utility.py 19.1 KB
Newer Older
L
LDOUBLEV 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# 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.

import argparse
W
WenmuZhou 已提交
16
import os
W
WenmuZhou 已提交
17
import sys
L
LDOUBLEV 已提交
18 19
import cv2
import numpy as np
L
LDOUBLEV 已提交
20 21
import json
from PIL import Image, ImageDraw, ImageFont
22
import math
W
WenmuZhou 已提交
23
from paddle import inference
24 25 26
import time
from ppocr.utils.logging import get_logger
logger = get_logger()
L
LDOUBLEV 已提交
27 28 29 30 31 32 33


def parse_args():
    def str2bool(v):
        return v.lower() in ("true", "t", "1")

    parser = argparse.ArgumentParser()
W
WenmuZhou 已提交
34
    # params for prediction engine
L
LDOUBLEV 已提交
35 36 37
    parser.add_argument("--use_gpu", type=str2bool, default=True)
    parser.add_argument("--ir_optim", type=str2bool, default=True)
    parser.add_argument("--use_tensorrt", type=str2bool, default=False)
L
LDOUBLEV 已提交
38
    parser.add_argument("--use_fp16", type=str2bool, default=False)
L
LDOUBLEV 已提交
39
    parser.add_argument("--gpu_mem", type=int, default=500)
L
LDOUBLEV 已提交
40

W
WenmuZhou 已提交
41
    # params for text detector
L
LDOUBLEV 已提交
42 43 44
    parser.add_argument("--image_dir", type=str)
    parser.add_argument("--det_algorithm", type=str, default='DB')
    parser.add_argument("--det_model_dir", type=str)
W
WenmuZhou 已提交
45 46
    parser.add_argument("--det_limit_side_len", type=float, default=960)
    parser.add_argument("--det_limit_type", type=str, default='max')
L
LDOUBLEV 已提交
47

W
WenmuZhou 已提交
48
    # DB parmas
L
LDOUBLEV 已提交
49 50
    parser.add_argument("--det_db_thresh", type=float, default=0.3)
    parser.add_argument("--det_db_box_thresh", type=float, default=0.5)
W
WenmuZhou 已提交
51
    parser.add_argument("--det_db_unclip_ratio", type=float, default=1.6)
L
LDOUBLEV 已提交
52
    parser.add_argument("--max_batch_size", type=int, default=10)
L
LDOUBLEV 已提交
53
    parser.add_argument("--use_dilation", type=bool, default=False)
littletomatodonkey's avatar
littletomatodonkey 已提交
54
    parser.add_argument("--det_db_score_mode", type=str, default="fast")
W
WenmuZhou 已提交
55
    # EAST parmas
L
LDOUBLEV 已提交
56 57 58 59
    parser.add_argument("--det_east_score_thresh", type=float, default=0.8)
    parser.add_argument("--det_east_cover_thresh", type=float, default=0.1)
    parser.add_argument("--det_east_nms_thresh", type=float, default=0.2)

W
WenmuZhou 已提交
60
    # SAST parmas
L
licx 已提交
61 62
    parser.add_argument("--det_sast_score_thresh", type=float, default=0.5)
    parser.add_argument("--det_sast_nms_thresh", type=float, default=0.2)
63
    parser.add_argument("--det_sast_polygon", type=bool, default=False)
L
licx 已提交
64

W
WenmuZhou 已提交
65
    # params for text recognizer
L
LDOUBLEV 已提交
66 67
    parser.add_argument("--rec_algorithm", type=str, default='CRNN')
    parser.add_argument("--rec_model_dir", type=str)
T
fix bug  
tink2123 已提交
68 69
    parser.add_argument("--rec_image_shape", type=str, default="3, 32, 320")
    parser.add_argument("--rec_char_type", type=str, default='ch')
L
LDOUBLEV 已提交
70
    parser.add_argument("--rec_batch_num", type=int, default=6)
T
fix bug  
tink2123 已提交
71
    parser.add_argument("--max_text_length", type=int, default=25)
L
LDOUBLEV 已提交
72 73 74 75
    parser.add_argument(
        "--rec_char_dict_path",
        type=str,
        default="./ppocr/utils/ppocr_keys_v1.txt")
W
WenmuZhou 已提交
76 77
    parser.add_argument("--use_space_char", type=str2bool, default=True)
    parser.add_argument(
T
tink2123 已提交
78
        "--vis_font_path", type=str, default="./doc/fonts/simfang.ttf")
W
WenmuZhou 已提交
79
    parser.add_argument("--drop_score", type=float, default=0.5)
W
WenmuZhou 已提交
80

