forked from Wangt-CN/MTFN-RR-PyTorch-Code
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathengine.py
210 lines (164 loc) · 7.57 KB
/
engine.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
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
#encoding:utf-8
# -----------------------------------------------------------
# "Matching Images and Text with Multi-modal Tensor Fusion and Re-ranking"
# WangTan, XingXu, YangYang, Alan Hanjalic, HengtaoShen, JingkuanSong
# ACM Multimedia 2019, Nice, France
# Writen by WangTan, 2019
# ------------------------------------------------------------
import time
import torch
import numpy as np
import sys
from torch.autograd import Variable
import utils
import seq2vec
import tensorboard_logger as tb_logger
import logging
def train(train_loader, model, criterion, optimizer, epoch, print_freq=10):
# switch to train mode
model.train()
batch_time = utils.AverageMeter()
data_time = utils.AverageMeter()
train_logger = utils.LogCollector()
end = time.time()
for i, train_data in enumerate(train_loader):
images, captions, lengths, ids = train_data
batch_size = images.size(0)
margin = 0.2
# measure data loading time
data_time.update(time.time() - end)
model.logger = train_logger
input_visual = Variable(images)
input_text = Variable(captions)
if torch.cuda.is_available():
input_visual = input_visual.cuda()
input_text = input_text.cuda()
#target_answer = Variable(sample['answer'].cuda(async=True))
# compute output and loss
scores = model(input_visual, input_text)
torch.cuda.synchronize()
loss = utils.calcul_loss(scores, input_visual.size(0), margin)
# label_ = torch.eye(batch_size).long()
# label = Variable(label_).view(-1).cuda()
# loss_all = criterion(sims.view(-1, 2), label)
train_logger.update('L', loss.cpu().data.numpy())
optimizer.zero_grad()
loss.backward()
torch.cuda.synchronize()
optimizer.step()
torch.cuda.synchronize()
# measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
if i % print_freq == 0:
logging.info(
'Epoch: [{0}][{1}/{2}]\t'
'Time {batch_time.val:.3f}\t'
'{elog}\t'
.format(epoch, i, len(train_loader),
batch_time=batch_time,
# loss=train_logger['L'], loss_p=train_logger['loss_p'], loss_n=train_logger['loss_p'],
# acc=train_logger['acc'], rec=train_logger['rec'], pre=train_logger['pre'],
elog=str(train_logger)))
tb_logger.log_value('epoch', epoch, step=model.Eiters)
tb_logger.log_value('step', i, step=model.Eiters)
tb_logger.log_value('batch_time', batch_time.val, step=model.Eiters)
train_logger.tb_log(tb_logger, step=model.Eiters)
def validate(val_loader, model, criterion, optimizer, batch_size):
model.eval()
val_logger = utils.LogCollector()
model.logger = val_logger
start = time.time()
input_ii = torch.zeros(5070, 36, 2048)
input_visual = []
input_text = []
ids_ = []
# input_visual = np.zeros((len(val_loader.dataset), 49, 2048))
# input_text = np.zeros((len(val_loader.dataset), 2400))
d = np.zeros((1014, 5070))
for i, val_data in enumerate(val_loader):
images, captions, lengths, ids = val_data
input_ii[ids] = images
# input_visual.append(images)
input_text.append(captions)
ids_.append(ids)
input_ii = input_ii[[i for i in range(0, 5070, 5)]]
input_visual = [input_ii[batch_size*i:min(batch_size*(i+1), 1014)] for i in range(1014//batch_size + 1)]
del input_ii
for j in range(len(input_visual)):
for k in range(len(input_text)):
