utility.py 24.0 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
17
import sys
文幕地方's avatar
文幕地方 已提交
18
import platform
L
LDOUBLEV 已提交
19 20
import cv2
import numpy as np
21
import paddle
L
LDOUBLEV 已提交
22
from PIL import Image, ImageDraw, ImageFont
23
import math
W
WenmuZhou 已提交
24
from paddle import inference
L
LDOUBLEV 已提交
25 26
import time
from ppocr.utils.logging import get_logger
W
WenmuZhou 已提交
27

L
LDOUBLEV 已提交
28

29 30
def str2bool(v):
    return v.lower() in ("true", "t", "1")
L
LDOUBLEV 已提交
31 32


W
WenmuZhou 已提交
33
def init_args():
L
LDOUBLEV 已提交
34
    parser = argparse.ArgumentParser()
W
WenmuZhou 已提交
35
    # params for prediction engine
L
LDOUBLEV 已提交
36 37 38
    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)
39
    parser.add_argument("--min_subgraph_size", type=int, default=15)
L
LDOUBLEV 已提交
40
    parser.add_argument("--precision", type=str, default="fp32")
41
    parser.add_argument("--gpu_mem", type=int, default=500)
L
LDOUBLEV 已提交
42

W
WenmuZhou 已提交
43
    # params for text detector
L
LDOUBLEV 已提交
44 45 46
    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
47 48
    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
49

W
WenmuZhou 已提交
50
    # DB parmas
L
LDOUBLEV 已提交
51
    parser.add_argument("--det_db_thresh", type=float, default=0.3)
L
LDOUBLEV 已提交
52 53
    parser.add_argument("--det_db_box_thresh", type=float, default=0.6)
    parser.add_argument("--det_db_unclip_ratio", type=float, default=1.5)
L
LDOUBLEV 已提交
54
    parser.add_argument("--max_batch_size", type=int, default=10)
55
    parser.add_argument("--use_dilation", type=str2bool, default=False)
56
    parser.add_argument("--det_db_score_mode", type=str, default="fast")
W
WenmuZhou 已提交
57
    # EAST parmas
L
LDOUBLEV 已提交
58 59 60 61
    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
62
    # SAST parmas
L
licx 已提交
63 64
    parser.add_argument("--det_sast_score_thresh", type=float, default=0.5)
    parser.add_argument("--det_sast_nms_thresh", type=float, default=0.2)
65
    parser.add_argument("--det_sast_polygon", type=str2bool, default=False)
L
licx 已提交
66

W
WenmuZhou 已提交
67 68 69 70
    # PSE parmas
    parser.add_argument("--det_pse_thresh", type=float, default=0)
    parser.add_argument("--det_pse_box_thresh", type=float, default=0.85)
    parser.add_argument("--det_pse_min_area", type=float, default=16)
W
WenmuZhou 已提交
71
    parser.add_argument("--det_pse_box_type", type=str, default='box')
W
WenmuZhou 已提交
72 73
    parser.add_argument("--det_pse_scale", type=int, default=1)

W
WenmuZhou 已提交
74
    # params for text recognizer
L
LDOUBLEV 已提交
75 76
    parser.add_argument("--rec_algorithm", type=str, default='CRNN')
    parser.add_argument("--rec_model_dir", type=str)
T
fix bug  
tink2123 已提交
77
    parser.add_argument("--rec_image_shape", type=str, default="3, 32, 320")
78
    parser.add_argument("--rec_batch_num", type=int, default=6)
T
fix bug  
tink2123 已提交
79
    parser.add_argument("--max_text_length", type=int, default=25)
L
LDOUBLEV 已提交
80 81 82 83
    parser.add_argument(
        "--rec_char_dict_path",
        type=str,
        default="./ppocr/utils/ppocr_keys_v1.txt")
W
WenmuZhou 已提交
84 85
    parser.add_argument("--use_space_char", type=str2bool, default=True)
    parser.add_argument(
T
tink2123 已提交
86
        "--vis_font_path", type=str, default="./doc/fonts/simfang.ttf")
W
WenmuZhou 已提交
87
    parser.add_argument("--drop_score", type=float, default=0.5)
W
WenmuZhou 已提交
88

J
Jethong 已提交
89 90 91 92 93 94 95 96 97
    # 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
98
        "--e2e_char_dict_path", type=str, default="./ppocr/utils/ic15_dict.txt")
J
Jethong 已提交
99
    parser.add_argument("--e2e_pgnet_valid_set", type=str, default='totaltext')
100
    parser.add_argument("--e2e_pgnet_mode", type=str, default='fast')
J
Jethong 已提交
101

W
WenmuZhou 已提交
102 103 104 105 106
    # 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'])
107
    parser.add_argument("--cls_batch_num", type=int, default=6)
W
WenmuZhou 已提交
108 109 110
    parser.add_argument("--cls_thresh", type=float, default=0.9)

    parser.add_argument("--enable_mkldnn", type=str2bool, default=False)
L
LDOUBLEV 已提交
111
    parser.add_argument("--cpu_threads", type=int, default=10)
W
WenmuZhou 已提交
112
    parser.add_argument("--use_pdserving", type=str2bool, default=False)
113 114 115 116 117 118 119
    parser.add_argument("--warmup", type=str2bool, default=False)