J
Jethong 已提交
81 82 83 84 85 86 87 88 89
    # params for e2e
    parser.add_argument("--e2e_algorithm", type=str, default='PGNet')
    parser.add_argument("--e2e_model_dir", type=str)
    parser.add_argument("--e2e_limit_side_len", type=float, default=768)
    parser.add_argument("--e2e_limit_type", type=str, default='max')

    # PGNet parmas
    parser.add_argument("--e2e_pgnet_score_thresh", type=float, default=0.5)
    parser.add_argument(
J
Jethong 已提交
90
        "--e2e_char_dict_path", type=str, default="./ppocr/utils/ic15_dict.txt")
J
Jethong 已提交
91
    parser.add_argument("--e2e_pgnet_valid_set", type=str, default='totaltext')
J
Jethong 已提交
92
    parser.add_argument("--e2e_pgnet_polygon", type=bool, default=True)
J
Jethong 已提交
93
    parser.add_argument("--e2e_pgnet_mode", type=str, default='fast')
J
Jethong 已提交
94

W
WenmuZhou 已提交
95 96 97 98 99
    # params for text classifier
    parser.add_argument("--use_angle_cls", type=str2bool, default=False)
    parser.add_argument("--cls_model_dir", type=str)
    parser.add_argument("--cls_image_shape", type=str, default="3, 48, 192")
    parser.add_argument("--label_list", type=list, default=['0', '180'])
L
LDOUBLEV 已提交
100
    parser.add_argument("--cls_batch_num", type=int, default=6)
W
WenmuZhou 已提交
101 102 103
    parser.add_argument("--cls_thresh", type=float, default=0.9)

    parser.add_argument("--enable_mkldnn", type=str2bool, default=False)
104
    parser.add_argument("--cpu_threads", type=int, default=10)
W
WenmuZhou 已提交
105 106
    parser.add_argument("--use_pdserving", type=str2bool, default=False)

littletomatodonkey's avatar
littletomatodonkey 已提交
107
    parser.add_argument("--use_mp", type=str2bool, default=False)
108 109 110
    parser.add_argument("--total_process_num", type=int, default=1)
    parser.add_argument("--process_id", type=int, default=0)

L
LDOUBLEV 已提交
111 112 113
    return parser.parse_args()


W
WenmuZhou 已提交
114 115 116 117 118
def create_predictor(args, mode, logger):
    if mode == "det":
        model_dir = args.det_model_dir
    elif mode == 'cls':
        model_dir = args.cls_model_dir
J
Jethong 已提交
119
    elif mode == 'rec':
W
WenmuZhou 已提交
120
        model_dir = args.rec_model_dir
J
Jethong 已提交
121 122
    else:
        model_dir = args.e2e_model_dir
W
WenmuZhou 已提交
123 124 125 126

    if model_dir is None:
        logger.info("not find {} model file path {}".format(mode, model_dir))
        sys.exit(0)
文幕地方's avatar
文幕地方 已提交
127 128
    model_file_path = model_dir + "/inference.pdmodel"
    params_file_path = model_dir + "/inference.pdiparams"
W
WenmuZhou 已提交
129 130 131 132 133 134 135
    if not os.path.exists(model_file_path):
        logger.info("not find model file path {}".format(model_file_path))
        sys.exit(0)
    if not os.path.exists(params_file_path):
        logger.info("not find params file path {}".format(params_file_path))
        sys.exit(0)

W
WenmuZhou 已提交
136
    config = inference.Config(model_file_path, params_file_path)
W
WenmuZhou 已提交
137 138 139

