label_ops.py 18.6 KB
Newer Older
W
WenmuZhou 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
from __future__ import unicode_literals

import numpy as np
T
tink2123 已提交
21
import string
L
LDOUBLEV 已提交
22
import json
W
WenmuZhou 已提交
23 24 25 26 27 28 29 30 31 32 33 34 35


class ClsLabelEncode(object):
    def __init__(self, label_list, **kwargs):
        self.label_list = label_list

    def __call__(self, data):
        label = data['label']
        if label not in self.label_list:
            return None
        label = self.label_list.index(label)
        data['label'] = label
        return data
W
WenmuZhou 已提交
36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55


class DetLabelEncode(object):
    def __init__(self, **kwargs):
        pass

    def __call__(self, data):
        label = data['label']
        label = json.loads(label)
        nBox = len(label)
        boxes, txts, txt_tags = [], [], []
        for bno in range(0, nBox):
            box = label[bno]['points']
            txt = label[bno]['transcription']
            boxes.append(box)
            txts.append(txt)
            if txt in ['*', '###']:
                txt_tags.append(True)
            else:
                txt_tags.append(False)
M
MissPenguin 已提交
56
        boxes = self.expand_points_num(boxes)
L
LDOUBLEV 已提交
57 58
        #boxes = np.array(boxes, dtype=np.float32)
        #txt_tags = np.array(txt_tags, dtype=np.bool)
W
WenmuZhou 已提交
59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74

        data['polys'] = boxes
        data['texts'] = txts
        data['ignore_tags'] = txt_tags
        return data

    def order_points_clockwise(self, pts):
        rect = np.zeros((4, 2), dtype="float32")
        s = pts.sum(axis=1)
        rect[0] = pts[np.argmin(s)]
        rect[2] = pts[np.argmax(s)]
        diff = np.diff(pts, axis=1)
        rect[1] = pts[np.argmin(diff)]
        rect[3] = pts[np.argmax(diff)]
        return rect

M
MissPenguin 已提交
75 76 77 78 79 80 81 82 83 84 85
    def expand_points_num(self, boxes):
        max_points_num = 0
        for box in boxes:
            if len(box) > max_points_num:
                max_points_num = len(box)
        ex_boxes = []
        for box in boxes:
            ex_box = box + [box[-1]] * (max_points_num - len(box))
            ex_boxes.append(ex_box)
        return ex_boxes

W
WenmuZhou 已提交
86 87 88 89 90 91 92 93 94

class BaseRecLabelEncode(object):
    """ Convert between text-label and text-index """

    def __init__(self,
                 max_text_length,
                 character_dict_path=None,
                 character_type='ch',
                 use_space_char=False):
M
MissPenguin 已提交
95
        support_character_type = [
T
tink2123 已提交
96 97
            'ch', 'en', 'EN_symbol', 'french', 'german', 'japan', 'korean',
            'EN', 'it', 'xi', 'pu', 'ru', 'ar', 'ta', 'ug', 'fa', 'ur', 'rs',
T
tink2123 已提交
98
            'oc', 'rsc', 'bg', 'uk', 'be', 'te', 'ka', 'chinese_cht', 'hi',
T
tink2123 已提交
99
            'mr', 'ne', 'latin', 'arabic', 'cyrillic', 'devanagari'
M
MissPenguin 已提交
100
        ]
W
WenmuZhou 已提交
101
        assert character_type in support_character_type, "Only {} are supported now but get {}".format(
M
MissPenguin 已提交
102
            support_character_type, character_type)
W
WenmuZhou 已提交
103 104

