-
Notifications
You must be signed in to change notification settings - Fork 6
/
dataset.py
353 lines (304 loc) · 13.2 KB
/
dataset.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
import os
import json
import pickle as cPickle
import numpy as np
import utils
import h5py
import torch
from torch.utils.data import Dataset
from sklearn.metrics.pairwise import cosine_similarity
from tqdm import tqdm
import pickle
from torch.utils.data.sampler import Sampler
class Dictionary(object):
def __init__(self, word2idx=None, idx2word=None):
if word2idx is None:
word2idx = {}
if idx2word is None:
idx2word = []
self.word2idx = word2idx
self.idx2word = idx2word
@property
def ntoken(self):
return len(self.word2idx)
@property
def padding_idx(self):
return len(self.word2idx)
def tokenize(self, sentence, add_word):
sentence = sentence.lower()
sentence = sentence.replace(',', '').replace('?', '').replace('\'s', ' \'s').replace('-',
' ').replace('.',
'').replace(
'"', '').replace('n\'t', ' not').replace('$', ' dollar ')
words = sentence.split()
tokens = []
if add_word:
for w in words:
if '-' in w:
print(w)
tokens.append(self.add_word(w))
else:
for w in words:
if w in self.word2idx:
tokens.append(self.word2idx[w])
else:
tokens.append(len(self.word2idx))
return tokens
def dump_to_file(self, path):
cPickle.dump([self.word2idx, self.idx2word], open(path, 'wb'))
print('dictionary dumped to %s' % path)
@classmethod
def load_from_file(cls, path):
print('loading dictionary from %s' % path)
word2idx, idx2word = cPickle.load(open(path, 'rb'))
d = cls(word2idx, idx2word)
return d
def add_word(self, word):
if word not in self.word2idx:
self.idx2word.append(word)
self.word2idx[word] = len(self.idx2word) - 1
return self.word2idx[word]
def __len__(self):
return len(self.idx2word)
def _create_entry(img, question, answer):
answer.pop('image_id')
answer.pop('question_id')
entry = {
'question_id': question['question_id'],
'image_id': question['image_id'],
'image': img,
'question': question['question'],
'answer': answer}
return entry
class SelfCriticalDataset(Dataset):
def __init__(self, split,
hint_type,
dictionary,
opt,
load_img=True,
discard_items_without_hints=False,
discard_items_with_hints=False,
ignore_counting_questions=False):
super(SelfCriticalDataset, self).__init__()
self.split = split
self.hint_type = hint_type
self.dictionary = dictionary # questions' dictionary
self.load_img = load_img
self.opt = opt
self.data_dir = opt.data_dir
self.discard_items_without_hints = discard_items_without_hints
self.discard_items_with_hints = discard_items_with_hints
if hint_type is None and self.discard_items_without_hints:
raise Exception("Cannot discard items without hints because hint_type is not specified")
if hint_type is None and self.discard_items_with_hints:
raise Exception("Cannot discard items with hints because hint_type is not specified")
self.ignore_counting_questions = ignore_counting_questions
# Load data
self.qid_to_target = self.get_qid_to_target()
ans2label_path = os.path.join(self.data_dir, 'processed', 'trainval_ans2label.pkl')
label2ans_path = os.path.join(self.data_dir, 'processed', 'trainval_label2ans.pkl')
self.ans2label = cPickle.load(open(ans2label_path, 'rb'))
self.label2ans = cPickle.load(open(label2ans_path, 'rb'))
self.num_ans_candidates = len(self.ans2label)
self.image_id2ix = {}
self.datalen = self.get_datalen()
print(f"split {self.split} len {self.datalen}")
def get_qid_to_target(self):
train_target = cPickle.load(open(os.path.join(self.data_dir, 'processed', f'train_target.pkl'), 'rb'))
val_target = cPickle.load(open(os.path.join(self.data_dir, 'processed', f'val_target.pkl'), 'rb'))
target = train_target + val_target
qid_to_target = {}
for t in target:
question_id = t['question_id']
assert question_id not in qid_to_target
qid_to_target[question_id] = t
return qid_to_target
def init_data(self):
self.hf = {}
self.features = {}
self.spatials = {}
self.cls_scores = {}
self.attr_scores = {}
print('loading features from h5 file')
self.load_data('train')
self.load_data('val')
count = self.init_vqx()
print(f"{self.split} count {count}")
self.tokenize()
self.tensorize()
self.v_dim = 2048 # self.features.size(2)
self.s_dim = 36 # self.spatials.size(2)
def load_data(self, split):
self.image_id2ix[split] = cPickle.load(open(os.path.join(self.data_dir, f'{split}36_imgid2img.pkl'), 'rb'))
h5_path = os.path.join(self.data_dir, '%s36.hdf5' % split)
self.hf[split] = h5py.File(h5_path, 'r')
self.features[split] = self.hf[split].get('image_features')
self.spatials[split] = self.hf[split].get('spatial_features')
self.cls_scores[split] = np.array(self.hf[split].get('cls_score'))
self.attr_scores[split] = np.array(self.hf[split].get('attr_score'))
def get_hint_fname(self):
if self.hint_type is None or len(self.hint_type) == 0:
return None
else:
hint_fname = f'hints/{self.split}_{self.hint_type}.pkl'
print(f"loading hints from hint_fname")
return hint_fname
def get_questions(self):
if self.opt.dataset == 'vqacp2':
f = os.path.join(self.data_dir, f'vqacp_v2_{self.split}_questions.json')
return json.load(open(f))
elif self.opt.dataset == 'vqa2':
year = '2015' if self.split == 'test' else '2014'
f = os.path.join(self.data_dir, f'v2_OpenEnded_mscoco_{self.split}{year}_questions.json')
