Creating Artistic Images using Neural Style Transfer

The repository uses Neural Style Transfer. It takes as input two images: content image (the main input image), and style image (the filter). It then creates a third image using high level features (contents) of the content image and lower level features (styles) of the style image.

The detailed workflow is explained in style_transfer.ipynb. An in-depth analysis of the results and comparision with the Prisma app is also shown in images folder.

Summary of results:

Content	Style	Output

References

• [1] https://arxiv.org/abs/1508.06576
• [2] https://arxiv.org/abs/1603.08155
• [3] Deep Learning with Python by Francois Chollet

The files are created in Google Colab, and input and output images are fed through Google Drive. Link is included.