        self.max_text_len = max_text_length
T
tink2123 已提交
105 106
        self.beg_str = "sos"
        self.end_str = "eos"
W
WenmuZhou 已提交
107 108 109
        if character_type == "en":
            self.character_str = "0123456789abcdefghijklmnopqrstuvwxyz"
            dict_character = list(self.character_str)
T
tink2123 已提交
110
        elif character_type == "EN_symbol":
T
tink2123 已提交
111 112 113 114
            # same with ASTER setting (use 94 char).
            self.character_str = string.printable[:-6]
            dict_character = list(self.character_str)
        elif character_type in support_character_type:
W
WenmuZhou 已提交
115
            self.character_str = ""
T
tink2123 已提交
116 117
            assert character_dict_path is not None, "character_dict_path should not be None when character_type is {}".format(
                character_type)
W
WenmuZhou 已提交
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
            with open(character_dict_path, "rb") as fin:
                lines = fin.readlines()
                for line in lines:
                    line = line.decode('utf-8').strip("\n").strip("\r\n")
                    self.character_str += line
            if use_space_char:
                self.character_str += " "
            dict_character = list(self.character_str)
        self.character_type = character_type
        dict_character = self.add_special_char(dict_character)
        self.dict = {}
        for i, char in enumerate(dict_character):
            self.dict[char] = i
        self.character = dict_character

    def add_special_char(self, dict_character):
        return dict_character

    def encode(self, text):
        """convert text-label into text-index.
        input:
            text: text labels of each image. [batch_size]

        output:
            text: concatenated text index for CTCLoss.
                    [sum(text_lengths)] = [text_index_0 + text_index_1 + ... + text_index_(n - 1)]
            length: length of each text. [batch_size]
        """
W
WenmuZhou 已提交
146
        if len(text) == 0 or len(text) > self.max_text_len:
W
WenmuZhou 已提交
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
            return None
        if self.character_type == "en":
            text = text.lower()
        text_list = []
        for char in text:
            if char not in self.dict:
                # logger = get_logger()
                # logger.warning('{} is not in dict'.format(char))
                continue
            text_list.append(self.dict[char])
        if len(text_list) == 0:
            return None
        return text_list


class CTCLabelEncode(BaseRecLabelEncode):
    """ Convert between text-label and text-index """

    def __init__(self,
                 max_text_length,
                 character_dict_path=None,
                 character_type='ch',
                 use_space_char=False,
                 **kwargs):
        super(CTCLabelEncode,
              self).__init__(max_text_length, character_dict_path,
                             character_type, use_space_char)

    def __call__(self, data):
        text = data['label']
        text = self.encode(text)
        if text is None:
            return None
        data['length'] = np.array(len(text))
        text = text + [0] * (self.max_text_len - len(text))
        data['label'] = np.array(text)
        return data

    def add_special_char(self, dict_character):
        dict_character = ['blank'] + dict_character
        return dict_character


J
Jethong 已提交
190
class E2ELabelEncodeTest(BaseRecLabelEncode):
J
Jethong 已提交
191 192 193 194 195 196
    def __init__(self,
                 max_text_length,
                 character_dict_path=None,
                 character_type='EN',
                 use_space_char=False,
                 **kwargs):
J
Jethong 已提交
197
        super(E2ELabelEncodeTest,
J
Jethong 已提交
198 199 200 201
              self).__init__(max_text_length, character_dict_path,
                             character_type, use_space_char)

    def __call__(self, data):
J
Jethong 已提交
202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219
        import json
        padnum = len(self.dict)
        label = data['label']
        label = json.loads(label)
        nBox = len(label)
        boxes, txts, txt_tags = [], [], []
        for bno in range(0, nBox):
            box = label[bno]['points']
            txt = label[bno]['transcription']
            boxes.append(box)
            txts.append(txt)
            if txt in ['*', '###']:
                txt_tags.append(True)
            else:
                txt_tags.append(False)
        boxes = np.array(boxes, dtype=np.float32)
        txt_tags = np.array(txt_tags, dtype=np.bool)
        data['polys'] = boxes
J
Jethong 已提交
220
        data['ignore_tags'] = txt_tags
J
Jethong 已提交
221
        temp_texts = []
J
Jethong 已提交
222
        for text in txts:
J
Jethong 已提交
223 224 225 226
            text = text.lower()
            text = self.encode(text)
            if text is None:
                return None
J
Jethong 已提交
227 228
            text = text + [padnum] * (self.max_text_len - len(text)
                                      )  # use 36 to pad
J
Jethong 已提交
229 230 231 232 233
            temp_texts.append(text)
        data['texts'] = np.array(temp_texts)
        return data


