wellyzhang / metastyle Goto Github PK

I wonder why only the 'fast_weights(in_weights)' are updated excluding conv&deconv layers in inner loop. (line 166 in main.py)

Is it make the learning faster or something?

when I try to use fast train:
$python src/main.py train --content-dataset test2017 --style-image images\style\sketch.png --model MetaStyle.pth --cuda 1
,i got the error:
" following arguments are required: --style-dataset"
please tell me what to do

Hi,

Thanks for an amazing work!
You may also be interested in the work of our group on style transfer.
Here are some of our recent papers:

A Style-Aware Content Loss for Real-time HD Style Transfer (ECCV19, Oral)
https://compvis.github.io/adaptive-style-transfer/

A Content Transformation Block for Image Style Transfer (CVPR19)
https://compvis.github.io/content-targeted-style-transfer/

Content and Style Disentanglement for Artistic Style Transfer (ICCV19)
https://github.com/CompVis/content-style-disentangled-ST

We would highly appreciate if cite our works in your paper.

Best,
Artsiom

Is this model can realize arbitrary style?