Image Search API

This project implements an image search API using FastAPI and Qdrant for vector similarity search. It allows users to upload an image and find similar images from a pre-indexed collection.

Features

• Image similarity search using vector embeddings
• FastAPI backend for handling image upload and search requests
• Integration with Qdrant for efficient vector similarity search
• CORS middleware for cross-origin requests
• Static file serving for frontend assets

Prerequisites

• Python 3.11+
• Docker (optional, for containerization)

Installation

1. Clone the repository:

git clone <repository-url>
cd <project-directory>

2. Install the required dependencies:

pip install -r requirements.txt

Configuration

The project uses environment variables for configuration. You can set these in a .env file or directly in your environment:

QDRANT_URL=http://localhost:6333/
QDRANT_API_KEY=your_api_key
COLLECTION_NAME=nocnoc
EMBEDDINGS_MODEL=Qdrant/clip-ViT-B-32-vision

Usage

1. Start the FastAPI server:

python main.py

2. The API will be available at http://localhost:8000.

3. To search for similar images, send a POST request to /image/search with the image file in the request body.

API Endpoints

• POST /image/search: Upload an image to search for similar images in the collection.

Docker Support

A Dockerfile is provided for containerization. To build and run the Docker image:

docker build -t image-search-api .
docker run -p 8000:8000 image-search-api

Project Structure

• main.py: FastAPI application and main entry point
• config.py: Configuration settings and environment variables
• image_searcher.py: Image search functionality using Qdrant
• requirements.txt: Python dependencies
• Dockerfile: Docker configuration for containerization
• static/: Directory for static frontend assets (if applicable)

Data Processing

The project includes a Jupyter notebook (image_search_processing.ipynb) for data preprocessing and indexing. This notebook demonstrates how to:

1. Load and preprocess image data
2. Generate image embeddings using a pre-trained model
3. Index the embeddings in Qdrant for efficient similarity search

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

[Specify your license here]