sys.stdout.write('\r>> shard_distance batch (%d,%d)' % (j, k))
input_v = input_visual[j]
input_t = input_text[k]
batch_size_v = input_v.size(0)
batch_size_t = input_t.size(0)
ims = Variable(input_v).cuda()
txs = Variable(input_t).cuda()
sums = model(ims, txs)
# sums = sums.view(batch_size_v, batch_size_t)
d[batch_size*j:min(batch_size*(j+1), 1014), batch_size*k:min(batch_size*(k+1), 5070)] = sums.cpu().data.numpy()
sys.stdout.write('\n')
# np.save('flickr1_stage1_d', d)
end = time.time()
print("calculate similarity time:", end - start)
(r1i, r5i, r10i, medri, meanri), _ = utils.acc_i2t(d)
logging.info("Image to text: %.1f, %.1f, %.1f, %.1f, %.1f" %
(r1i, r5i, r10i, medri, meanri))
(r1t, r5t, r10t, medrt, meanrt), _ = utils.acc_t2i(d)
logging.info("Text to image: %.1f, %.1f, %.1f, %.1f, %.1f" %
(r1t, r5t, r10t, medrt, meanrt))
currscore = r1t + r5t + r10t + r1i + r5i + r10i
tb_logger.log_value('r1i', r1i, step=model.Eiters)
tb_logger.log_value('r5i', r5i, step=model.Eiters)
tb_logger.log_value('r10i', r10i, step=model.Eiters)
tb_logger.log_value('medri', medri, step=model.Eiters)
tb_logger.log_value('meanri', meanri, step=model.Eiters)
tb_logger.log_value('r1t', r1t, step=model.Eiters)
tb_logger.log_value('r5t', r5t, step=model.Eiters)
tb_logger.log_value('r10t', r10t, step=model.Eiters)
tb_logger.log_value('medrt', medrt, step=model.Eiters)
tb_logger.log_value('meanrt', meanrt, step=model.Eiters)
tb_logger.log_value('rsum', currscore, step=model.Eiters)
return currscore
# for (1000, 5000)
def validate2(val_loader, model, criterion, optimizer):
model.eval()
val_logger = utils.LogCollector()
model.logger = val_logger
start = time.time()
input_visual = np.zeros((len(val_loader.dataset), 36, 2048))
input_text = np.zeros((len(val_loader.dataset), 47), dtype=np.int64)
maxl = 0
for i, val_data in enumerate(val_loader):
images, captions, lengths, ids, maxlength = val_data
if maxlength > maxl:
maxl = maxlength
batch_size = images.size(0)
input_visual[ids] = (images.numpy().copy())
input_text[ids,:captions.size(1)] = (captions.numpy().copy())
input_visual = np.array([input_visual[i] for i in range(0, len(input_visual), 5)])
input_visual = input_visual[:896]
input_text = input_text[:4480]
d = utils.shard_dis(input_visual, input_text, model)
# sim = utils.shard_dis(input_visual, input_text, model)
end = time.time()
print("calculate similarity time:", end - start)
(r1i, r5i, r10i, medri, meanri), _ = utils.acc_i2t2(d)
logging.info("Image to text: %.1f, %.1f, %.1f, %.1f, %.1f" %
(r1i, r5i, r10i, medri, meanri))
(r1t, r5t, r10t, medrt, meanrt), _ = utils.acc_t2i2(d)
logging.info("Text to image: %.1f, %.1f, %.1f, %.1f, %.1f" %
(r1t, r5t, r10t, medrt, meanrt))
currscore = r1t + r5t + r10t + r1i + r5i + r10i
tb_logger.log_value('r1i', r1i, step=model.Eiters)
tb_logger.log_value('r5i', r5i, step=model.Eiters)
tb_logger.log_value('r10i', r10i, step=model.Eiters)
tb_logger.log_value('medri', medri, step=model.Eiters)
tb_logger.log_value('meanri', meanri, step=model.Eiters)
tb_logger.log_value('r1t', r1t, step=model.Eiters)
tb_logger.log_value('r5t', r5t, step=model.Eiters)
tb_logger.log_value('r10t', r10t, step=model.Eiters)
tb_logger.log_value('medrt', medrt, step=model.Eiters)
tb_logger.log_value('meanrt', meanrt, step=model.Eiters)
tb_logger.log_value('rsum', currscore, step=model.Eiters)
return currscore