Training a feed-forward network for fast neural style transfer of images is
proven to be successful. However, the naive extension to process video frame by
frame is prone to producing flickering results. We propose the first end-to-end
network for online video style transfer, which gen