Skip to content

Latest commit

 

History

History
21 lines (15 loc) · 1.05 KB

README.md

File metadata and controls

21 lines (15 loc) · 1.05 KB

Neural Style Transfer

Neural Style Transfer model that combines a content image with a style reference image to yield artistic renders.

The result was achieved by optimizing a Style Content Model. In this model we preprocess the input on a pretrained VGG19 model (trained on imagenet), then extract the intermediate layer activations to obtain weight activations learned from our content image, followed by computing the style-content loss and updating our model weights to acheive a smoother combined image.

Example

Given as input content image and style reference image, respectively

with neural style transfer we can generate image

Neural Style Transfer Output

References