J
Jethong 已提交
234
class E2ELabelEncodeTrain(object):
J
Jethong 已提交
235 236
    def __init__(self, **kwargs):
        pass
J
Jethong 已提交
237 238

    def __call__(self, data):
J
Jethong 已提交
239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257
        import json
        label = data['label']
        label = json.loads(label)
        nBox = len(label)
        boxes, txts, txt_tags = [], [], []
        for bno in range(0, nBox):
            box = label[bno]['points']
            txt = label[bno]['transcription']
            boxes.append(box)
            txts.append(txt)
            if txt in ['*', '###']:
                txt_tags.append(True)
            else:
                txt_tags.append(False)
        boxes = np.array(boxes, dtype=np.float32)
        txt_tags = np.array(txt_tags, dtype=np.bool)

        data['polys'] = boxes
        data['texts'] = txts
J
Jethong 已提交
258
        data['ignore_tags'] = txt_tags
J
Jethong 已提交
259 260 261
        return data


W
WenmuZhou 已提交
262 263 264 265 266 267 268 269 270 271 272 273 274 275
class AttnLabelEncode(BaseRecLabelEncode):
    """ Convert between text-label and text-index """

    def __init__(self,
                 max_text_length,
                 character_dict_path=None,
                 character_type='ch',
                 use_space_char=False,
                 **kwargs):
        super(AttnLabelEncode,
              self).__init__(max_text_length, character_dict_path,
                             character_type, use_space_char)

    def add_special_char(self, dict_character):
L
LDOUBLEV 已提交
276 277 278
        self.beg_str = "sos"
        self.end_str = "eos"
        dict_character = [self.beg_str] + dict_character + [self.end_str]
W
WenmuZhou 已提交
279 280
        return dict_character

L
LDOUBLEV 已提交
281 282
    def __call__(self, data):
        text = data['label']
W
WenmuZhou 已提交
283
        text = self.encode(text)
L
LDOUBLEV 已提交
284 285
        if text is None:
            return None
L
LDOUBLEV 已提交
286
        if len(text) >= self.max_text_len:
L
LDOUBLEV 已提交
287 288 289
            return None
        data['length'] = np.array(len(text))
        text = [0] + text + [len(self.character) - 1] + [0] * (self.max_text_len
T
tink2123 已提交
290
                                                               - len(text) - 2)
L
LDOUBLEV 已提交
291 292 293 294 295 296 297
        data['label'] = np.array(text)
        return data

    def get_ignored_tokens(self):
        beg_idx = self.get_beg_end_flag_idx("beg")
        end_idx = self.get_beg_end_flag_idx("end")
        return [beg_idx, end_idx]
W
WenmuZhou 已提交
298 299 300 301 302 303 304 305 306 307

    def get_beg_end_flag_idx(self, beg_or_end):
        if beg_or_end == "beg":
            idx = np.array(self.dict[self.beg_str])
        elif beg_or_end == "end":
            idx = np.array(self.dict[self.end_str])
        else:
            assert False, "Unsupport type %s in get_beg_end_flag_idx" \
                          % beg_or_end
        return idx
T
tink2123 已提交
308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329


class SRNLabelEncode(BaseRecLabelEncode):
    """ Convert between text-label and text-index """

    def __init__(self,
                 max_text_length=25,
                 character_dict_path=None,
                 character_type='en',
                 use_space_char=False,
                 **kwargs):
        super(SRNLabelEncode,
              self).__init__(max_text_length, character_dict_path,
                             character_type, use_space_char)

    def add_special_char(self, dict_character):
        dict_character = dict_character + [self.beg_str, self.end_str]
        return dict_character

    def __call__(self, data):
        text = data['label']
        text = self.encode(text)
T
tink2123 已提交
330
        char_num = len(self.character)
T
tink2123 已提交
331 332 333 334 335
        if text is None:
            return None
        if len(text) > self.max_text_len:
            return None
        data['length'] = np.array(len(text))
T
tink2123 已提交
336
        text = text + [char_num - 1] * (self.max_text_len - len(text))
T
tink2123 已提交
337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353
        data['label'] = np.array(text)
        return data

    def get_ignored_tokens(self):
        beg_idx = self.get_beg_end_flag_idx("beg")
        end_idx = self.get_beg_end_flag_idx("end")
        return [beg_idx, end_idx]

    def get_beg_end_flag_idx(self, beg_or_end):
        if beg_or_end == "beg":
            idx = np.array(self.dict[self.beg_str])
        elif beg_or_end == "end":
            idx = np.array(self.dict[self.end_str])
        else:
            assert False, "Unsupport type %s in get_beg_end_flag_idx" \
                          % beg_or_end
        return idx
M
MissPenguin 已提交
354

