/
train.py
187 lines (168 loc) · 6.64 KB
/
train.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
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
import os
import sys
import yaml
import argparse
import numpy as np
from PIL import Image
import torch
import torch.nn as nn
import torch.nn.functional as F
from torchvision import datasets
from torchvision.utils import save_image
from torch.utils.data import DataLoader
from torch.autograd import Variable
import torchvision.transforms as transforms
from nets.fusion import PUGAN, Gradient_Difference_Loss
from nets.commons import Weights_Normal, Gradient_Penalty
from utils.data_utils import GetTrainingPairs, GetValImage
import matplotlib.pyplot as plt
from Par.model import DNet, TtoDNet, AENet
from math import exp
## get configs and training options
parser = argparse.ArgumentParser()
parser.add_argument("--cfg_file", type=str, default="train.yaml")
args = parser.parse_args()
with open(args.cfg_file) as f:
cfg = yaml.load(f, Loader=yaml.FullLoader)
# get info from config file
dataset_name = cfg["dataset_name"]
dataset_path = cfg["dataset_path"]
channels = cfg["chans"]
img_width = cfg["im_width"]
img_height = cfg["im_height"]
val_interval = cfg["val_interval"]
ckpt_interval = cfg["ckpt_interval"]
from_epoch = cfg["from_epoch"]
num_epochs = cfg["num_epochs"]
batch_size = cfg["batch_size"]
lr_rate = cfg["lr_rate"]
num_critic = cfg["num_critic"]
lambda_gp = cfg["gp_weight"]
lambda_gp1 = cfg["gp1_weight"]
lambda_1 = cfg["l1_weight"]
lambda_2 = cfg["lg_weight"]
lambda_d = cfg["lambda_d"]
lambda_d1 = cfg["lambda_d1"]
model_path_d = cfg["path_d"]
model_path_t = cfg["path_t"]
model_path_a = cfg["path_a"]
model_path_g_from = cfg["path_g"]
model_path_d_from = cfg["path_D"]
model_path_d1_from = cfg["path_D1"]
version = cfg["version"]
## create dir for model
checkpoint_dir = "checkpoints/"
os.makedirs(checkpoint_dir, exist_ok=True)
L1_G = torch.nn.L1Loss() # l1 loss term
L1_gp = Gradient_Penalty() # wgan_gp loss term
L_gdl = Gradient_Difference_Loss() # GDL loss term
# Initialize generator and discriminator
ugan_ = PUGAN()
generator = ugan_.netG
discriminator = ugan_.netD
discriminator1 = ugan_.netD1
generator.cuda()
discriminator.cuda()
discriminator1.cuda()
L1_gp.cuda()
L1_G.cuda()
L_gdl.cuda()
Tensor = torch.cuda.FloatTensor
if (from_epoch == 0):
generator.apply(Weights_Normal)
discriminator.apply(Weights_Normal)
discriminator1.apply(Weights_Normal)
else:
generator.load_state_dict(torch.load(model_path_g_from))
discriminator.load_state_dict(torch.load(model_path_d_from))
discriminator1.load_state_dict(torch.load(model_path_d1_from))
optimizer_G = torch.optim.Adam(filter(lambda model: model.requires_grad, generator.parameters()), lr=lr_rate)
optimizer_D = torch.optim.Adam(discriminator.parameters(), lr=lr_rate)
optimizer_D1 = torch.optim.Adam(discriminator1.parameters(), lr=lr_rate)
Dnet = DNet().cuda().eval()
Ttodnet = TtoDNet().cuda().eval()
coenet = AENet().cuda().eval()
Dnet.load_state_dict(torch.load(model_path_d))
Ttodnet.load_state_dict(torch.load(model_path_t))
coenet.load_state_dict(torch.load(model_path_a))
