-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathloss.py
411 lines (343 loc) · 15.2 KB
/
loss.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
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
import torch
from torch.autograd import Variable
import torch.nn.functional as F
import torch.nn as nn
import numpy as np
import pdb
def cosine_sim(im, s):
"""Cosine similarity between all the image and sentence pairs
"""
return im.mm(s.t())
def order_sim(im, s):
"""Order embeddings similarity measure $max(0, s-im)$
"""
YmX = (s.unsqueeze(1).expand(s.size(0), im.size(0), s.size(1))
- im.unsqueeze(0).expand(s.size(0), im.size(0), s.size(1)))
score = -YmX.clamp(min=0).pow(2).sum(2).sqrt().t()
return score
def euclidean_sim(im, s):
"""Order embeddings similarity measure $max(0, s-im)$
"""
YmX = (s.unsqueeze(1).expand(s.size(0), im.size(0), s.size(1))
- im.unsqueeze(0).expand(s.size(0), im.size(0), s.size(1)))
score = -YmX.pow(2).sum(2).t()
return score
class improveTripletLoss(nn.Module):
"""
triplet ranking loss
"""
def __init__(self, margin=0, measure=False, max_violation=False, cost_style='sum', direction='all'):
super(TripletLoss, self).__init__()
self.margin = margin
self.cost_style = cost_style
self.direction = direction
if measure == 'order':
self.sim = order_sim
elif measure == 'euclidean':
self.sim = euclidean_sim
else:
self.sim = cosine_sim
self.max_violation = max_violation
def forward(self, s, im, s_idx, t_idx):
# compute image-sentence score matrix
scores = self.sim(im, s)
seten_sim = self.sim(s, s)
video_sim = self.sim(im, im)
diagonal = scores.diag().view(im.size(0), 1)
d1 = diagonal.expand_as(scores)
d2 = diagonal.t().expand_as(scores)
# clear diagonals
mask = torch.eye(scores.size(0)) > .5
I = Variable(mask)
# generate_index
inter_domain_pair_index = []
intra_domain_pair_index = []
if len(s_idx) > len(t_idx):
add_ele = len(s_idx) - len(t_idx)
else:
add_ele = len(t_idx) - len(s_idx)
for ii in range(scores.size(0)):
if len(s_idx) > len(t_idx):
if ii in s_idx:
inter_domain_pair_index.append(t_idx+[ii]*add_ele)
intra_domain_pair_index.append(s_idx)
else:
inter_domain_pair_index.append(s_idx)
intra_domain_pair_index.append(t_idx+[ii]*add_ele)
elif len(s_idx) < len(t_idx):
if ii in s_idx:
inter_domain_pair_index.append(t_idx)
intra_domain_pair_index.append(s_idx+[ii]*add_ele)
else:
inter_domain_pair_index.append(s_idx+[ii]*add_ele)
intra_domain_pair_index.append(t_idx)
else:
if ii in s_idx:
inter_domain_pair_index.append(t_idx)
intra_domain_pair_index.append(s_idx)
else:
inter_domain_pair_index.append(s_idx)
intra_domain_pair_index.append(t_idx)
inter_domain_pair_index = Variable(torch.LongTensor(inter_domain_pair_index), volatile=True).cuda()
intra_domain_pair_index = Variable(torch.LongTensor(intra_domain_pair_index), volatile=True).cuda()
# for re_idx in range(scores.size(0)):
# if re_idx in s_idx:
# for xx in t_idx:
# I[re_idx, xx] = True
# else:
# for xx in s_idx:
# I[re_idx, xx] = True
if torch.cuda.is_available():
I = I.cuda()
cost_s = None
cost_im = None
# compare every diagonal score to scores in its column
if self.direction in ['i2t', 'all']:
# caption retrieval
cost_s = (self.margin + scores - d1).clamp(min=0)
cost_s = cost_s.masked_fill_(I, 0)
# compare every diagonal score to scores in its row
if self.direction in ['t2i', 'all']:
# image retrieval
cost_im = (self.margin + scores - d2).clamp(min=0)
cost_im = cost_im.masked_fill_(I, 0)
# keep the maximum violating negative for each query
if self.max_violation:
if cost_s is not None:
# max_idx = cost_s.max(1)[1].cpu().data.numpy()
# not_match_idx = []
# for re_idx in range(len(max_idx)):
# if re_idx in s_idx:
# if max_idx[re_idx] not in s_idx:
# not_match_idx.append(re_idx)
# else:
# if max_idx[re_idx] not in t_idx:
# not_match_idx.append(re_idx)
inter_pair_sim = torch.gather(cost_s, 1, inter_domain_pair_index)
intra_pair_sim = torch.gather(cost_s, 1, intra_domain_pair_index)
intra_mask = intra_pair_sim.max(1)[0] > inter_pair_sim.max(1)[0]
inter_mask = intra_pair_sim.max(1)[0] < inter_pair_sim.max(1)[0]
