vgg.py

# Copyright 2020 by Andrey Ignatov. All Rights Reserved.

from torchvision import models
import torch.nn as nn
import torch
from torchvision import models
import torch.nn as nn

CONTENT_LAYER = 'relu_16'

class VGG16_perceptual(torch.nn.Module):
    def __init__(self, requires_grad=False):
        super(VGG16_perceptual, self).__init__()
        vgg_pretrained_features = models.vgg16(pretrained=True).features
        self.slice1 = torch.nn.Sequential()
        self.slice2 = torch.nn.Sequential()
        self.slice3 = torch.nn.Sequential()
        self.slice4 = torch.nn.Sequential()
        for x in range(2):
            self.slice1.add_module(str(x), vgg_pretrained_features[x])
        for x in range(2, 4):
            self.slice2.add_module(str(x), vgg_pretrained_features[x])
        for x in range(4, 14):
            self.slice3.add_module(str(x), vgg_pretrained_features[x])
        for x in range(14, 21):
            self.slice4.add_module(str(x), vgg_pretrained_features[x])
        if not requires_grad:
            for param in self.parameters():
                param.requires_grad = False

    def forward(self, X):
        h = self.slice1(X)
        h_relu1_1 = h
        h = self.slice2(h)
        h_relu1_2 = h
        h = self.slice3(h)
        h_relu3_2 = h
        h = self.slice4(h)
        h_relu4_2 = h
        return h_relu1_1, h_relu1_2, h_relu3_2, h_relu4_2
    
def vgg_19(device):

    vgg_19 = models.vgg19(pretrained=True).features
    model = nn.Sequential()

    i = 0
    for layer in vgg_19.children():
        if isinstance(layer, nn.Conv2d):
            i += 1
            name = 'conv_{}'.format(i)
        elif isinstance(layer, nn.ReLU):
            name = 'relu_{}'.format(i)
            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 == CONTENT_LAYER:
            break

    model = model.to(device)
    model = torch.nn.DataParallel(model)

    for param in model.parameters():
        param.requires_grad = False

    for param in vgg_19.parameters():
        param.requires_grad = False

    return model