[Google_Bootcamp_Day23]

Neural style transfer

Cost function

Content cost function

• Assume you use hidden layer l to compute content cost (choose middle layer l)
• By using pre-trained ConvNet, forward propagate image C and image G.
• Let a(l)(c) and a(l)(G) be the activation of layer l on the images
• If a(l)(C) and a(l)(G) are similar, both images have similar content.

Style cost function

• Assume you use layer l’s activation to measure “style”
• Define style as correlation between activations across channels (compute how correlated are the activations across different channels in each positions)
• What does it mean when two channels are highly correlated?
  • correlation tells you which of these high level texture components tend to occur or not occur together in part of an image
  • if highly correlated, then high level texture components occur together in part of an image
• this gives you a measure of how similar is the style of the generated image to the style of the input style image

Style matrix

Final Style cost function

[source] https://www.coursera.org/learn/convolutional-neural-networks