-
Notifications
You must be signed in to change notification settings - Fork 8
/
train_module.py
174 lines (147 loc) · 5.61 KB
/
train_module.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
import torch
import torch.nn as nn
from torch.autograd import Variable
import torch.utils.data as Data
import torch.nn.functional as F
import torchvision
import cv2
from loss import *
import numpy as np
import cv2
import random
import time
import os
import argparse
from models import *
from func import *
from data.dataset_util import RainDataset
from torch.utils.data import DataLoader
from tensorboardX import SummaryWriter
class trainer:
def __init__(self, opt):
self.net_D = Discriminator().cuda()
self.net_G = Generator().cuda()
self.optim1 = torch.optim.Adam(filter(lambda p : p.requires_grad, self.net_G.parameters()), lr = opt.lr, betas = (0.5,0.99))
self.optim2 = torch.optim.Adam(filter(lambda p : p.requires_grad, self.net_D.parameters()), lr = opt.lr, betas = (0.5,0.99))
self.start = opt.load
self.iter = opt.iter
self.batch_size = opt.batch_size
train_dataset = RainDataset(opt)
valid_dataset = RainDataset(opt, is_eval=True)
train_size = len(train_dataset)
valid_size = len(valid_dataset)
self.train_loader = DataLoader(train_dataset, batch_size=opt.batch_size, shuffle=True)
self.valid_loader = DataLoader(valid_dataset, batch_size=opt.batch_size)
print("# train set : {}".format(train_size))
print("# eval set : {}".format(valid_size))
self.expr_dir = opt.checkpoint_dir
#Attention Loss
self.criterionAtt = AttentionLoss(theta=0.8, iteration=4)
#GAN Loss
self.criterionGAN = GANLoss(real_label=1.0, fake_label=0.0)
#Perceptual Loss
self.criterionPL = PerceptualLoss()
#Multiscale Loss
self.criterionML = MultiscaleLoss(ld = [0.6,0.8,1.0],batch=self.batch_size)
#MAP Loss
self.criterionMAP = MAPLoss(gamma = 0.05)
#MSE Loss
self.criterionMSE = nn.MSELoss().cuda()
self.out_path = './weight/'
def forward_process(self,I_,GT, is_train=True):
M_ = []
for i in range(I_.shape[0]):
M_.append(get_mask(np.array(I_[i]),np.array(GT[i])))
M_ = np.array(M_)
M_ = torch_variable(M_, is_train)
I_ = torch_variable(I_, is_train)
GT_ = torch_variable(GT, is_train)
A_, t1, t2, t3 = self.net_G(I_)
# print 'mask len', len(A_)
S_ = [t1,t2,t3]
O_ = t3
loss = self.criterionMSE(O_,GT_.detach())
if is_train:
#attention_loss
loss_att = self.criterionAtt(A_,M_.detach())
#perceptual_loss O_: generation, T_: GT
loss_PL = self.criterionPL(O_, GT_.detach())
#Multiscale_loss
loss_ML = self.criterionML(S_,GT)
# print('t3', t3.shape)
# D(Fake)
D_map_O, D_fake = self.net_D(t3)
# D(Real)
# GT = torch_variable(GT,is_train, is_grad=True)
D_map_R, D_real = self.net_D(GT_)
loss_MAP = self.criterionMAP(D_map_O, D_map_R, A_[-1].detach())
# 1 - D_real
# 0 - D_fake
# loss_GAN_fake = self.criterionGAN(D_fake,is_real=False)
# loss_GAN_real = self.criterionGAN(D_real,is_real=True)
# loss_gen_D = torch.log(1.0-loss_GAN_fake)
loss_fake = self.criterionGAN(D_fake,is_real=False) # BCE 1, D_fake -(log(1-fake))
loss_real = self.criterionGAN(D_real,is_real=True) # BCE 0, D_real -log(real)
#D_real, 1
loss_D = loss_real+loss_fake + loss_MAP
# print (loss_gen_D), (loss_att), (loss_ML), (loss_PL)
loss_G = 0.01 * (-loss_fake) + loss_att + loss_ML + loss_PL
output = [loss_G, loss_D, loss_PL, loss_ML, loss_att, loss_MAP, loss]
else: # validation
output = loss
return output
def train_start(self):
loss_sum = 0.
valid_loss_sum = 0.
# I_: input raindrop image
# A_: attention map(Mask_list) from ConvLSTM
# M_: mask GT
# O_: output image of the autoencoder
# T_: GT
writer = SummaryWriter()
count = 0
before_loss = 10000000
for epoch in range(self.start, self.iter+1):
for i, data in enumerate(self.train_loader):
count+=1
I_, GT_ = data
# print 'GT:',GT_.shape
loss_G, loss_D, loss_PL, loss_ML, loss_att, loss_MAP, MSE_loss= self.forward_process(I_,GT_)
# print loss_G
self.optim1.zero_grad()
loss_G.backward(retain_graph=True)
self.optim1.step()
self.optim2.zero_grad()
loss_D.backward()
self.optim2.step()
if count % 20==0:
print('count: '+str(count)+' loss G: {:.4f}'.format(float(loss_G.data[0]))+' loss_D: {:.4f}'.format(float(loss_D.data[0]))+' loss_MSE: {:.4f}'.format(MSE_loss.data[0]))
print('loss_PL:{:.4f}'.format(float(loss_PL.data[0]))+' loss_ML:{:.4f}'.format(float(loss_ML.data[0]))+' loss_Att:{:.4f}'.format(float(loss_att.data[0]))+' loss_MAP:{:.4f}'.format(float(loss_MAP.data[0])))
writer.add_scalar('loss_G', float(loss_G.data[0]), count)
writer.add_scalar('loss_D', float(loss_D.data[0]), count)
step = 0
for i, data in enumerate(self.valid_loader):
I_, GT_ = data
if i == 0:
valid_loss_sum = self.forward_process(I_,GT_, is_train=False)
else:
valid_loss_sum += self.forward_process(I_,GT_, is_train=False)
step+=1
print('epoch_'+str(epoch)+'valid_loss:{} '.format(valid_loss_sum.data[0]/step)+'\n')
writer.add_scalar('validation_loss', float(valid_loss_sum.data[0])/step, epoch)
valid_loss_sum = float(valid_loss_sum.data[0])/step
if before_loss/valid_loss_sum >1.01:
before_loss = valid_loss_sum
print("save model " + "!"*10)
if not os.path.exists(self.out_path):
os.system('mkdir -p {}'.format(self.out_path))
w_name = 'G_epoch:{}_loss:{}.pth'.format(epoch,valid_loss_sum)
save_path = os.path.join(self.out_path,w_name)
torch.save(self.net_G.state_dict(), save_path)
w_name = 'D_epoch:{}_loss:{}.pth'.format(epoch,valid_loss_sum)
save_path = os.path.join(self.out_path,w_name)
torch.save(self.net_D.state_dict(), save_path)
valid_loss_sum = 0.
writer.export_scalars_to_json("./attention_video_restoration.json")
writer.close()
return