This Style Transfer implementation is based on the paper of Gatys' : Image Style Transfer Using Convolutional Neural Networks
The implementation used Keras and Tensorflow API.
Compare to the Gatys' paper, there are 3 main differences:
- the adam optimizer is used insteand of BFGS
- the total variance loss is also used (which was not present in the paper)
- The convolution layers used come from VGG16 instead of VGG19