reasons, do not place real personal credentials in source code, README files, or Git history.

If you later add external integrations (e.g., cloud storage, paid APIs), use environment variables and a .env file that is excluded from version control.

---

🛠️ Troubleshooting

Error processing image: ...

• Confirm Ollama is running:

  ollama list

• Ensure gemma3:12b is installed:

  ollama pull gemma3:12b

• Re-launch Streamlit after model download.

Missing icon error (./assets/gemma3.png)

Create assets/ and add gemma3.png, or adjust the image path in app.py.

---

🗺️ Roadmap Ideas

• Batch OCR for multiple images
• Download results as .md / .txt
• Bounding-box OCR visualization overlay
• Language selection and prompt presets
• Dockerized deployment profile

---

🤝 Contributing

Contributions are welcome.

1. Fork the repo
2. Create a feature branch
3. Commit changes
4. Open a pull request with a clear description

---

📄 License

This project is licensed under the terms in LICENSE.