Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/xlassix/style-transfer
A plain implementation of Style Transfer in pytorch using resnet50
https://github.com/xlassix/style-transfer
Last synced: 19 days ago
JSON representation
A plain implementation of Style Transfer in pytorch using resnet50
- Host: GitHub
- URL: https://github.com/xlassix/style-transfer
- Owner: xlassix
- Created: 2019-11-27T19:36:56.000Z (about 5 years ago)
- Default Branch: master
- Last Pushed: 2019-11-27T19:44:01.000Z (about 5 years ago)
- Last Synced: 2024-10-18T21:04:15.578Z (2 months ago)
- Homepage:
- Size: 43.4 MB
- Stars: 1
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
## trained on kagggle
## Import required modules
---
### [#1]
```python
import os
from distutils.dir_util import copy_tree
copy_tree("../input/vgg19dcbb9e9dpth/","/tmp/checkpoints")
os.environ['TORCH_HOME'] = '/tmp'
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
import numpy as np
from PIL import Image
import matplotlib.pyplot as plt
import torchvision.transforms as transforms
import torchvision.models as models
%pylab inline
pylab.rcParams['figure.figsize'] = (20, 20)
import copy
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
pylab.rcParams['figure.figsize'] = (20, 20)
```
# Setting paramater for optimization
##
---
### [#2]
```python
#imsize = 2048 if torch.cuda.is_available() else 480
#this reduces the image quality in case a gpu is unavailable
imsize = 2048 if torch.cuda.is_available() else 1080
loader = transforms.Compose([
transforms.Scale(imsize),# scale imported image
transforms.CenterCrop(imsize),
transforms.ToTensor()]) # transform it into a torch tensor
def image_loader(image_name):
image = Image.open(image_name)
# fake batch dimension required to fit network's input dimensions
image = loader(image).unsqueeze(0)
return image.to(device, torch.float)
style_img = image_loader(r"/kaggle/input/picture/Beautiful-Computer-Wallpaper-HD-1080p.jpg")
content_img = image_loader(r"//kaggle/input/picture/unicorn-desktop-wallpaper-fantasy-art-pure-mythology-high-quality-2.jpg")
print(style_img.size(),content_img.size())
assert(style_img.size() == content_img.size())
```
---
### [#3]
```python
unloader = transforms.ToPILImage() # reconvert into PIL image
plt.ion()
def imshow(tensor, title=None):
image = tensor.cpu().clone() # we clone the tensor to not do changes on it
image = image.squeeze(0) # remove the fake batch dimension
image = unloader(image)
plt.imshow(image)
if title is not None:
plt.title(title)
plt.pause(0.001) # pause a bit so that plots are updated
plt.figure()
imshow(style_img, title='Style Image')
plt.figure()
imshow(content_img, title='Content Image')
```
---
### [#4]
```python
```
Output:
[style_img]
[content_img]
## define a filter to any (m x n) dimension as edge_detection
---
### [#5]
```python
class ContentLoss(nn.Module):
def __init__(self, target,):
super(ContentLoss, self).__init__()
# we 'detach' the target content from the tree used
# to dynamically compute the gradient: this is a stated value,
# not a variable. Otherwise the forward method of the criterion
# will throw an error.
self.target = target.detach()
def forward(self, input):
self.loss = F.mse_loss(input, self.target)
return input
def gram_matrix(input):
a, b, c, d = input.size() # a=batch size(=1)
# b=number of feature maps
# (c,d)=dimensions of a f. map (N=c*d)
features = input.view(a * b, c * d) # resise F_XL into \hat F_XL
G = torch.mm(features, features.t()) # compute the gram product
# we 'normalize' the values of the gram matrix
# by dividing by the number of element in each feature maps.
return G.div(a * b * c * d)
class StyleLoss(nn.Module):
def __init__(self, target_feature):
super(StyleLoss, self).__init__()
self.target = gram_matrix(target_feature).detach()
def forward(self, input):
G = gram_matrix(input)
self.loss = F.mse_loss(G, self.target)
return input
class Normalization(nn.Module):
def __init__(self, mean, std):
super(Normalization, self).__init__()
# .view the mean and std to make them [C x 1 x 1] so that they can
# directly work with image Tensor of shape [B x C x H x W].
# B is batch size. C is number of channels. H is height
self.mean = torch.tensor(mean).view(-1, 1, 1)
self.std = torch.tensor(std).view(-1, 1, 1)
def forward(self, img):
# normalize img
return (img - self.mean) / self.std
# desired depth layers to compute style/content losses :
content_layers_default = ['conv_4']
style_layers_default = ['conv_1', 'conv_2', 'conv_3', 'conv_4', 'conv_5']
def get_style_model_and_losses(cnn, normalization_mean, normalization_std,
style_img, content_img,
content_layers=content_layers_default,
style_layers=style_layers_default):
cnn = copy.deepcopy(cnn)
# normalization module
normalization = Normalization(normalization_mean, normalization_std).to(device)
# just in order to have an iterable access to or list of content/syle
# losses
content_losses = []
style_losses = []
model = nn.Sequential(normalization)
i = 0 # increment every time we see a conv
for layer in cnn.children():
if isinstance(layer, nn.Conv2d):
i += 1
name = 'conv_{}'.format(i)
elif isinstance(layer, nn.ReLU):
name = 'relu_{}'.format(i)
