-
Notifications
You must be signed in to change notification settings - Fork 0
/
data.py
164 lines (129 loc) · 6.43 KB
/
data.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
import random
import datasets
from datasets import Dataset
import pandas as pd
from datasets import load_dataset
import json
from datasets import Features, Sequence, ClassLabel, Value, Array2D
datasets.logging.set_verbosity(datasets.logging.ERROR)
def load(task_name, tokenizer, max_seq_length=512, is_id=False):
print("Loading {}".format(task_name))
if task_name == "rvl_cdip":
datasets = load_id()
elif task_name == 'ood':
datasets = load_ood()
def encode_example(example, pad_token_box=[0, 0, 0, 0]):
words = example['words']
normalized_word_boxes = example['bbox']
assert len(words) == len(normalized_word_boxes)
token_boxes = []
for word, box in zip(words, normalized_word_boxes):
word_tokens = tokenizer.tokenize(word)
token_boxes.extend([box] * len(word_tokens))
# Truncation of token_boxes
special_tokens_count = 2
if len(token_boxes) > max_seq_length - special_tokens_count:
token_boxes = token_boxes[: (max_seq_length - special_tokens_count)]
# add bounding boxes of cls + sep tokens
token_boxes = [[0, 0, 0, 0]] + token_boxes + [[1000, 1000, 1000, 1000]]
encoding = tokenizer(' '.join(words), padding='max_length', truncation=True)
# Padding of token_boxes up the bounding boxes to the sequence length.
input_ids = tokenizer(' '.join(words), truncation=True)["input_ids"]
padding_length = max_seq_length - len(input_ids)
token_boxes += [pad_token_box] * padding_length
encoding['bbox'] = token_boxes
encoding['label'] = example['label']
assert len(encoding['input_ids']) == max_seq_length
assert len(encoding['attention_mask']) == max_seq_length
assert len(encoding['token_type_ids']) == max_seq_length
assert len(encoding['bbox']) == max_seq_length
return encoding
def encode_ood_example(example, pad_token_box=[0, 0, 0, 0]):
words = example['words']
normalized_word_boxes = example['bbox']
assert len(words) == len(normalized_word_boxes)
token_boxes = []
for word, box in zip(words, normalized_word_boxes):
word_tokens = tokenizer.tokenize(word)
token_boxes.extend([box] * len(word_tokens))
# Truncation of token_boxes
special_tokens_count = 2
if len(token_boxes) > max_seq_length - special_tokens_count:
token_boxes = token_boxes[: (max_seq_length - special_tokens_count)]
# add bounding boxes of cls + sep tokens
token_boxes = [[0, 0, 0, 0]] + token_boxes + [[1000, 1000, 1000, 1000]]
encoding = tokenizer(' '.join(words), padding='max_length', truncation=True)
# Padding of token_boxes up the bounding boxes to the sequence length.
input_ids = tokenizer(' '.join(words), truncation=True)["input_ids"]
padding_length = max_seq_length - len(input_ids)
token_boxes += [pad_token_box] * padding_length
encoding['bbox'] = token_boxes
encoding['label'] = 0
assert len(encoding['input_ids']) == max_seq_length
assert len(encoding['attention_mask']) == max_seq_length
assert len(encoding['token_type_ids']) == max_seq_length
assert len(encoding['bbox']) == max_seq_length
return encoding
features = Features({
'input_ids': Sequence(feature=Value(dtype='int64')),
'bbox': Array2D(dtype="int64", shape=(512, 4)),
'attention_mask': Sequence(Value(dtype='int64')),
'token_type_ids': Sequence(Value(dtype='int64')),
'label': ClassLabel(num_classes=16),
'image_dir': Value(dtype='string'),
'words': Sequence(feature=Value(dtype='string')),
})
ood_features = Features({
'input_ids': Sequence(feature=Value(dtype='int64')),
'bbox': Array2D(dtype="int64", shape=(512, 4)),
'attention_mask': Sequence(Value(dtype='int64')),
'token_type_ids': Sequence(Value(dtype='int64')),
'label': ClassLabel(num_classes=1),
'image_dir': Value(dtype='string'),
'words': Sequence(feature=Value(dtype='string')),
})
if 'train' in datasets and is_id:
train_dataset = datasets['train'].map(lambda example: encode_example(example), features=features)
train_dataset.set_format(type="torch",
columns=['input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'label'])
else:
train_dataset = None
if 'validation' in datasets and is_id:
dev_dataset = datasets['validation'].map(lambda example: encode_example(example), features=features)
dev_dataset.set_format(type="torch", columns=['input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'label'])
else:
dev_dataset = None
if 'test' in datasets and is_id:
test_dataset = datasets['test'].map(lambda example: encode_example(example), features=features)
test_dataset.set_format(type="torch",
columns=['input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'label'])
elif 'test' in datasets and not is_id:
test_dataset = datasets['test'].map(lambda example: encode_ood_example(example), features=ood_features)
test_dataset.set_format(type="torch",
columns=['input_ids', 'bbox', 'attention_mask', 'token_type_ids', 'label'])
else:
test_dataset = None
return train_dataset, dev_dataset, test_dataset
def parse_json(example):
json_file = example['image_dir']
with open(json_file, 'r') as file:
ocr_result = json.load(file)
return ocr_result
def load_id():
train_df = pd.read_csv("data/processed_train.csv")
val_df = pd.read_csv("data/processed_val.csv")
test_df = pd.read_csv("data/processed_test.csv")
# train_temp = Dataset.from_pandas(train_df.iloc[0:50])
# val_temp = Dataset.from_pandas(val_df.iloc[0:20])
# test_temp = Dataset.from_pandas(test_df.iloc[0:20])
updated_train = Dataset.from_pandas(train_df[0:100000]).map(parse_json)
updated_val = Dataset.from_pandas(val_df).map(parse_json)
updated_test = Dataset.from_pandas(test_df).map(parse_json)
datasets = {'train': updated_train, 'validation': updated_val, 'test': updated_test}
return datasets
def load_ood():
ood_df = pd.read_csv("data/processed_ood.csv")
ood_df = Dataset.from_pandas(ood_df)
updated_ood = ood_df.map(parse_json)
datasets = {'test': updated_ood}
return datasets