-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathpt_gan.py
160 lines (140 loc) · 4.73 KB
/
pt_gan.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
import torch
import torchvision
import torch.nn as nn
import torch.nn.functional as F
import torchvision.datasets as dsets
import torchvision.transforms as transforms
from torch.autograd import Variable
from sklearn.cross_validation import train_test_split
import numpy as np
import time
import sys
###################### ADD PATH TO THE DEEP PRIOR PACKAGE HERE
sys.path.append('../../DeepPrior/src/')
sys.path.append('../../utils/')
from data.dataset import NYUDataset
from data.importers import NYUImporter
import math
J =14
n_conv = 5
n_filters = 8
n_pool = 3
n_pixels_1 = 128
n_pixels_2 = 128
size_out_1 = 9
size_out_2 = 9
C = 1e-4
Nepoch = 400
batchsize = 256
nrand = 0
beta = 1.0
alpha = 0.5
use_gpu = 0
class Generator(nn.Module):
def __init__(self):
super(Generator, self).__init__()
self.fc1 = nn.Linear(size_out_1*size_out_2*n_filters + nrand, 1024)
self.fc2 = nn.Linear(1024, 1024)
self.fc3 = nn.Linear(1024, 3*J)
self.conv1 = nn.Conv2d(1, n_filters, n_conv)
self.conv2 = nn.Conv2d(n_filters,n_filters,n_conv)
self.conv3 = nn.Conv2d(n_filters, n_filters, n_conv)
self.mp1 = nn.MaxPool2d(n_pool)
self.mp2 = nn.MaxPool2d(n_pool)
def forward(self, x):
h = F.relu(self.conv1(x))
h = self.mp1(h)
h = F.relu(self.conv2(h))
h = self.mp2(h)
h = F.relu(self.conv2(h))
print type(h)
data = h.data
data.resize_(x.size(0), size_out_1*size_out_2*n_filters)
h = Variable(data)
h = F.relu(self.fc1(h))
h = F.relu(self.fc2(h))
h = F.relu(self.fc3(h))
return h
class Discriminator(nn.Module):
def __init__(self):
super(Discriminator, self).__init__()
self.conv1 = nn.Conv2d(1, n_filters, n_conv)
self.conv2 = nn.Conv2d(n_filters,n_filters,n_conv)
self.conv3 = nn.Conv2d(n_filters, n_filters, n_conv)
self.mp1 = nn.MaxPool2d(n_pool)
self.mp2 = nn.MaxPool2d(n_pool)
self.fc1 = nn.Linear(size_out_1*size_out_2*n_filters + nrand + 3 * J, 200)
self.fc2 = nn.Linear(200, 200)
self.fc3 = nn.Linear(200, 1)
def forward(self, x, z):
h = F.relu(self.conv1(x))
h = self.mp1(h)
h = F.relu(self.conv2(h))
h = self.mp2(h)
h = F.relu(self.conv2(h))
print type(h)
data = h.data
data.resize_(h.size(0), size_out_1*size_out_2*n_filters)
h = Variable(data)
print z.size(), h.size()
h_extended = torch.cat([h, z],1)
h = F.relu(self.fc1(h_extended))
h = F.relu(self.fc2(h))
h = F.sigmoid(self.fc3(h))
return h
if __name__ == '__main__':
print "start"
generator = Generator()
discriminator = Discriminator()
print "cuda"
discriminator.cuda()
generator.cuda()
print "opt"
criterion = nn.BCELoss()
d_optimizer = torch.optim.Adam(discriminator.parameters(), lr=0.0005)
g_optimizer = torch.optim.Adam(generator.parameters(), lr=0.0005)
#loading data
di = NYUImporter('../../DeepPrior/data/NYU')
Seq = di.loadSequence('test')
trainDataset = NYUDataset([Seq])
X_train, Y_train = trainDataset.imgStackDepthOnly('test')
Y_train = np.reshape(Y_train, (Y_train.shape[0], Y_train.shape[1]* Y_train.shape[2]))
x_train, x_val, y_train, y_val = train_test_split(X_train, Y_train, test_size = 0.2)
N = y_train.shape[0]
for epoch in range(Nepoch):
sum_dis_loss = np.float32(0)
sum_gen_loss = np.float32(0)
#xp.random.shuffle(train_data)
for i in range(0, N, batchsize):
input_images = torch.FloatTensor(x_train[i:i+batchsize])
images = Variable(input_images.cuda())
real_poses = Variable(torch.FloatTensor(y_train[i:i+batchsize])).cuda()
real_labels = Variable(torch.ones(images.size(0))).cuda()
fake_labels = Variable(torch.zeros(images.size(0))).cuda()
#train discriminator
discriminator.zero_grad()
outputs = discriminator(images, real_poses)
real_loss = criterion(outputs, real_labels)
real_score = outputs
print "real loss" , real_loss
noise = Variable(torch.randn(images.size(0), 1, images.size(2), images.size(3))).cuda()
print images.size()
fake_poses = generator(noise)
outputs = discriminator(images,fake_poses)
fake_loss = criterion(outputs, fake_labels)
fake_score = outputs
d_loss = real_loss + fake_loss
d_loss.backward()
d_optimizer.step()
print "fake_loss", fake_loss
# Train the generator
generator.zero_grad()
noise = Variable(torch.randn(images.size(0), 1, images.size(2), images.size(3))).cuda()
fake_poses = generator(noise)
outputs = discriminator(images, fake_poses)
g_loss = criterion(outputs, real_labels)
g_loss.backward()
g_optimizer.step()
print('Epoch [%d/%d], Step[%d/%d], d_loss: %.4f, g_loss: %.4f, '
'D(x): %.2f, D(G(z)): %.2f' %(epoch, 200, i+1, 600, d_loss.data[0],
g_loss.data[0],real_score.data.mean(), fake_score.cpu().data.mean()))