## Data pipeline
transforms_ = [
transforms.Resize((img_height, img_width), Image.BICUBIC),
transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)),
]
dataloader = DataLoader(
GetTrainingPairs(dataset_path, transforms_=transforms_),
batch_size=batch_size,
shuffle=True,
num_workers=8,
)
## Training pipeline
lossD = []
lossD1 = []
lossG = []
i = 1
for epoch in range(from_epoch, num_epochs):
for i, batch in enumerate(dataloader):
# Model inputs
imgs_distorted = Variable(batch["A"].type(Tensor))
imgs_good_gt = Variable(batch["B"].type(Tensor))
optimizer_D.zero_grad()
with torch.no_grad():
depth = Dnet(imgs_distorted)
ae = coenet(imgs_distorted)
tm = Ttodnet(ae, depth)[0]
imgs_fake = generator(imgs_distorted, tm)
pred_real = discriminator(imgs_good_gt)
pred_fake = discriminator(imgs_fake)
loss_D = -torch.mean(pred_real) + torch.mean(pred_fake) # wgan
gradient_penalty = L1_gp(discriminator, imgs_good_gt.data, imgs_fake.data)
loss_D += lambda_gp * gradient_penalty # Eq.2 paper
loss_D.backward()
optimizer_D.step()
lossD.append(loss_D.item() / (len(imgs_distorted)))
## Train Discriminator
optimizer_D1.zero_grad()
with torch.no_grad():
depth_real = torch.cat([Dnet(imgs_good_gt), imgs_good_gt], 1)
with torch.no_grad():
depth_fake = torch.cat([Dnet(imgs_fake), imgs_fake], 1)
pred_real1 = discriminator1(depth_real)
pred_fake1 = discriminator1(depth_fake)
loss_D1 = -torch.mean(pred_real1) + torch.mean(pred_fake1) # wgan
gradient_penalty = L1_gp(discriminator1, depth_real.data, depth_fake.data)
loss_D1 += lambda_gp1 * gradient_penalty # Eq.2 paper
loss_D1.backward()
optimizer_D1.step()
lossD1.append(loss_D1.item() / (len(imgs_distorted)))
optimizer_G.zero_grad()
## Train Generator at 1:num_critic rate
if i % num_critic == 0:
with torch.no_grad():
depth = Dnet(imgs_distorted)
ae = coenet(imgs_distorted)
tm = Ttodnet(ae, depth)[0]
imgs_fake = generator(imgs_distorted, tm)
pred_fake = discriminator(imgs_fake.detach())
loss_gen = -torch.mean(pred_fake)
with torch.no_grad():
depth_fake = torch.cat([Dnet(imgs_fake), imgs_fake], 1)
pred_fake1 = discriminator1(depth_fake.detach())
loss_gen1 = -torch.mean(pred_fake1)
loss_1 = L1_G(imgs_fake, imgs_good_gt)
loss_gdl = L_gdl(imgs_fake, imgs_good_gt)
loss_G = lambda_d * loss_gen + lambda_d1 * loss_gen1 + lambda_1 * loss_1 + lambda_2 * loss_gdl
loss_G.backward()
optimizer_G.step()
lossG.append(loss_G.item() / (len(imgs_distorted)))
## Print log
if not i % 50:
sys.stdout.write("\r[Epoch %d/%d: batch %d/%d] [DLoss: %.3f, D1Loss: %.3f, 1Loss: %.3f, gdlLoss: %.3f, GLoss: %.3f] "
% (epoch, num_epochs, i, len(dataloader),
loss_D.item(), loss_D1.item(),loss_1.item(), loss_gdl.item(), loss_G.item()
))
## Save model checkpoints
if ((epoch + 1) % ckpt_interval == 0):
torch.save(generator.state_dict(),"./checkpoints/UIEB/ugan_generator_" + str(version) + str(epoch) + ".pth")
torch.save(discriminator.state_dict(),"./checkpoints/UIEB/ugan_discriminator_" + str(version) + str(epoch) + ".pth")
torch.save(discriminator1.state_dict(),"./checkpoints/UIEB/ugan_discriminator1_" + str(version) + str(epoch) + ".pth")