🖼️ Image Comparison API

A FastAPI-based microservice that compares two base64-encoded images using Google's Gemini (Generative AI) and returns a similarity score along with an explanation of differences and similarities.

🚀 Features

• Accepts two images in base64 format.
• Uses Google Gemini (gemini-1.5-pro) to analyze and compare the images.
• Returns:
  • Similarity Score (0 to 100)
  • Human-readable explanation of the differences and similarities.

🧠 Powered By

• FastAPI
• Google Generative AI
• Pillow (PIL)

📦 Installation

1. Clone the Repository

git clone https://github.com/raks-javac/image-comparison-api.git
cd image-comparison-api

2. Create a virtual environment and install dependencies

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

3. Set up Environment Variables

Create a .env file in the root directory and add your Google API key:

GOOGLE_API_KEY=your_google_api_key_here

4. Run the App

uvicorn main:app --reload

The app will be available at: http://localhost:8000

🛠️ API Endpoints

GET /

Check if the API is running.

Response:

{
  "message": "Image Comparison API is running"
}

POST /compare

Compares two base64-encoded images.

Request Body:

{
  "image1": "data:image/png;base64,iVBORw0K...",
  "image2": "data:image/png;base64,iVBORw0K..."
}

Response:

{
  "similarity_score": 87.5,
  "explanation": "Both images share similar structure and content, but differ slightly in color and background elements."
}

📁 Project Structure

.
├── main.py                     # Main FastAPI app
├── models/
│   └── image_models.py         # Pydantic request/response models
├── .env                        # Environment variables
└── requirements.txt            # Python dependencies

✅ Requirements

• Python 3.8+
• Valid Google API key with access to Gemini 1.5 Pro
• Dependencies listed in requirements.txt:
  • fastapi
  • python-dotenv
  • google-generativeai
  • pillow
  • uvicorn

🤝 Contributing

Contributions are welcome! Please fork the repo and submit a pull request.

📄 License

This project is licensed under the MIT License.

🙋‍♂️ Author

Your Name
GitHub: @Raks-Javac