DescraPix

This project uses OpenAI's CLIP model to find images in a folder that match a user provided text description. It can analyze a folder of images, compute their similarity to a given description, and display or save the best matches.

Technologies Used

1. Streamlit: For the web application interface.
2. PyTorch: For leveraging the CLIP model.
3. Hugging Face Transformers: To load the pretrained CLIP model and processor.
4. Pillow (PIL): For image processing.
5. shutil: For file copying and management.

Future Improvements

1. Add support for more image formats (e.g., GIFs, BMP).

2. Enhance the user interface with drag-and-drop functionality for folder selection.

3. Integrate GPU acceleration for faster image processing.

4. Extend the application to process videos by extracting frames and performing similarity checks.

Contact

For any inquiries or feedback, please contact:

Email: vishnumkpn@gmail.com

GitHub: VishnuMuraliKr