def create_perceptualnet(opt):
# Get the first 15 layers of vgg16, which is conv3_3
perceptualnet = network.PerceptualNet()
# Pre-trained VGG-16
pretrained_dict = torch.load(opt.perceptual_path)
load_dict(perceptualnet, pretrained_dict)
# It does not gradient
for param in perceptualnet.parameters():
param.requires_grad = False
return perceptualnet
def create_perceptualnet():
# Pre-trained VGG-16
vgg16 = torch.load('vgg16_pretrained.pth')
# Get the first 16 layers of vgg16, which is conv3_3
perceptualnet = network.PerceptualNet()
# Update the parameters
load_dict(perceptualnet, vgg16)
# It does not gradient
for param in perceptualnet.parameters():
param.requires_grad = False
return perceptualnet
def create_perceptualnet():
# Get the first 15 layers of vgg16, which is conv3_3
perceptualnet = network.PerceptualNet()
# Pre-trained VGG-16
vgg16 = torch.load('./vgg16_pretrained.pth')
load_dict(perceptualnet, vgg16)
# It does not gradient
for param in perceptualnet.parameters():
param.requires_grad = False
print('Perceptual network is created!')
return perceptualnet
def create_perceptualnet():
# Initialize the network
perceptualnet = network.PerceptualNet()
vgg16 = tv.models.vgg16(pretrained = True)
# Init the network
load_dict(perceptualnet, vgg16)
print('PerceptualNet is created!')
# It does not gradient
for param in perceptualnet.parameters():
param.requires_grad = False
return perceptualnet
def create_perceptualnet():
# Get the first 15 layers of vgg16, which is conv3_3
perceptualnet = network.PerceptualNet()
print('Perceptual network is created!')
return perceptualnet