return json.load(open(f))['questions']
def get_annotations(self):
if self.opt.dataset == 'vqacp2':
f = os.path.join(self.data_dir, f'vqacp_v2_{self.split}_annotations.json')
return json.load(open(f))
elif self.opt.dataset == 'vqa2':
year = '2015' if self.split == 'test' else '2014'
f = os.path.join(self.data_dir, f'v2_mscoco_{self.split}{year}_annotations.json')
return json.load(open(f))['annotations']
def get_datalen(self):
hint_fname = self.get_hint_fname()
count = 0
questions = self.get_questions()
if hint_fname is not None:
self.hint = cPickle.load(open(os.path.join(self.data_dir, hint_fname), 'rb'))
for i in tqdm(range(len(questions))):
question = questions[i]
question_id = question['question_id']
if self.ignore_counting_questions:
if ('how many' in question['question']) or ('how much' in question['question']):
continue
# if hint type is not specified, or if the dataset is being asked to return entire dataset, then
if self.discard_items_without_hints and question_id not in self.hint.keys():
continue
elif self.discard_items_with_hints and question_id in self.hint.keys():
continue
count += 1
return count
def init_vqx(self):
hint_fname = self.get_hint_fname()
count = 0
# self.entriess = cPickle.load(open(self.dataroot + '/VQAcp_caption_' + self.split + 'dataset.pkl', 'rb'))
self.questions = self.get_questions()
self.entries = {}
print(f"split {self.split} questions {len(self.questions)}")
if hint_fname is not None:
self.hint = cPickle.load(open(os.path.join(self.data_dir, hint_fname), 'rb'))
for i in tqdm(range(len(self.questions))):
question = self.questions[i]
image_id = question['image_id']
if image_id in self.image_id2ix['train']:
split = 'train'
else:
split = 'val'
question_id = question['question_id']
if self.ignore_counting_questions:
if ('how many' in question['question']) or ('how much' in question['question']):
continue
if self.discard_items_without_hints and question_id not in self.hint.keys():
continue
elif self.discard_items_with_hints and question_id in self.hint.keys():
continue
elif self.hint_type is not None and question_id in self.hint.keys():
hint = self.hint[question_id]
hint_a = np.zeros((36))
obj_cls = np.array(self.cls_scores[split][self.image_id2ix[split][image_id]][:, 0])
hint_o = obj_cls.astype('float')
hint_flag = 1
else:
hint_a = np.zeros((36))
hint_o = np.zeros((36))
hint = np.zeros((36))
hint_flag = 0
new_entry = {'image': self.image_id2ix[split][image_id],
'image_id': image_id,
'question_id': question_id,
'question': question['question'],
'answer': self.qid_to_target[question_id],
'hint': hint,
'hint_a': hint_a,
'hint_o': hint_o,
'hint_flag': hint_flag}
self.entries[count] = new_entry
count += 1
print(f"split {self.split} init_vqx count {count}")
return count
def tokenize(self, max_length=14):
"""Tokenizes the questions.
This will add q_token in each entry of the dataset.
-1 represent nil, and should be treated as padding_idx in embedding
"""
for e_id in range(len(self.entries)):
entry = self.entries[e_id]
tokens = self.dictionary.tokenize(entry['question'], False)
tokens = tokens[:max_length]
if len(tokens) < max_length:
# Note here we pad in front of the sentence
padding = [self.dictionary.padding_idx] * (max_length - len(tokens))
tokens = padding + tokens
utils.assert_eq(len(tokens), max_length)
entry['q_token'] = tokens
def tensorize(self):
for e_id in range(len(self.entries)):
entry = self.entries[e_id]
question = torch.from_numpy(np.array(entry['q_token']))
entry['q_token'] = question
answer = entry['answer']
labels = np.array(answer['labels'])
scores = np.array(answer['scores'], dtype=np.float32)
if labels is None:
entry['answer']['labels'] = None
entry['answer']['scores'] = None
elif len(labels):
labels = torch.from_numpy(labels)
scores = torch.from_numpy(scores)
entry['answer']['labels'] = labels
entry['answer']['scores'] = scores
else:
entry['answer']['labels'] = None
entry['answer']['scores'] = None
def __getitem__(self, index):
# features
if not hasattr(self, 'hf'):
self.init_data()
entry = self.entries[index]
imgid = entry['image_id']
if imgid in self.image_id2ix['train']:
split = 'train'
else:
split = 'val'
qid = entry['question_id']
if self.load_img:
obj_nodes = torch.from_numpy(np.array(self.features[split][entry['image']]))
else:
obj_nodes = torch.zeros(36, 2048)
hint_score = torch.from_numpy(entry['hint'])
hint_flag = entry['hint_flag']
if self.load_img:
hint_o = torch.from_numpy(entry['hint_o'])
else:
hint_o = torch.zeros_like(hint_score)
question = entry['q_token']
answer = entry['answer']
labels = answer['labels']
scores = answer['scores']
target = torch.zeros(self.num_ans_candidates)
hint_score = hint_score.float().unsqueeze(1)
if labels is not None:
target.scatter_(0, labels, scores)
return obj_nodes, question, target, hint_score, hint_o, qid, imgid, hint_flag
def __len__(self):
return self.datalen
class RandomSubsetSampler(Sampler):
def __init__(self, data_source, subset_size):
print(f"Creating RandomSubsetSampler of subset size {subset_size} and total size {len(data_source)}")
self.data_source = data_source
self.subset_size = subset_size
self.subset = torch.randperm(len(self.data_source))[:self.subset_size]
def __iter__(self):
return iter(self.subset.tolist())
def __len__(self):
return self.subset_size