migrate to pytorch 1.0.

use torch.no_grad() instead of volatile variables.
update example images to also use relu5_1 stylization.
Point models download to new file.
Remove model conversion script.
add .gitignore

.gitignore

@@ -0,0 +1,12 @@
+models
+models.zip
+models_pytorch.zip
+
+# python artifacts
+__pychache__
+*.pyc
+
+# Temporary Files
+*~
+*.swp
+*.swo

Readme.md

@@ -7,7 +7,7 @@ Official Torch implementation can be found [here](https://github.com/Yijunmaveri
 ## Prerequisites
 - [Pytorch](http://pytorch.org/)
 - [torchvision](https://github.com/pytorch/vision)
-- Pretrained encoder and decoder [models](https://drive.google.com/file/d/1M5KBPfqrIUZqrBZf78CIxLrMUT4lD4t9/view?usp=sharing) for image reconstruction only (download and uncompress them under models/)
+- Pretrained encoder and decoder [models](http://pascal.inrialpes.fr/data2/archetypal_style/models_pytorch.zip) for image reconstruction only (download and uncompress them under models/)
 - CUDA + CuDNN
 
 ## Prepare images

WCT.py

@@ -8,23 +8,18 @@
 from Loader import Dataset
 from util import *
 import scipy.misc
-from torch.utils.serialization import load_lua
 import time
 
 parser = argparse.ArgumentParser(description='WCT Pytorch')
 parser.add_argument('--contentPath',default='images/content',help='path to train')
 parser.add_argument('--stylePath',default='images/style',help='path to train')
 parser.add_argument('--workers', default=2, type=int, metavar='N',help='number of data loading workers (default: 4)')
-parser.add_argument('--vgg1', default='models/vgg_normalised_conv1_1.t7', help='Path to the VGG conv1_1')
-parser.add_argument('--vgg2', default='models/vgg_normalised_conv2_1.t7', help='Path to the VGG conv2_1')
-parser.add_argument('--vgg3', default='models/vgg_normalised_conv3_1.t7', help='Path to the VGG conv3_1')
-parser.add_argument('--vgg4', default='models/vgg_normalised_conv4_1.t7', help='Path to the VGG conv4_1')
-parser.add_argument('--vgg5', default='models/vgg_normalised_conv5_1.t7', help='Path to the VGG conv5_1')
-parser.add_argument('--decoder5', default='models/feature_invertor_conv5_1.t7', help='Path to the decoder5')
-parser.add_argument('--decoder4', default='models/feature_invertor_conv4_1.t7', help='Path to the decoder4')
-parser.add_argument('--decoder3', default='models/feature_invertor_conv3_1.t7', help='Path to the decoder3')
-parser.add_argument('--decoder2', default='models/feature_invertor_conv2_1.t7', help='Path to the decoder2')
-parser.add_argument('--decoder1', default='models/feature_invertor_conv1_1.t7', help='Path to the decoder1')
+parser.add_argument('--encoder', default='models/vgg19_normalized.pth.tar', help='Path to the VGG conv1_1')
+parser.add_argument('--decoder5', default='models/vgg19_normalized_decoder5.pth.tar', help='Path to the decoder5')
+parser.add_argument('--decoder4', default='models/vgg19_normalized_decoder4.pth.tar', help='Path to the decoder4')
+parser.add_argument('--decoder3', default='models/vgg19_normalized_decoder3.pth.tar', help='Path to the decoder3')
+parser.add_argument('--decoder2', default='models/vgg19_normalized_decoder2.pth.tar', help='Path to the decoder2')
+parser.add_argument('--decoder1', default='models/vgg19_normalized_decoder1.pth.tar', help='Path to the decoder1')
 parser.add_argument('--cuda', action='store_true', help='enables cuda')
 parser.add_argument('--batch_size', type=int, default=1, help='batch size')
 parser.add_argument('--fineSize', type=int, default=512, help='resize image to fineSize x fineSize,leave it to 0 if not resize')
@@ -45,70 +40,75 @@
                                      batch_size=1,
                                      shuffle=False)
 
-wct = WCT(args)
-def styleTransfer(contentImg,styleImg,imname,csF):
+def styleTransfer(wct, contentImg, styleImg, imname):
 
-    sF5 = wct.e5(styleImg)
-    cF5 = wct.e5(contentImg)
-    sF5 = sF5.data.cpu().squeeze(0)
-    cF5 = cF5.data.cpu().squeeze(0)
-    csF5 = wct.transform(cF5,sF5,csF,args.alpha)
+    sF5 = wct.encoder(styleImg, 'relu5_1')
+    cF5 = wct.encoder(contentImg, 'relu5_1')
+    sF5 = sF5.cpu().squeeze(0)
+    cF5 = cF5.cpu().squeeze(0)
+    csF5 = wct.transform(cF5,sF5,args.alpha)
+    csF5 = csF5.to(next(wct.parameters()).device)
     Im5 = wct.d5(csF5)
 
