- Implemented Neural Style Transfer techniques to create visually compelling images by merging the content of one image with arbitrary artistic style of another.
- Developed a Multi-level stylization approach using a pre-trained encoder-decoder architecture based on VGG19.
- Incorporated feature transformation with alpha as the style weight parameter to precisely control the stylistic transfer effects.
- Leveraged powerful techniques such as Adaptive Instance Normalization (AdaIN) and Whitening and Coloring transforms (WCT) within the feature transformation process for image reconstruction network.
- To deploy this project run the following file : 3Adain-2WCT.ipynb
- Make sure the content Image , Style Image and Output Image paths are clearly mentioned in the notebook.
- Example 1
- Example 2
- Example 3
Below are some research papers which have been instrumental in doing this project :
- Neural Style Transfer: A Review
- Arbitrary Style Transfer in Real-time with Adaptive Instance Normalization
- Universal Style Transfer via Feature Transforms
- A Neural Algorithm of Artistic Style
If you found value in this project, please give it a star! Your support is greatly appreciated.