chainer-gogh

Implementation of "A neural algorithm of Artistic style" (http://arxiv.org/abs/1508.06576)

: https://research.preferred.jp/2015/09/chainer-gogh/

(VGG, lam=0.0075, after 5000 iteration)

Usage:

Chainer

pip install chainer

https://github.com/pfnet/chainer

• NIN https://gist.github.com/mavenlin/d802a5849de39225bcc6

(-m nin) * VGG https://gist.github.com/ksimonyan/211839e770f7b538e2d8#file-readme-md

(-m vgg, -m vgg_chainer) vgg_chainer(chainer 1.19)

• GoogLeNet https://github.com/BVLC/caffe/tree/master/models/bvlc_googlenet

NIN(-m googlenet)

• illustration2vec http://illustration2vec.net/ (pre-trained model for tag prediction, version 2.0)

VGG(-m i2v)

CPU

python chainer-gogh.py -m nin -i input.png -s style.png -o output_dir -g -1

GPU

python chainer-gogh.py -m nin -i input.png -s style.png -o output_dir -g GPU

VGG

python chainer-gogh.py -m vgg_chainer -i input.png -s style.png -o output_dir -g 0 --width 256

-m nin

ninvgg, vgg_chainer, googlenet, i2v

• input.txt

input0.png style0.png
input1.png style1.png
...

chainer-gogh-multi.py

python chainer-gogh-multi.py -i input.txt

VGGGPU

• --lr:

• --lam: inputstyle

• alpha, beta: models.py

• *