    if args.use_gpu:
        config.enable_use_gpu(args.gpu_mem, 0)
L
LDOUBLEV 已提交
140 141
        if args.use_tensorrt:
            config.enable_tensorrt_engine(
L
LDOUBLEV 已提交
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
                precision_mode=inference.PrecisionType.Float32,
                max_batch_size=args.max_batch_size,
                min_subgraph_size=3)  # skip the minmum trt subgraph 
        if mode == "det" and "mobile" in model_file_path:
            min_input_shape = {
                "x": [1, 3, 50, 50],
                "conv2d_92.tmp_0": [1, 96, 20, 20],
                "conv2d_91.tmp_0": [1, 96, 10, 10],
                "nearest_interp_v2_1.tmp_0": [1, 96, 10, 10],
                "nearest_interp_v2_2.tmp_0": [1, 96, 20, 20],
                "nearest_interp_v2_3.tmp_0": [1, 24, 20, 20],
                "nearest_interp_v2_4.tmp_0": [1, 24, 20, 20],
                "nearest_interp_v2_5.tmp_0": [1, 24, 20, 20],
                "elementwise_add_7": [1, 56, 2, 2],
                "nearest_interp_v2_0.tmp_0": [1, 96, 2, 2]
            }
            max_input_shape = {
                "x": [1, 3, 2000, 2000],
                "conv2d_92.tmp_0": [1, 96, 400, 400],
                "conv2d_91.tmp_0": [1, 96, 200, 200],
                "nearest_interp_v2_1.tmp_0": [1, 96, 200, 200],
                "nearest_interp_v2_2.tmp_0": [1, 96, 400, 400],
                "nearest_interp_v2_3.tmp_0": [1, 24, 400, 400],
                "nearest_interp_v2_4.tmp_0": [1, 24, 400, 400],
                "nearest_interp_v2_5.tmp_0": [1, 24, 400, 400],
                "elementwise_add_7": [1, 56, 400, 400],
                "nearest_interp_v2_0.tmp_0": [1, 96, 400, 400]
            }
            opt_input_shape = {
                "x": [1, 3, 640, 640],
                "conv2d_92.tmp_0": [1, 96, 160, 160],
                "conv2d_91.tmp_0": [1, 96, 80, 80],
                "nearest_interp_v2_1.tmp_0": [1, 96, 80, 80],
                "nearest_interp_v2_2.tmp_0": [1, 96, 160, 160],
                "nearest_interp_v2_3.tmp_0": [1, 24, 160, 160],
                "nearest_interp_v2_4.tmp_0": [1, 24, 160, 160],
                "nearest_interp_v2_5.tmp_0": [1, 24, 160, 160],
                "elementwise_add_7": [1, 56, 40, 40],
                "nearest_interp_v2_0.tmp_0": [1, 96, 40, 40]
            }
        if mode == "det" and "server" in model_file_path:
            min_input_shape = {
                "x": [1, 3, 50, 50],
                "conv2d_59.tmp_0": [1, 96, 20, 20],
                "nearest_interp_v2_2.tmp_0": [1, 96, 20, 20],
                "nearest_interp_v2_3.tmp_0": [1, 24, 20, 20],
                "nearest_interp_v2_4.tmp_0": [1, 24, 20, 20],
                "nearest_interp_v2_5.tmp_0": [1, 24, 20, 20]
            }
            max_input_shape = {
                "x": [1, 3, 2000, 2000],
                "conv2d_59.tmp_0": [1, 96, 400, 400],
                "nearest_interp_v2_2.tmp_0": [1, 96, 400, 400],
                "nearest_interp_v2_3.tmp_0": [1, 24, 400, 400],
                "nearest_interp_v2_4.tmp_0": [1, 24, 400, 400],
                "nearest_interp_v2_5.tmp_0": [1, 24, 400, 400]
            }
            opt_input_shape = {
                "x": [1, 3, 640, 640],
                "conv2d_59.tmp_0": [1, 96, 160, 160],
                "nearest_interp_v2_2.tmp_0": [1, 96, 160, 160],
                "nearest_interp_v2_3.tmp_0": [1, 24, 160, 160],
                "nearest_interp_v2_4.tmp_0": [1, 24, 160, 160],
                "nearest_interp_v2_5.tmp_0": [1, 24, 160, 160]
            }
        elif mode == "rec":
            min_input_shape = {"x": [args.rec_batch_num, 3, 32, 10]}
            max_input_shape = {"x": [args.rec_batch_num, 3, 32, 2000]}
            opt_input_shape = {"x": [args.rec_batch_num, 3, 32, 320]}
        elif mode == "cls":
            min_input_shape = {"x": [args.rec_batch_num, 3, 48, 10]}
            max_input_shape = {"x": [args.rec_batch_num, 3, 48, 2000]}
            opt_input_shape = {"x": [args.rec_batch_num, 3, 48, 320]}

        config.set_trt_dynamic_shape_info(min_input_shape, max_input_shape,
                                          opt_input_shape)

