Neural Style Transfer implemented in PyTorch.
NST is a peculiar application of CNNs that is used to impose the style of one image on a target image without distorting its contents.
Many photo editing apps use this model in background of their filters.
I have used a pretrained VGG-19 model as the backbone. Through training we aim to minimise the total loss, which is a weighted sum of the content loss and the style loss.
By tuning the weights of the loss function we can adjust how much of the style/content we want to keep in the output.