DF Image Search 🦆

DF Image Search is a powerful, locally-hosted AI image search engine utilizing OpenAI's CLIP (via OpenCLIP) to allow you to search your local image collections using natural language or image-to-image queries.

(DF Image Search Interface - Indexing and Searching)

✨ Features

• Natural Language Search: "A dog playing in the snow", "Cyberpunk city at night".
• Image-to-Image Search: Find similar images by dragging and dropping a reference image.
• Privacy First: Everything runs locally on your machine. No images are uploaded to the cloud.
• GPU Acceleration: Leveraging PyTorch and CUDA for blazing fast indexing and searching.
• Modern Web UI: A sleek, dark-themed interface with glassmorphism design.
• File Management: Open, Copy, Move, and Delete files directly from the search results.
• Progress Tracking: Real-time progress bar for image indexing.

🚀 Installation

1. Clone the repository:

   git clone https://github.com/yourusername/df-image-search.git
   cd df-image-search

2. Install Dependencies: Recommendation: Use a virtual environment (venv or conda).

   pip install -r requirements.txt

   Note: You need to have PyTorch installed with CUDA support if you want GPU acceleration. Visit pytorch.org for the command specific to your system.

🦆 Usage

1. Start the Server: Simply double-click start_df_search.bat (Windows)

   OR run via terminal:

   python web_app.py

2. Index Your Images:

   • The Web UI will open automatically in your browser (usually http://localhost:8000).
   • Click on the Gear/Settings Icon.
   • Enter the full path to your image folder.
   • Click Start Indexing.
   • Wait for the progress bar to complete.

3. Search:

   • Type a description in the text box or drag an image onto the drop zone.
   • Enjoy exploring your collection!

🛠 Tech Stack

• Backend: Python, FastAPI, PyTorch, OpenCLIP
• Frontend: Vanilla JavaScript, CSS3 (Glassmorphism), HTML5
• Model: ViT-B-32 (default), supports swapping to other OpenCLIP models.

🤝 Contributing

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

📄 License

This project is open-source and available under the MIT License.