stn_max_sim = torch.gather(torch.gather(seten_sim, 1, inter_domain_pair_index), 1, inter_pair_sim.max(1)[1].unsqueeze(1)).squeeze()
stn_sim_mask_for_inter = stn_max_sim < -0.5
# print(inter_mask * stn_sim_mask_for_inter)
loss_for_intra = torch.masked_select(intra_pair_sim.max(1)[0], intra_mask).sum()
loss_for_inter = torch.masked_select(inter_pair_sim.max(1)[0], inter_mask * stn_sim_mask_for_inter).sum() + \
torch.masked_select(intra_pair_sim.max(1)[0], inter_mask * (stn_sim_mask_for_inter^1)).sum()
cost_s = loss_for_intra + loss_for_inter
if cost_im is not None:
# cost_im = cost_im.max(0)[0]
inter_im_pair_sim = torch.gather(cost_im, 0, inter_domain_pair_index.t())
intra_im_pair_sim = torch.gather(cost_im, 0, intra_domain_pair_index.t())
intra_mask_im = intra_im_pair_sim.max(0)[0] > inter_im_pair_sim.max(0)[0]
inter_mask_im = intra_im_pair_sim.max(0)[0] < inter_im_pair_sim.max(0)[0]
stn_max_sim_im = torch.gather(torch.gather(video_sim, 0, inter_domain_pair_index.t()), 0, inter_im_pair_sim.max(0)[1].unsqueeze(0)).squeeze()
stn_sim_mask_for_inter_im = stn_max_sim_im < -0.5
# print(inter_mask_im * stn_sim_mask_for_inter_im)
loss_for_intra_im = torch.masked_select(intra_im_pair_sim.max(0)[0], intra_mask_im).sum()
loss_for_inter_im = torch.masked_select(inter_im_pair_sim.max(0)[0], inter_mask_im * stn_sim_mask_for_inter_im).sum() + \
torch.masked_select(intra_im_pair_sim.max(0)[0], inter_mask_im * (stn_sim_mask_for_inter_im^1)).sum()
cost_im = loss_for_intra_im + loss_for_inter_im
if cost_s is None:
cost_s = Variable(torch.zeros(1)).cuda()
if cost_im is None:
cost_im = Variable(torch.zeros(1)).cuda()
if self.cost_style == 'sum':
return cost_s + cost_im
else:
return cost_s.mean() + cost_im.mean()
class WeightedTripletLoss(nn.Module):
"""
triplet ranking loss
"""
def __init__(self, margin=0, measure=False, max_violation=False, cost_style='sum', direction='all'):
super(WeightedTripletLoss, self).__init__()
self.margin = margin
self.cost_style = cost_style
self.direction = direction
if measure == 'order':
self.sim = order_sim
elif measure == 'euclidean':
self.sim = euclidean_sim
else:
self.sim = cosine_sim
self.max_violation = max_violation
def forward(self, s, im, s_prob_matrix, im_prob_matrix):
# compute the probabity that the sample belongs to the source domain
s_softmax = F.softmax(s_prob_matrix, dim=1)
s_prob_tensor = s_softmax[:, 1]
im_softmax = F.softmax(im_prob_matrix, dim=1)
im_prob_tensor = im_softmax[:, 1]
print('s', s_prob_tensor)
print('im', im_prob_tensor)
sys.exit(0)
# compute image-sentence score matrix
scores = self.sim(im, s)
diagonal = scores.diag().view(im.size(0), 1)
d1 = diagonal.expand_as(scores)
d2 = diagonal.t().expand_as(scores)
# clear diagonals
mask = torch.eye(scores.size(0)) > .5
I = Variable(mask)
if torch.cuda.is_available():
I = I.cuda()
cost_s = None
cost_im = None
# compare every diagonal score to scores in its column
if self.direction in ['i2t', 'all']:
# caption retrieval
cost_s = (self.margin + scores - d1).clamp(min=0)
cost_s = cost_s.masked_fill_(I, 0)
# compare every diagonal score to scores in its row
if self.direction in ['t2i', 'all']:
# image retrieval
cost_im = (self.margin + scores - d2).clamp(min=0)
cost_im = cost_im.masked_fill_(I, 0)
# keep the maximum violating negative for each query
if self.max_violation:
if cost_s is not None:
cost_s = cost_s.max(1)[0]
cost_s = cost_s * (0 + s_prob_tensor)
if cost_im is not None:
cost_im = cost_im.max(0)[0]
cost_im = cost_im * (0 + im_prob_tensor)
if cost_s is None:
cost_s = Variable(torch.zeros(1)).cuda()
if cost_im is None:
cost_im = Variable(torch.zeros(1)).cuda()
if self.cost_style == 'sum':
return cost_s.sum() + cost_im.sum()
else:
return cost_s.mean() + cost_im.mean()
class TripletLoss(nn.Module):
"""
triplet ranking loss
"""
def __init__(self, margin=0, measure=False, max_violation=False, cost_style='sum', direction='all'):
super(TripletLoss, self).__init__()
self.margin = margin
self.cost_style = cost_style
self.direction = direction
if measure == 'order':
