This project implements a neural style transfer algorithm as described in the seminal paper "Image Style Transfer Using Convolutional Neural Networks" by Gatys et al. It uses PyTorch and a pre-trained VGG19 network to transfer the artistic style from one image (the style image) onto the content of another image (the content image), effectively merging these aspects to create a new, stylistically altered image.
- VGG19 for Feature Extraction: Utilizes the VGG19 model, a deep convolutional neural network pre-trained on the ImageNet dataset, for extracting style and content features from images.
- Style Transfer: Combines the content of one image with the style of another using the technique of Gatys et al.
- Gram Matrix Representation: Employs Gram matrices to capture and apply artistic styles from the style image to the content image.
- Flexible Image Inputs: Supports image inputs from various sources, including URLs and local files.
- Customizable Parameters: Allows tweaking of style weights, content weights, and other parameters to explore different styles.
To get started with this project, you'll need to set up your Python environment and install the required libraries.
Python 3.x PyTorch Torchvision PIL (Python Imaging Library) NumPy Matplotlib
###steps
- Clone the Repository
git clone https://github.com/your-username/style-transfer.git cd style-transfer
pip install torch torchvision numpy matplotlib pillow
## Usage
To perform style transfer using this model, follow these steps:
# 1. Navigate to the Project Directory
cd path/to/style-transfer
# 2. Run the Notebook
jupyter notebook style_transfer.ipynb
Follow the instructions in the notebook to input your content and style images, and observe the style transfer process.
## Contributing
Contributions to this project are welcome. Please fork the repository and submit a pull request with your proposed changes.
## License
This project is open-source and available under the [MIT License](LICENSE).