# The in-place version doesn't play very nicely with the ContentLoss
# and StyleLoss we insert below. So we replace with out-of-place
# ones here.
layer = nn.ReLU(inplace=False)
elif isinstance(layer, nn.MaxPool2d):
name = 'pool_{}'.format(i)
elif isinstance(layer, nn.BatchNorm2d):
name = 'bn_{}'.format(i)
else:
raise RuntimeError('Unrecognized layer: {}'.format(layer.__class__.__name__))
model.add_module(name, layer)
if name in content_layers:
# add content loss:
target = model(content_img).detach()
content_loss = ContentLoss(target)
model.add_module("content_loss_{}".format(i), content_loss)
content_losses.append(content_loss)
if name in style_layers:
# add style loss:
target_feature = model(style_img).detach()
style_loss = StyleLoss(target_feature)
model.add_module("style_loss_{}".format(i), style_loss)
style_losses.append(style_loss)
# now we trim off the layers after the last content and style losses
for i in range(len(model) - 1, -1, -1):
if isinstance(model[i], ContentLoss) or isinstance(model[i], StyleLoss):
break
model = model[:(i + 1)]
return model, style_losses, content_losses
input_img = content_img.clone()
# if you want to use white noise instead uncomment the below line:
# input_img = torch.randn(content_img.data.size(), device=device)
# add the original input image to the figure:
plt.figure()
imshow(input_img)
```
---
### [#6]
```python
def get_input_optimizer(input_img):
# this line to show that input is a parameter that requires a gradient
optimizer = optim.LBFGS([input_img.requires_grad_()])
return optimizer
def run_style_transfer(cnn, normalization_mean, normalization_std,
content_img, style_img, input_img, num_steps=300,
style_weight=1000000, content_weight=1):
"""Run the style transfer."""
print('Building the style transfer model..')
model, style_losses, content_losses = get_style_model_and_losses(cnn,normalization_mean, normalization_std, style_img, content_img)
optimizer = get_input_optimizer(input_img)
print('Optimizing..')
run = [0]
while run[0] <= num_steps:
def closure():
# correct the values of updated input image
input_img.data.clamp_(0, 1)
optimizer.zero_grad()
model(input_img)
style_score = 0
content_score = 0
for sl in style_losses:
style_score += sl.loss
for cl in content_losses:
content_score += cl.loss
style_score *= style_weight
content_score *= content_weight
loss = style_score + content_score
loss.backward()
run[0] += 1
if run[0] % 50 == 0:
print("run {}:".format(run))
print('Style Loss : {:4f} Content Loss: {:4f}'.format(
style_score.item(), content_score.item()))
print()
imshow(input_img, title='current modification Image')
return style_score + content_score
optimizer.step(closure)
# a last correction...
input_img.data.clamp_(0, 1)
return input_img
```
[]
```python
cnn = models.vgg19(pretrained=True).features.to(device).eval()
cnn_normalization_mean = torch.tensor([0.5, 0.5, 0.5])
cnn_normalization_std = torch.tensor([0.229, 0.224, 0.225])
output = run_style_transfer(cnn, cnn_normalization_mean, cnn_normalization_std,content_img, style_img, input_img,num_steps=800)
plt.figure()
imshow(output, title='Output Image')
# sphinx_gallery_thumbnail_number = 4
plt.ioff()
plt.show()
```
Output:
Optimizing..
run [50]:
Style Loss : 259.097107 Content Loss: 4.290390
---
run [100]:
Style Loss : 179.582535 Content Loss: 5.312847
---
run [150]:
Style Loss : 87.066132 Content Loss: 6.564795
---
run [200]:
Style Loss : 20.309290 Content Loss: 5.657081
---
run [250]:
Style Loss : 10.864233 Content Loss: 4.735281
---
run [300]:
Style Loss : 7.285984 Content Loss: 4.068372
---
run [350]:
Style Loss : 5.060086 Content Loss: 3.667866
---
run [400]:
Style Loss : 3.686680 Content Loss: 3.429209
---
run [450]:
Style Loss : 2.733267 Content Loss: 3.259518
---
run [500]:
Style Loss : 1.996152 Content Loss: 3.139938
---
run [550]:
Style Loss : 1.502804 Content Loss: 3.055034
---
run [600]:
Style Loss : 1.131459 Content Loss: 3.011957
---
run [650]:
Style Loss : 0.853874 Content Loss: 2.961558
---
run [700]:
Style Loss : 0.653365 Content Loss: 2.888459
---
run [750]:
Style Loss : 0.530083 Content Loss: 2.789802
---
run [800]:
Style Loss : 0.445384 Content Loss: 2.696638
---