-    sF4 = wct.e4(styleImg)
-    cF4 = wct.e4(Im5)
-    sF4 = sF4.data.cpu().squeeze(0)
-    cF4 = cF4.data.cpu().squeeze(0)
-    csF4 = wct.transform(cF4,sF4,csF,args.alpha)
+    sF4 = wct.encoder(styleImg, 'relu4_1')
+    cF4 = wct.encoder(Im5, 'relu4_1')
+    sF4 = sF4.cpu().squeeze(0)
+    cF4 = cF4.cpu().squeeze(0)
+    csF4 = wct.transform(cF4,sF4,args.alpha)
+    csF4 = csF4.to(next(wct.parameters()).device)
     Im4 = wct.d4(csF4)
 
-    sF3 = wct.e3(styleImg)
-    cF3 = wct.e3(Im4)
-    sF3 = sF3.data.cpu().squeeze(0)
-    cF3 = cF3.data.cpu().squeeze(0)
-    csF3 = wct.transform(cF3,sF3,csF,args.alpha)
+    sF3 = wct.encoder(styleImg, 'relu3_1')
+    cF3 = wct.encoder(Im4, 'relu3_1')
+    sF3 = sF3.cpu().squeeze(0)
+    cF3 = cF3.cpu().squeeze(0)
+    csF3 = wct.transform(cF3,sF3,args.alpha)
+    csF3 = csF3.to(next(wct.parameters()).device)
     Im3 = wct.d3(csF3)
 
-    sF2 = wct.e2(styleImg)
-    cF2 = wct.e2(Im3)
-    sF2 = sF2.data.cpu().squeeze(0)
-    cF2 = cF2.data.cpu().squeeze(0)
-    csF2 = wct.transform(cF2,sF2,csF,args.alpha)
+    sF2 = wct.encoder(styleImg, 'relu2_1')
+    cF2 = wct.encoder(Im3, 'relu2_1')
+    sF2 = sF2.cpu().squeeze(0)
+    cF2 = cF2.cpu().squeeze(0)
+    csF2 = wct.transform(cF2,sF2,args.alpha)
+    csF2 = csF2.to(next(wct.parameters()).device)
     Im2 = wct.d2(csF2)
 
-    sF1 = wct.e1(styleImg)
-    cF1 = wct.e1(Im2)
-    sF1 = sF1.data.cpu().squeeze(0)
-    cF1 = cF1.data.cpu().squeeze(0)
-    csF1 = wct.transform(cF1,sF1,csF,args.alpha)
+    sF1 = wct.encoder(styleImg, 'relu1_1')
+    cF1 = wct.encoder(Im2, 'relu1_1')
+    sF1 = sF1.cpu().squeeze(0)
+    cF1 = cF1.cpu().squeeze(0)
+    csF1 = wct.transform(cF1,sF1,args.alpha)
+    csF1 = csF1.to(next(wct.parameters()).device)
     Im1 = wct.d1(csF1)
     # save_image has this wired design to pad images with 4 pixels at default.
-    vutils.save_image(Im1.data.cpu().float(),os.path.join(args.outf,imname))
+    vutils.save_image(Im1.cpu().float(),os.path.join(args.outf,imname))
     return
 
-avgTime = 0
-cImg = torch.Tensor()
-sImg = torch.Tensor()
-csF = torch.Tensor()
-csF = Variable(csF)
-if(args.cuda):
-    cImg = cImg.cuda(args.gpu)
-    sImg = sImg.cuda(args.gpu)
-    csF = csF.cuda(args.gpu)
-    wct.cuda(args.gpu)
-for i,(contentImg,styleImg,imname) in enumerate(loader):
-    imname = imname[0]
-    print('Transferring ' + imname)
-    if (args.cuda):
-        contentImg = contentImg.cuda(args.gpu)
-        styleImg = styleImg.cuda(args.gpu)
-    cImg = Variable(contentImg,volatile=True)
-    sImg = Variable(styleImg,volatile=True)
-    start_time = time.time()
-    # WCT Style Transfer
-    styleTransfer(cImg,sImg,imname,csF)
-    end_time = time.time()
-    print('Elapsed time is: %f' % (end_time - start_time))
-    avgTime += (end_time - start_time)
+def main():
+  wct = WCT(args)
+  if(args.cuda):
+      wct.cuda(args.gpu)
 
-print('Processed %d images. Averaged time is %f' % ((i+1),avgTime/(i+1)))
+  avgTime = 0
+  for i,(contentImg,styleImg,imname) in enumerate(loader):
+      if(args.cuda):
+          contentImg = contentImg.cuda(args.gpu)
+          styleImg = styleImg.cuda(args.gpu)
+      imname = imname[0]
+      print('Transferring ' + imname)
+      if (args.cuda):
+          contentImg = contentImg.cuda(args.gpu)
+          styleImg = styleImg.cuda(args.gpu)
+      start_time = time.time()
+      # WCT Style Transfer
+      styleTransfer(wct, contentImg, styleImg, imname)
+      end_time = time.time()
+      print('Elapsed time is: %f' % (end_time - start_time))
+      avgTime += (end_time - start_time)
+
+  print('Processed %d images. Averaged time is %f' % ((i+1),avgTime/(i+1)))
+
+if __name__ == '__main__':
+  with torch.no_grad():
+    main()