-
Notifications
You must be signed in to change notification settings - Fork 2
/
sketch2faceg_model.py
executable file
·141 lines (121 loc) · 7.44 KB
/
sketch2faceg_model.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
import torch
from .base_model import BaseModel
from . import networks
class Sketch2FaceGModel(BaseModel):
""" This class adapts the pix2pix model, for learning a mapping from input images to output images given paired data.
The model training requires '--dataset_mode facesg' dataset.
By default, it uses a '--netG unet256' U-Net generator,
a '--netD basic' discriminator (PatchGAN),
and a '--gan_mode' vanilla GAN loss (the cross-entropy objective used in the orignal GAN paper).
pix2pix paper based on: https://arxiv.org/pdf/1611.07004.pdf
"""
@staticmethod
def modify_commandline_options(parser, is_train=True):
"""Add new dataset-specific options, and rewrite default values for existing options.
Parameters:
parser -- original option parser
is_train (bool) -- whether training phase or test phase. You can use this flag to add training-specific or test-specific options.
Returns:
the modified parser.
For pix2pix, we do not use image buffer
The training objective is: GAN Loss + lambda_L1 * ||G(A)-B||_1
By default, we use vanilla GAN loss, UNet with batchnorm, and aligned datasets.
"""
# changing the default values to match the pix2pix paper (https://phillipi.github.io/pix2pix/)
parser.set_defaults(norm='batch', netG='unet_256', dataset_mode='aligned')
if is_train:
parser.set_defaults(pool_size=0, gan_mode='vanilla')
parser.add_argument('--lambda_L1', type=float, default=100.0, help='weight for L1 loss')
return parser
def __init__(self, opt):
"""Initialize the sketch2face class.
Parameters:
opt (Option class)-- stores all the experiment flags; needs to be a subclass of BaseOptions
"""
BaseModel.__init__(self, opt)
# specify the training losses you want to print out. The training/test scripts will call <BaseModel.get_current_losses>
self.loss_names = ['G_GAN', 'G_L1', 'G_2_L1', 'D_real', 'D_fake']
# specify the images you want to save/display. The training/test scripts will call <BaseModel.get_current_visuals>
self.visual_names = ['real_A', 'fake_B', 'fake_B_2', 'real_B', 'real_gray_B']
# specify the models you want to save to the disk. The training/test scripts will call <BaseModel.save_networks> and <BaseModel.load_networks>
if self.isTrain:
self.model_names = ['G', 'G_2', 'D']
else: # during test time, only load G and G_2
self.model_names = ['G', 'G_2']
# define networks (both generator and discriminator)
self.netG = networks.define_G(opt.input_nc, opt.output_nc, opt.ngf, opt.netG, opt.norm,
not opt.no_dropout, opt.init_type, opt.init_gain, self.gpu_ids)
self.netG_2 = networks.define_G(opt.input_nc, opt.output_nc, opt.ngf, opt.netG, opt.norm,
not opt.no_dropout, opt.init_type, opt.init_gain, self.gpu_ids)
if self.isTrain: # define a discriminator; conditional GANs need to take both input and output images; Therefore, #channels for D is input_nc + output_nc
self.netD = networks.define_D(opt.input_nc + opt.output_nc, opt.ndf, opt.netD,
opt.n_layers_D, opt.norm, opt.init_type, opt.init_gain, self.gpu_ids)
if self.isTrain:
# define loss functions
self.criterionGAN = networks.GANLoss(opt.gan_mode).to(self.device)
self.criterionL1 = torch.nn.L1Loss()
# initialize optimizers; schedulers will be automatically created by function <BaseModel.setup>.
self.optimizer_G = torch.optim.Adam(self.netG.parameters(), lr=opt.lr, betas=(opt.beta1, 0.999))
self.optimizer_G_2 = torch.optim.Adam(self.netG_2.parameters(), lr=opt.lr, betas=(opt.beta1, 0.999))
self.optimizer_D = torch.optim.Adam(self.netD.parameters(), lr=opt.lr, betas=(opt.beta1, 0.999))
self.optimizers.append(self.optimizer_G)
self.optimizers.append(self.optimizer_G_2)
self.optimizers.append(self.optimizer_D)
def set_input(self, input):
"""Unpack input data from the dataloader and perform necessary pre-processing steps.
Parameters:
input (dict): include the data itself and its metadata information.
The option 'direction' can be used to swap images in domain A and domain B.
"""
assert self.opt.direction in ['sketch2face', 'face2sketch']
sketch2face = self.opt.direction == 'sketch2face'
self.real_A = input['sketch' if sketch2face else 'photo'].to(self.device)
self.real_B = input['photo' if sketch2face else 'sketch'].to(self.device)
self.real_gray_B = input['gray'].to(self.device)
self.image_paths = input['sketch_path' if sketch2face else 'photo_path']
def forward(self):
"""Run forward pass; called by both functions <optimize_parameters> and <test>."""
self.fake_B = self.netG(self.real_A) # G(A)
self.fake_B_2 = self.netG_2(self.fake_B) # iterative refinement #1
def backward_D(self):
"""Calculate GAN loss for the discriminator"""
# Fake; stop backprop to the generator by detaching fake_B
fake_AB = torch.cat((self.real_A, self.fake_B_2), 1) # we use conditional GANs; we need to feed both input and output to the discriminator
pred_fake = self.netD(fake_AB.detach())
self.loss_D_fake = self.criterionGAN(pred_fake, False)
# Real
real_AB = torch.cat((self.real_A, self.real_B), 1)
pred_real = self.netD(real_AB)
self.loss_D_real = self.criterionGAN(pred_real, True)
# combine loss and calculate gradients
self.loss_D = (self.loss_D_fake + self.loss_D_real) * 0.5
self.loss_D.backward()
def backward_G(self):
"""Calculate GAN and L1 loss for the generator"""
# First, G(A) should fake the discriminator
fake_AB = torch.cat((self.real_A, self.fake_B_2), 1)
pred_fake = self.netD(fake_AB)
self.loss_G_GAN = self.criterionGAN(pred_fake, True)
# Second, G(A) = B
self.loss_G_L1 = self.criterionL1(self.fake_B, self.real_gray_B) * self.opt.lambda_L1
# Do iterative refinement: colorize
self.loss_G_2_L1 = self.criterionL1(self.fake_B_2, self.real_B) * self.opt.lambda_L1
# combine loss and calculate gradients
self.loss_G = self.loss_G_GAN + self.loss_G_L1 + self.loss_G_2_L1
self.loss_G.backward()
def optimize_parameters(self):
self.forward() # compute fake images: G(A)
# update D
self.set_requires_grad(self.netD, True) # enable backprop for D
self.optimizer_D.zero_grad() # set D's gradients to zero
self.backward_D() # calculate gradients for D
self.optimizer_D.step() # update D's weights
# update G
self.set_requires_grad(self.netD, False) # D requires no gradients when optimizing G
self.optimizer_G.zero_grad() # set G's gradients to zero
#self.optimizer_G.step() # update G's weights
# update G_2
self.optimizer_G_2.zero_grad() # set G's gradients to zero
self.backward_G() # calculate gradients for G
self.optimizer_G.step() # update G's weights
self.optimizer_G_2.step() # update G's weights