图片风格转换

在此之前

我们一定要搞明白模型背后的数学原理

我们的Loss损失函数是什么

计算每个featuremap之间的欧及里的距离可以知道２个图片的相近度

计算featuremap之间的协方差　可以计算２个图片的风格相似度

优化就是优化　生成的图片和内容图片每个层里面的featuremap之间的欧及里的距离

生成图片和要转换的风格图片　每层里面featuremap之间的协方差之间的距离

 

代码实现

import numpy as np

import torch
import torch.nn as nn
from torch.autograd import Variable
import torch.optim as optim

import PIL
from PIL import Image
import matplotlib.pyplot as plt

import torchvision.transforms as transforms
import torchvision.models as models

# cuda

use_cuda = torch.cuda.is_available()
dtype = torch.cuda.FloatTensor if use_cuda else torch.FloatTensor

# images
"""
图片的加载和预处理
"""
imsize = 200  # desired size of the output image

loader = transforms.Compose([
    transforms.Scale(imsize),  # scale imported image
    transforms.ToTensor()])  # transform it into a torch tensor


def image_loader(image_name):
    image = Image.open(image_name)
    image = Variable(loader(image))
    # fake batch dimension required to fit network's input dimensions
    image = image.unsqueeze(0)
    return image


style = image_loader("./images/picasso.jpg").type(dtype)
content = image_loader("./images/dancing.jpg").type(dtype)

assert style.size() == content.size(),"we need to import style and content images of the same size"


# display

unloader = transforms.ToPILImage()  # reconvert into PIL image


def toimage(tensor):
    image = tensor.clone().cpu()  # we clone the tensor to not do changes on it
    image = image.view(3, imsize, imsize)  # remove the fake batch dimension
    image = unloader(image)
    return image


# content loss
"""
图片２个mapfeature之间的欧及里得距离可以判断２个图片的内容相似度
每个mapfeature之间的协方差矩阵则表示２个图片之间风格相似度
这就是这个loss的理论基础
生成一个图片使得　与内容图片内容相近　　风格相近　　分别算２个不不同loss即可
"""
class ContentLoss(nn.Module):
    def __init__(self, target, weight):
        super(ContentLoss, self).__init__()
        # we 'detach' the target content from the tree used
        #
        self.target = target