L
LDOUBLEV 已提交
355

M
MissPenguin 已提交
356 357
class TableLabelEncode(object):
    """ Convert between text-label and text-index """
L
LDOUBLEV 已提交
358 359 360 361 362 363 364 365

    def __init__(self,
                 max_text_length,
                 max_elem_length,
                 max_cell_num,
                 character_dict_path,
                 span_weight=1.0,
                 **kwargs):
M
MissPenguin 已提交
366 367 368
        self.max_text_length = max_text_length
        self.max_elem_length = max_elem_length
        self.max_cell_num = max_cell_num
L
LDOUBLEV 已提交
369 370
        list_character, list_elem = self.load_char_elem_dict(
            character_dict_path)
M
MissPenguin 已提交
371 372 373 374 375 376 377 378 379
        list_character = self.add_special_char(list_character)
        list_elem = self.add_special_char(list_elem)
        self.dict_character = {}
        for i, char in enumerate(list_character):
            self.dict_character[char] = i
        self.dict_elem = {}
        for i, elem in enumerate(list_elem):
            self.dict_elem[elem] = i
        self.span_weight = span_weight
L
LDOUBLEV 已提交
380

M
MissPenguin 已提交
381 382 383 384 385
    def load_char_elem_dict(self, character_dict_path):
        list_character = []
        list_elem = []
        with open(character_dict_path, "rb") as fin:
            lines = fin.readlines()
W
WenmuZhou 已提交
386
            substr = lines[0].decode('utf-8').strip("\r\n").split("\t")
M
MissPenguin 已提交
387 388
            character_num = int(substr[0])
            elem_num = int(substr[1])
389

M
MissPenguin 已提交
390
            for cno in range(1, 1+character_num):
W
WenmuZhou 已提交
391
                character = lines[cno].decode('utf-8').strip("\r\n")
M
MissPenguin 已提交
392 393
                list_character.append(character)
            for eno in range(1+character_num, 1+character_num+elem_num):
W
WenmuZhou 已提交
394
                elem = lines[eno].decode('utf-8').strip("\r\n")
M
MissPenguin 已提交
395 396
                list_elem.append(elem)
        return list_character, list_elem
L
LDOUBLEV 已提交
397

M
MissPenguin 已提交
398 399 400 401 402
    def add_special_char(self, list_character):
        self.beg_str = "sos"
        self.end_str = "eos"
        list_character = [self.beg_str] + list_character + [self.end_str]
        return list_character
L
LDOUBLEV 已提交
403

M
MissPenguin 已提交
404 405 406 407 408 409
    def get_span_idx_list(self):
        span_idx_list = []
        for elem in self.dict_elem:
            if 'span' in elem:
                span_idx_list.append(self.dict_elem[elem])
        return span_idx_list
L
LDOUBLEV 已提交
410

M
MissPenguin 已提交
411 412 413 414 415 416 417 418
    def __call__(self, data):
        cells = data['cells']
        structure = data['structure']['tokens']
        structure = self.encode(structure, 'elem')
        if structure is None:
            return None
        elem_num = len(structure)
        structure = [0] + structure + [len(self.dict_elem) - 1]
L
LDOUBLEV 已提交
419 420
        structure = structure + [0] * (self.max_elem_length + 2 - len(structure)
                                       )
M
MissPenguin 已提交
421 422 423 424 425
        structure = np.array(structure)
        data['structure'] = structure
        elem_char_idx1 = self.dict_elem['<td>']
        elem_char_idx2 = self.dict_elem['<td']
        span_idx_list = self.get_span_idx_list()
L
LDOUBLEV 已提交
426 427
        td_idx_list = np.logical_or(structure == elem_char_idx1,
                                    structure == elem_char_idx2)
M
MissPenguin 已提交
428
        td_idx_list = np.where(td_idx_list)[0]
L
LDOUBLEV 已提交
429 430 431