W
WenmuZhou 已提交
219 220
    else:
        config.disable_gpu()
221 222 223 224 225
        if hasattr(args, "cpu_threads"):
            config.set_cpu_math_library_num_threads(args.cpu_threads)
        else:
            config.set_cpu_math_library_num_threads(
                10)  # default cpu threads as 10
W
WenmuZhou 已提交
226 227 228 229 230
        if args.enable_mkldnn:
            # cache 10 different shapes for mkldnn to avoid memory leak
            config.set_mkldnn_cache_capacity(10)
            config.enable_mkldnn()

L
LDOUBLEV 已提交
231 232
    # enable memory optim
    config.enable_memory_optim()
W
WenmuZhou 已提交
233 234
    config.disable_glog_info()

W
WenmuZhou 已提交
235 236
    config.delete_pass("conv_transpose_eltwiseadd_bn_fuse_pass")
    config.switch_use_feed_fetch_ops(False)
W
WenmuZhou 已提交
237

W
WenmuZhou 已提交
238 239
    # create predictor
    predictor = inference.create_predictor(config)
W
WenmuZhou 已提交
240 241
    input_names = predictor.get_input_names()
    for name in input_names:
W
WenmuZhou 已提交
242
        input_tensor = predictor.get_input_handle(name)
W
WenmuZhou 已提交
243 244 245
    output_names = predictor.get_output_names()
    output_tensors = []
    for output_name in output_names:
W
WenmuZhou 已提交
246
        output_tensor = predictor.get_output_handle(output_name)
W
WenmuZhou 已提交
247 248 249 250
        output_tensors.append(output_tensor)
    return predictor, input_tensor, output_tensors


J
Jethong 已提交
251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266
def draw_e2e_res(dt_boxes, strs, img_path):
    src_im = cv2.imread(img_path)
    for box, str in zip(dt_boxes, strs):
        box = box.astype(np.int32).reshape((-1, 1, 2))
        cv2.polylines(src_im, [box], True, color=(255, 255, 0), thickness=2)
        cv2.putText(
            src_im,
            str,
            org=(int(box[0, 0, 0]), int(box[0, 0, 1])),
            fontFace=cv2.FONT_HERSHEY_COMPLEX,
            fontScale=0.7,
            color=(0, 255, 0),
            thickness=1)
    return src_im


L
LDOUBLEV 已提交
267
def draw_text_det_res(dt_boxes, img_path):
L
LDOUBLEV 已提交
268 269 270 271
    src_im = cv2.imread(img_path)
    for box in dt_boxes:
        box = np.array(box).astype(np.int32).reshape(-1, 2)
        cv2.polylines(src_im, [box], True, color=(255, 255, 0), thickness=2)
L
LDOUBLEV 已提交
272
    return src_im
L
LDOUBLEV 已提交
273 274


L
LDOUBLEV 已提交
275 276
def resize_img(img, input_size=600):
    """
L
LDOUBLEV 已提交
277
    resize img and limit the longest side of the image to input_size
L
LDOUBLEV 已提交
278 279 280 281 282
    """
    img = np.array(img)
    im_shape = img.shape
    im_size_max = np.max(im_shape[0:2])
    im_scale = float(input_size) / float(im_size_max)
W
WenmuZhou 已提交
283 284
    img = cv2.resize(img, None, None, fx=im_scale, fy=im_scale)
    return img
L
LDOUBLEV 已提交
285 286


