Semantic Image Search Engine

This project leverages PyTorch and OpenAI CLIP for advanced text-based semantic image search capabilities. The backend is built using FastAPI, ensuring a robust and high-performance server, while the frontend is developed with Next.js, providing a seamless and responsive user experience.

Features

• Text-based semantic image search using OpenAI CLIP
• Frontend built with React (Next.js) and styled with Tailwind CSS
• Backend built with FastAPI and powered by PyTorch
• Preloaded random images for demonstration purposes

Technologies Used

• Backend: FastAPI, PyTorch, Hugging Face Transformers
• Frontend: Next.js, Tailwind CSS, React
• Other: Node.js, Python, OpenAI CLIP

Prerequisites

Before running the project, make sure you have the following installed:

• Python 3.10+: Download
• Node.js: Download
• Unsplash Developer Access Key (optional, for fetching new random images): Get Access Key

Installation

Clone the Repo

git clone https://github.com/your-username/your-repo.git
cd your-repo

Backend Setup

1. Navigate to the backend directory:

   cd backend

2. Create a virtual environment:

   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

4. Start the FastAPI server:

   uvicorn call:app --reload

   By default, the server will run at http://127.0.0.1:8000.

Frontend Setup

1. Navigate to the frontend directory:

   cd frontend/semantic-image-search

2. Install dependencies:

   npm install

3. Start the development server:

   npm run dev

   The frontend will be available at http://localhost:3000.

Using Different Images

If you’d like to use different images for the project:

1. Replace the images in the random-images folder:
   • Use the provided script to fetch random images from Unsplash:

     python backend/get-random-images.py

     Note: You need an Unsplash Developer Access Key to use the script. Add it to the script by replacing the placeholder ACCESS_KEY with your access key.
2. Alternatively, manually replace the images in the random-images folder with your own collection.

Usage

1. Open the frontend in your browser: http://localhost:3000.
2. Search for images using text queries such as "beautiful sunset," "expansive ocean," or "snowy mountains."
3. Results will display a set of images matching the query from the preloaded random dataset.