        structure_mask = np.ones(
            (self.max_elem_length + 2, 1), dtype=np.float32)
M
MissPenguin 已提交
432
        bbox_list = np.zeros((self.max_elem_length + 2, 4), dtype=np.float32)
L
LDOUBLEV 已提交
433 434
        bbox_list_mask = np.zeros(
            (self.max_elem_length + 2, 1), dtype=np.float32)
M
MissPenguin 已提交
435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460
        img_height, img_width, img_ch = data['image'].shape
        if len(span_idx_list) > 0:
            span_weight = len(td_idx_list) * 1.0 / len(span_idx_list)
            span_weight = min(max(span_weight, 1.0), self.span_weight)
        for cno in range(len(cells)):
            if 'bbox' in cells[cno]:
                bbox = cells[cno]['bbox'].copy()
                bbox[0] = bbox[0] * 1.0 / img_width
                bbox[1] = bbox[1] * 1.0 / img_height
                bbox[2] = bbox[2] * 1.0 / img_width
                bbox[3] = bbox[3] * 1.0 / img_height
                td_idx = td_idx_list[cno]
                bbox_list[td_idx] = bbox
                bbox_list_mask[td_idx] = 1.0
                cand_span_idx = td_idx + 1
                if cand_span_idx < (self.max_elem_length + 2):
                    if structure[cand_span_idx] in span_idx_list:
                        structure_mask[cand_span_idx] = span_weight

        data['bbox_list'] = bbox_list
        data['bbox_list_mask'] = bbox_list_mask
        data['structure_mask'] = structure_mask
        char_beg_idx = self.get_beg_end_flag_idx('beg', 'char')
        char_end_idx = self.get_beg_end_flag_idx('end', 'char')
        elem_beg_idx = self.get_beg_end_flag_idx('beg', 'elem')
        elem_end_idx = self.get_beg_end_flag_idx('end', 'elem')
L
LDOUBLEV 已提交
461 462 463 464 465
        data['sp_tokens'] = np.array([
            char_beg_idx, char_end_idx, elem_beg_idx, elem_end_idx,
            elem_char_idx1, elem_char_idx2, self.max_text_length,
            self.max_elem_length, self.max_cell_num, elem_num
        ])
M
MissPenguin 已提交
466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516
        return data

    def encode(self, text, char_or_elem):
        """convert text-label into text-index.
        """
        if char_or_elem == "char":
            max_len = self.max_text_length
            current_dict = self.dict_character
        else:
            max_len = self.max_elem_length
            current_dict = self.dict_elem
        if len(text) > max_len:
            return None
        if len(text) == 0:
            if char_or_elem == "char":
                return [self.dict_character['space']]
            else:
                return None
        text_list = []
        for char in text:
            if char not in current_dict:
                return None
            text_list.append(current_dict[char])
        if len(text_list) == 0:
            if char_or_elem == "char":
                return [self.dict_character['space']]
            else:
                return None
        return text_list

    def get_ignored_tokens(self, char_or_elem):
        beg_idx = self.get_beg_end_flag_idx("beg", char_or_elem)
        end_idx = self.get_beg_end_flag_idx("end", char_or_elem)
        return [beg_idx, end_idx]

    def get_beg_end_flag_idx(self, beg_or_end, char_or_elem):
        if char_or_elem == "char":
            if beg_or_end == "beg":
                idx = np.array(self.dict_character[self.beg_str])
            elif beg_or_end == "end":
                idx = np.array(self.dict_character[self.end_str])
            else:
                assert False, "Unsupport type %s in get_beg_end_flag_idx of char" \
                              % beg_or_end
        elif char_or_elem == "elem":
            if beg_or_end == "beg":
                idx = np.array(self.dict_elem[self.beg_str])
            elif beg_or_end == "end":
                idx = np.array(self.dict_elem[self.end_str])
            else:
                assert False, "Unsupport type %s in get_beg_end_flag_idx of elem" \
L
LDOUBLEV 已提交
517
                              % beg_or_end
M
MissPenguin 已提交
518 519
        else:
            assert False, "Unsupport type %s in char_or_elem" \
L
LDOUBLEV 已提交
520
                              % char_or_elem
M
MissPenguin 已提交
521
        return idx