W
WenmuZhou 已提交
287 288 289 290 291
def draw_ocr(image,
             boxes,
             txts=None,
             scores=None,
             drop_score=0.5,
L
LDOUBLEV 已提交
292
             font_path="./doc/fonts/simfang.ttf"):
293 294 295
    """
    Visualize the results of OCR detection and recognition
    args:
L
LDOUBLEV 已提交
296
        image(Image|array): RGB image
297 298 299 300
        boxes(list): boxes with shape(N, 4, 2)
        txts(list): the texts
        scores(list): txxs corresponding scores
        drop_score(float): only scores greater than drop_threshold will be visualized
W
WenmuZhou 已提交
301
        font_path: the path of font which is used to draw text
302 303 304
    return(array):
        the visualized img
    """
L
LDOUBLEV 已提交
305 306
    if scores is None:
        scores = [1] * len(boxes)
W
WenmuZhou 已提交
307 308 309 310
    box_num = len(boxes)
    for i in range(box_num):
        if scores is not None and (scores[i] < drop_score or
                                   math.isnan(scores[i])):
L
LDOUBLEV 已提交
311
            continue
W
WenmuZhou 已提交
312
        box = np.reshape(np.array(boxes[i]), [-1, 1, 2]).astype(np.int64)
L
LDOUBLEV 已提交
313
        image = cv2.polylines(np.array(image), [box], True, (255, 0, 0), 2)
W
WenmuZhou 已提交
314
    if txts is not None:
L
LDOUBLEV 已提交
315
        img = np.array(resize_img(image, input_size=600))
316
        txt_img = text_visual(
W
WenmuZhou 已提交
317 318 319 320 321 322
            txts,
            scores,
            img_h=img.shape[0],
            img_w=600,
            threshold=drop_score,
            font_path=font_path)
323
        img = np.concatenate([np.array(img), np.array(txt_img)], axis=1)
L
LDOUBLEV 已提交
324 325
        return img
    return image
326 327


W
WenmuZhou 已提交
328 329 330 331 332 333
def draw_ocr_box_txt(image,
                     boxes,
                     txts,
                     scores=None,
                     drop_score=0.5,
                     font_path="./doc/simfang.ttf"):
334 335 336
    h, w = image.height, image.width
    img_left = image.copy()
    img_right = Image.new('RGB', (w, h), (255, 255, 255))
337 338

    import random
L
LDOUBLEV 已提交
339

340 341 342
    random.seed(0)
    draw_left = ImageDraw.Draw(img_left)
    draw_right = ImageDraw.Draw(img_right)
W
WenmuZhou 已提交
343 344 345
    for idx, (box, txt) in enumerate(zip(boxes, txts)):
        if scores is not None and scores[idx] < drop_score:
            continue
T
tink2123 已提交
346 347
        color = (random.randint(0, 255), random.randint(0, 255),
                 random.randint(0, 255))
348
        draw_left.polygon(box, fill=color)
T
tink2123 已提交
349 350 351 352 353 354 355 356 357 358
        draw_right.polygon(
            [
                box[0][0], box[0][1], box[1][0], box[1][1], box[2][0],
                box[2][1], box[3][0], box[3][1]
            ],
            outline=color)
        box_height = math.sqrt((box[0][0] - box[3][0])**2 + (box[0][1] - box[3][
            1])**2)
        box_width = math.sqrt((box[0][0] - box[1][0])**2 + (box[0][1] - box[1][
            1])**2)
359 360
        if box_height > 2 * box_width:
            font_size = max(int(box_width * 0.9), 10)
W
WenmuZhou 已提交
361
            font = ImageFont.truetype(font_path, font_size, encoding="utf-8")
362 363 364
            cur_y = box[0][1]
            for c in txt:
                char_size = font.getsize(c)
T
tink2123 已提交
365 366
                draw_right.text(
                    (box[0][0] + 3, cur_y), c, fill=(0, 0, 0), font=font)
367 368 369
                cur_y += char_size[1]
        else:
            font_size = max(int(box_height * 0.8), 10)
W
WenmuZhou 已提交
370
            font = ImageFont.truetype(font_path, font_size, encoding="utf-8")
T
tink2123 已提交
371 372
            draw_right.text(
                [box[0][0], box[0][1]], txt, fill=(0, 0, 0), font=font)
373 374 375 376
    img_left = Image.blend(image, img_left, 0.5)
    img_show = Image.new('RGB', (w * 2, h), (255, 255, 255))
    img_show.paste(img_left, (0, 0, w, h))
    img_show.paste(img_right, (w, 0, w * 2, h))
377 378 379
    return np.array(img_show)


380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403
def str_count(s):
    """
    Count the number of Chinese characters,
    a single English character and a single number
    equal to half the length of Chinese characters.
    args:
        s(string): the input of string
    return(int):
        the number of Chinese characters
    """
    import string
    count_zh = count_pu = 0
    s_len = len(s)
    en_dg_count = 0
    for c in s:
        if c in string.ascii_letters or c.isdigit() or c.isspace():
            en_dg_count += 1
        elif c.isalpha():
            count_zh += 1
        else:
            count_pu += 1
    return s_len - math.ceil(en_dg_count / 2)


