Multimodal Search with Weaviate

This project demonstrates multimodal search capabilities using Weaviate, allowing for combined text and image search functionality.

Prerequisites

• Python 3.10 or higher
• Docker and Docker Compose
• uv package manager (recommended) or pip

Setup Instructions

1. Install uv (if not already installed)

# Install uv using pip
pip install uv

# Or using curl (recommended)
curl -LsSf https://astral.sh/uv/install.sh | sh

2. Initialize the Python Environment

# Install dependencies
uv sync

3. Start Weaviate Services

The project uses Weaviate with CLIP for multimodal vector search. Start the services using Docker Compose:

docker-compose up -d

This will start:

• Weaviate instance on port 8080
• CLIP inference API for image/text embeddings

4. Running the Notebook

1. Ensure your Python environment is activated
2. Launch Jupyter:

   jupyter notebook

3. Open create_collections.ipynb in your browser
4. Execute the notebook cells sequentially

Project Structure

• docker-compose.yml: Docker configuration for Weaviate and CLIP services
• create_collections.ipynb: Main notebook with code examples
• pyproject.toml: Python project dependencies

Troubleshooting

• If Weaviate services fail to start, ensure ports 8080 and 50051 are not in use
• Check Docker logs using: docker-compose logs
• Verify Weaviate is running: curl http://localhost:8080/v1/meta

Stopping the Services

To stop the Weaviate services:

docker-compose down