self.sim = order_sim
elif measure == 'euclidean':
self.sim = euclidean_sim
else:
self.sim = cosine_sim
self.max_violation = max_violation
def forward(self, s, im):
# compute image-sentence score matrix
scores = self.sim(im, s)
diagonal = scores.diag().view(im.size(0), 1)
d1 = diagonal.expand_as(scores)
d2 = diagonal.t().expand_as(scores)
# clear diagonals
mask = torch.eye(scores.size(0)) > .5
I = Variable(mask)
if torch.cuda.is_available():
I = I.cuda()
cost_s = None
cost_im = None
# compare every diagonal score to scores in its column
if self.direction in ['i2t', 'all']:
# caption retrieval
cost_s = (self.margin + scores - d1).clamp(min=0)
cost_s = cost_s.masked_fill_(I, 0)
# compare every diagonal score to scores in its row
if self.direction in ['t2i', 'all']:
# image retrieval
cost_im = (self.margin + scores - d2).clamp(min=0)
cost_im = cost_im.masked_fill_(I, 0)
# keep the maximum violating negative for each query
if self.max_violation:
if cost_s is not None:
cost_s = cost_s.max(1)[0]
if cost_im is not None:
cost_im = cost_im.max(0)[0]
if cost_s is None:
cost_s = Variable(torch.zeros(1)).cuda()
if cost_im is None:
cost_im = Variable(torch.zeros(1)).cuda()
if self.cost_style == 'sum':
return cost_s.sum() + cost_im.sum()
else:
return cost_s.mean() + cost_im.mean()
class MultidomainTripletLoss(nn.Module):
"""
triplet ranking loss
"""
def __init__(self, margin=0, measure=False, max_violation=False, cost_style='sum', direction='all'):
super(MultidomainTripletLoss, self).__init__()
self.margin = margin
self.cost_style = cost_style
self.direction = direction
if measure == 'order':
self.sim = order_sim
elif measure == 'euclidean':
self.sim = euclidean_sim
else:
self.sim = cosine_sim
self.max_violation = max_violation
def forward(self, s_s, t_s, s_im, t_im):
# compute image-sentence score matrix
# source N + target N
s = torch.cat((s_s, t_s), dim=0)
im = torch.cat((s_im, t_im), dim=0)
scores = self.sim(im, s)
diagonal = scores.diag().view(im.size(0), 1)
d1 = diagonal.expand_as(scores)
d2 = diagonal.t().expand_as(scores)
# clear diagonals
mask_for_i2t = torch.eye(scores.size(0))
mask_for_t2i = torch.eye(scores.size(0))
mask_for_i2t[s_s.size(0):,:] = 1
mask_for_t2i[:,s_s.size(0):] = 1
mask_for_i2t = mask_for_i2t > .5
mask_for_t2i = mask_for_t2i > .5
I_for_i2t = Variable(mask_for_i2t)
if torch.cuda.is_available():
I_for_i2t = I_for_i2t.cuda()
I_for_t2i = Variable(mask_for_t2i)
if torch.cuda.is_available():
I_for_t2i = I_for_t2i.cuda()
cost_s = None
cost_im = None
# compare every diagonal score to scores in its column
if self.direction in ['i2t', 'all']:
# caption retrieval
cost_s = (self.margin + scores - d1).clamp(min=0)
cost_s = cost_s.masked_fill_(I_for_i2t, 0)
# compare every diagonal score to scores in its row
if self.direction in ['t2i', 'all']:
# image retrieval
cost_im = (self.margin + scores - d2).clamp(min=0)
cost_im = cost_im.masked_fill_(I_for_t2i, 0)
# keep the maximum violating negative for each query
if self.max_violation:
if cost_s is not None:
cost_s_s = cost_s[:s_s.size(0), :s_s.size(0)].max(1)[0]
msk_s_t = (cost_s[:s_s.size(0), s_s.size(0):]-cost_s_s) < 0
cost_s_t = (cost_s[:s_s.size(0), s_s.size(0):]*msk_s_t.float()).max(1)[0]
# print(msk_s_t.float().sum().cpu().data[0])
if msk_s_t.float().sum().cpu().data[0] == s_s.size(0)*s_s.size(0):
lamda_s = 0.001
else:
lamda_s = 0
cost_s = cost_s_s + lamda_s*cost_s_t
if cost_im is not None:
cost_im_s = cost_im[:s_s.size(0), :s_s.size(0)].max(0)[0]
msk_im_t = (cost_im[s_s.size(0):, :s_s.size(0)]-cost_im_s) < 0
cost_im_t = (cost_im[s_s.size(0):, :s_s.size(0)]*msk_im_t.float()).max(0)[0]
# print(msk_im_t.float().sum().cpu().data[0])
if msk_im_t.float().sum().cpu().data[0] == s_s.size(0)*s_s.size(0):
lamda_im = 0.001
else:
lamda_im = 0
cost_im = cost_im_s + lamda_im*cost_im_t
if cost_s is None:
cost_s = Variable(torch.zeros(1)).cuda()
if cost_im is None:
cost_im = Variable(torch.zeros(1)).cuda()
if self.cost_style == 'sum':
return cost_s.sum() + cost_im.sum()
else:
return cost_s.mean() + cost_im.mean()