W
WenmuZhou 已提交
404 405 406 407 408 409
def text_visual(texts,
                scores,
                img_h=400,
                img_w=600,
                threshold=0.,
                font_path="./doc/simfang.ttf"):
410 411 412 413 414 415 416
    """
    create new blank img and draw txt on it
    args:
        texts(list): the text will be draw
        scores(list|None): corresponding score of each txt
        img_h(int): the height of blank img
        img_w(int): the width of blank img
W
WenmuZhou 已提交
417
        font_path: the path of font which is used to draw text
418 419 420 421 422 423 424 425 426
    return(array):
    """
    if scores is not None:
        assert len(texts) == len(
            scores), "The number of txts and corresponding scores must match"

    def create_blank_img():
        blank_img = np.ones(shape=[img_h, img_w], dtype=np.int8) * 255
        blank_img[:, img_w - 1:] = 0
L
LDOUBLEV 已提交
427 428
        blank_img = Image.fromarray(blank_img).convert("RGB")
        draw_txt = ImageDraw.Draw(blank_img)
429
        return blank_img, draw_txt
L
LDOUBLEV 已提交
430

431 432 433 434
    blank_img, draw_txt = create_blank_img()

    font_size = 20
    txt_color = (0, 0, 0)
W
WenmuZhou 已提交
435
    font = ImageFont.truetype(font_path, font_size, encoding="utf-8")
436 437 438

    gap = font_size + 5
    txt_img_list = []
L
LDOUBLEV 已提交
439
    count, index = 1, 0
440 441
    for idx, txt in enumerate(texts):
        index += 1
L
LDOUBLEV 已提交
442
        if scores[idx] < threshold or math.isnan(scores[idx]):
443 444 445 446 447 448 449 450 451 452 453
            index -= 1
            continue
        first_line = True
        while str_count(txt) >= img_w // font_size - 4:
            tmp = txt
            txt = tmp[:img_w // font_size - 4]
            if first_line:
                new_txt = str(index) + ': ' + txt
                first_line = False
            else:
                new_txt = '    ' + txt
L
LDOUBLEV 已提交
454
            draw_txt.text((0, gap * count), new_txt, txt_color, font=font)
455 456 457 458 459
            txt = tmp[img_w // font_size - 4:]
            if count >= img_h // gap - 1:
                txt_img_list.append(np.array(blank_img))
                blank_img, draw_txt = create_blank_img()
                count = 0
L
LDOUBLEV 已提交
460
            count += 1
461 462 463
        if first_line:
            new_txt = str(index) + ': ' + txt + '   ' + '%.3f' % (scores[idx])
        else:
L
LDOUBLEV 已提交
464
            new_txt = "  " + txt + "  " + '%.3f' % (scores[idx])
L
LDOUBLEV 已提交
465
        draw_txt.text((0, gap * count), new_txt, txt_color, font=font)
466
        # whether add new blank img or not
L
LDOUBLEV 已提交
467
        if count >= img_h // gap - 1 and idx + 1 < len(texts):
468 469 470
            txt_img_list.append(np.array(blank_img))
            blank_img, draw_txt = create_blank_img()
            count = 0
L
LDOUBLEV 已提交
471
        count += 1
472 473 474 475 476 477
    txt_img_list.append(np.array(blank_img))
    if len(txt_img_list) == 1:
        blank_img = np.array(txt_img_list[0])
    else:
        blank_img = np.concatenate(txt_img_list, axis=1)
    return np.array(blank_img)
L
LDOUBLEV 已提交
478 479


D
dyning 已提交
480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498
def base64_to_cv2(b64str):
    import base64
    data = base64.b64decode(b64str.encode('utf8'))
    data = np.fromstring(data, np.uint8)
    data = cv2.imdecode(data, cv2.IMREAD_COLOR)
    return data


def draw_boxes(image, boxes, scores=None, drop_score=0.5):
    if scores is None:
        scores = [1] * len(boxes)
    for (box, score) in zip(boxes, scores):
        if score < drop_score:
            continue
        box = np.reshape(np.array(box), [-1, 1, 2]).astype(np.int64)
        image = cv2.polylines(np.array(image), [box], True, (255, 0, 0), 2)
    return image


L
LDOUBLEV 已提交
499
if __name__ == '__main__':
L
LDOUBLEV 已提交
500
    pass