-
Notifications
You must be signed in to change notification settings - Fork 5
/
utils.py
177 lines (144 loc) · 5.25 KB
/
utils.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
import numpy as np
import torch
import collections
import torchvision.transforms as transforms
from PIL import Image
from torch.autograd import Variable
class ResizeNormalize(object):
def __init__(self, size, colored=False, interpolation=Image.BILINEAR):
self.size = size
self.colored = colored
self.interpolation = interpolation
self.toTensor = transforms.ToTensor()
def __call__(self, img):
img = img.resize(self.size, self.interpolation)
img = self.toTensor(img)
if not self.colored:
img.sub_(0.5).div_(0.5)
else:
pass
return img
class AlignCollate(object):
def __init__(self, im_h=32, im_w=100, keep_ratio=False, min_ratio=1):
self.im_h = im_h
self.im_w = im_w
self.keep_ratio = keep_ratio
self.min_ratio = min_ratio
def __call__(self, batch):
images, labels = zip(*batch)
im_h = self.im_h
im_w = self.im_w
if self.keep_ratio:
ratios = []
for image in images:
w, h = image.size
ratios.append(w / float(h))
ratios.sort()
max_ratio = ratios[-1]
im_w = int(np.floor(max_ratio * im_h))
im_w = max(im_h * self.min_ratio, im_w)
transform = ResizeNormalize(size=(im_w, im_h))
images = [transform(image) for image in images]
images = torch.cat([t.unsqueeze(0) for t in images], 0)
return images, labels
class strLabelConverter(object):
"""Convert between str and label.
NOTE:
Insert `blank` to the alphabet for CTC.
Args:
alphabet (str): set of the possible characters.
ignore_case (bool, default=True): whether or not to ignore all of the case.
"""
def __init__(self, alphabet, ignore_case=True):
self._ignore_case = ignore_case
if self._ignore_case:
alphabet = alphabet.lower()
self.alphabet = alphabet + '-' # for `-1` index
self.dict = {}
for i, char in enumerate(alphabet):
# NOTE: 0 is reserved for 'blank' required by wrap_ctc
self.dict[char] = i + 1
def encode(self, text):
"""Support batch or single str.
Args:
text (str or list of str): texts to convert.
Returns:
torch.IntTensor [length_0 + length_1 + ... length_{n - 1}]: encoded texts.
torch.IntTensor [n]: length of each text.
"""
if isinstance(text, str):
# print(text)
text_new = text
# flag = False
# for c in text:
# if flag and c != '\'':
# text_new += c
# if c == '\'':
# flag = not flag
# print(text_new)
text = [
self.dict[char.lower() if self._ignore_case else char]
for char in text_new
]
length = [len(text)]
elif isinstance(text, collections.Iterable):
length = [len(s) for s in text]
text = ''.join(text)
text, _ = self.encode(text)
return (torch.IntTensor(text), torch.IntTensor(length))
def decode(self, t, length, raw=False):
"""Decode encoded texts back into strs.
Args:
torch.IntTensor [length_0 + length_1 + ... length_{n - 1}]: encoded texts.
torch.IntTensor [n]: length of each text.
Raises:
AssertionError: when the texts and its length does not match.
Returns:
text (str or list of str): texts to convert.
"""
if length.numel() == 1:
length = length[0]
assert t.numel() == length, "text with length: {} does not match declared length: {}".format(t.numel(), length)
if raw:
return ''.join([self.alphabet[i - 1] for i in t])
else:
char_list = []
for i in range(length):
if t[i] != 0 and (not (i > 0 and t[i - 1] == t[i])):
char_list.append(self.alphabet[t[i] - 1])
return ''.join(char_list)
else:
# batch mode
assert t.numel() == length.sum(), "texts with length: {} does not match declared length: {}".format(t.numel(), length.sum())
texts = []
index = 0
for i in range(length.numel()):
l = length[i]
texts.append(
self.decode(
t[index:index + l], torch.IntTensor([l]), raw=raw))
index += l
return texts
class averager(object):
"""Compute average for `torch.Variable` and `torch.Tensor`. """
def __init__(self):
self.reset()
def add(self, v):
if isinstance(v, Variable):
count = v.data.numel()
v = v.data.sum()
elif isinstance(v, torch.Tensor):
count = v.numel()
v = v.sum()
self.n_count += count
self.sum += v
def reset(self):
self.n_count = 0
self.sum = 0
def val(self):
res = 0
if self.n_count != 0:
res = self.sum / float(self.n_count)
return res
def loadData(v, data):
v.resize_(data.size()).copy_(data)