def tokenize(self, max_length=config.max_question_len):
""" 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 entry in self.entries:
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 _load_dataset(dataroot, name, img_id2val, label2ans):
"""Load entries
img_id2val: dict {img_id -> val} val can be used to retrieve image or features
dataroot: root path of dataset
name: 'train', 'val', 'test'
"""
data_path = os.path.join(dataroot, name + 'set.json')
samples = json.load(open(data_path))
samples = sorted(samples, key=lambda x: x['qid'])
answer_path = os.path.join(dataroot, 'cache', '%s_target.pkl' % name)
answers = cPickle.load(open(answer_path, 'rb'))
answers = sorted(answers, key=lambda x: x['qid'])
utils.assert_eq(len(samples), len(answers))
entries = []
for sample, answer in zip(samples, answers):
utils.assert_eq(sample['qid'], answer['qid'])
utils.assert_eq(sample['image_name'], answer['image_name'])
img_id = sample['image_name']
if not COUNTING_ONLY or is_howmany(sample['question'], answer,
label2ans):
entries.append(_create_entry(img_id2val[img_id], sample, answer))
return entries
def _load_dataset(cache_path, name, img_id2val):
""" Load entries. img_id2val: dict {img_id -> val} ,
val can be used to retrieve image or features.
"""
train, val, test = False, False, False
if name == 'train':
train = True
elif name == 'val':
val = True
else:
test = True
question_path = utils.path_for(train=train,
val=val,
test=test,
question=True)
questions = json.load(open(question_path, 'r'))
if not config.cp_data:
questions = questions['questions']
questions = sorted(questions, key=lambda x: x['question_id'])
if test: # will be ignored anyway
answers = [{
'image_id': 0,
'question_id': 0,
'question_type': '',
'labels': [],
'scores': []
} for _ in range(len(questions))]
else:
answer_path = os.path.join(cache_path, '{}_target.json'.format(name))
answers = json.load(open(answer_path, 'r'))
answers = sorted(answers, key=lambda x: x['question_id'])
utils.assert_eq(len(questions), len(answers))
entries = []
for question, answer in zip(questions, answers):
if not test:
utils.assert_eq(question['question_id'], answer['question_id'])
utils.assert_eq(question['image_id'], answer['image_id'])
img_id = question['image_id']
entries.append(_create_entry(img_id2val[img_id], question, answer))
return entries