pfnet-research/chainer-gogh

chainer-gogh's Issues

Is there any English readme here?

hi@pfnet-research
Very good work!
I want to use this project,but I can read the readme in Japanese . Could you please give a simple english readme here?
Thank you very much!

Chainer GPU

For unknown reason, I cannot get the Chainer GPU to work, on Windows, py27. If I install chainer-master, then run the chainer-gogh, it keeps on crashing.

Pre-trained model Always download failed?

cu_00095@gtx1080:/chainer-gogh$ time python chainer-gogh.py -m vgg_chainer -i /home/cu_00095/chainer-gogh/input/photo.png -s /home/cu_00095/chainer-gogh/style/q.png -o /home/cu_00095/chainer-gogh/result -g 0
load model... vgg_chainer
Downloading from http://www.robots.ox.ac.uk/%7Evgg/software/very_deep/caffe/VGG_ILSVRC_16_layers.caffemodel...
Connection to proxy.chongdata.com closed by remote host.cu_00095@gtx1080:/chainer-gogh$ time python chainer-gogh.py -m vgg_chainer -i /home/cu_00095/chainer-gogh/input/photo.png -s /home/cu_00095/chainer-gogh/style/q.png -o /home/cu_00095/chainer-gogh/result -g 0
load model... vgg_chainer
Downloading from http://www.robots.ox.ac.uk/%7Evgg/software/very_deep/caffe/VGG_ILSVRC_16_layers.caffemodel...
Connection to proxy.chongdata.com closed by remote host.

I try to use vgg_chainer
But Pre-trained model Always download failed?now . I have download the VGG_ILSVRC_16_layers.caffemodel to the chainer root directory；
How should I change the code to skip the connect(Downloading from http://www.robots) step?
It have spend me all days to deal with this..Thanks

It is not obviously working with -v2

The simplest solution might be to install older version 1.2.4.
https://github.com/chainer/chainer/releases

Using nin model ,result is strange

Can you share me your parameters of nin model?My result is really strange,I need a better one

pillow get a float instead of an integer

trying (on Python-3.4 64bit, windows 10):

python chainer-gogh.py -m nin -i cat.png -s style_0.png -o zzz -g -1

I get an error I don't know what to do about:

load model... nin_imagenet.caffemodel
Traceback (most recent call last):
  File "chainer-gogh.py", line 196, in <module>
    img_content,nw,nh = image_resize(args.orig_img, W)
  File "chainer-gogh.py", line 50, in image_resize
    gogh = np.asarray(gogh.resize((new_w,new_h)))[:,:,:3].transpose(2, 0, 1)[::-1].astype(np.float32)
  File "python-3.4.4.amd64\lib\site-packages\PIL\Image.py", line 1554, in resize
    return self.convert('RGBa').resize(size, resample).convert('RGBA')
  File "python-3.4.4.amd64\lib\site-packages\PIL\Image.py", line 1556, in resize
    return self._new(self.im.resize(size, resample))
TypeError: integer argument expected, got float