    #
    parser.add_argument(
        "--draw_img_save_dir", type=str, default="./inference_results")
    parser.add_argument("--save_crop_res", type=str2bool, default=False)
    parser.add_argument("--crop_res_save_dir", type=str, default="./output")
W
WenmuZhou 已提交
120

121
    # multi-process
122
    parser.add_argument("--use_mp", type=str2bool, default=False)
123 124
    parser.add_argument("--total_process_num", type=int, default=1)
    parser.add_argument("--process_id", type=int, default=0)
125

126
    parser.add_argument("--benchmark", type=str2bool, default=False)
L
LDOUBLEV 已提交
127
    parser.add_argument("--save_log_path", type=str, default="./log_output/")
128

129
    parser.add_argument("--show_log", type=str2bool, default=True)
T
tink2123 已提交
130
    parser.add_argument("--use_onnx", type=str2bool, default=False)
W
WenmuZhou 已提交
131
    return parser
W
WenmuZhou 已提交
132

133

134
def parse_args():
W
WenmuZhou 已提交
135
    parser = init_args()
L
LDOUBLEV 已提交
136 137 138
    return parser.parse_args()


139 140 141 142 143
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
144
    elif mode == 'rec':
145
        model_dir = args.rec_model_dir
W
WenmuZhou 已提交
146 147
    elif mode == 'table':
        model_dir = args.table_model_dir
J
Jethong 已提交
148 149
    else:
        model_dir = args.e2e_model_dir
150 151 152 153

    if model_dir is None:
        logger.info("not find {} model file path {}".format(mode, model_dir))
        sys.exit(0)
T
tink2123 已提交
154 155 156 157 158 159 160 161
    if args.use_onnx:
        import onnxruntime as ort
        model_file_path = model_dir
        if not os.path.exists(model_file_path):
            raise ValueError("not find model file path {}".format(
                model_file_path))
        sess = ort.InferenceSession(model_file_path)
        return sess, sess.get_inputs()[0], None, None
L
LDOUBLEV 已提交
162

L
LDOUBLEV 已提交
163
    else:
T
tink2123 已提交
164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181
        model_file_path = model_dir + "/inference.pdmodel"
        params_file_path = model_dir + "/inference.pdiparams"
        if not os.path.exists(model_file_path):
            raise ValueError("not find model file path {}".format(
                model_file_path))
        if not os.path.exists(params_file_path):
            raise ValueError("not find params file path {}".format(
                params_file_path))

        config = inference.Config(model_file_path, params_file_path)

        if hasattr(args, 'precision'):
            if args.precision == "fp16" and args.use_tensorrt:
                precision = inference.PrecisionType.Half
            elif args.precision == "int8":
                precision = inference.PrecisionType.Int8
            else:
                precision = inference.PrecisionType.Float32
L
LDOUBLEV 已提交
182
        else:
T
tink2123 已提交
183 184 185 186 187
            precision = inference.PrecisionType.Float32

        if args.use_gpu:
            gpu_id = get_infer_gpuid()
            if gpu_id is None:
L
LDOUBLEV 已提交
188
                logger.warning(
L
LDOUBLEV 已提交
189
                    "GPU is not found in current device by nvidia-smi. Please check your device or ignore it if run on jeston."
T
tink2123 已提交
190 191 192 193
                )
            config.enable_use_gpu(args.gpu_mem, 0)
            if args.use_tensorrt:
                config.enable_tensorrt_engine(
L
LDOUBLEV 已提交
194
                    workspace_size=1 << 30,
T
tink2123 已提交
195 196 197 198
                    precision_mode=precision,
                    max_batch_size=args.max_batch_size,
                    min_subgraph_size=args.min_subgraph_size)
                # skip the minmum trt subgraph
L
fix trt  
LDOUBLEV 已提交
199
            use_dynamic_shape = True
L
fix  
LDOUBLEV 已提交
200
            if mode == "det":
T
tink2123 已提交
201 202 203 204 205 206 207 208 209 210 211 212 213 214 215
                min_input_shape = {
                    "x": [1, 3, 50, 50],
                    "conv2d_92.tmp_0": [1, 120, 20, 20],
                    "conv2d_91.tmp_0": [1, 24, 10, 10],
                    "conv2d_59.tmp_0": [1, 96, 20, 20],
                    "nearest_interp_v2_1.tmp_0": [1, 256, 10, 10],
                    "nearest_interp_v2_2.tmp_0": [1, 256, 20, 20],
                    "conv2d_124.tmp_0": [1, 256, 20, 20],
                    "nearest_interp_v2_3.tmp_0": [1, 64, 20, 20],
                    "nearest_interp_v2_4.tmp_0": [1, 64, 20, 20],
                    "nearest_interp_v2_5.tmp_0": [1, 64, 20, 20],
                    "elementwise_add_7": [1, 56, 2, 2],
                    "nearest_interp_v2_0.tmp_0": [1, 256, 2, 2]
                }
                max_input_shape = {
L
fix trt  
LDOUBLEV 已提交
216
                    "x": [1, 3, 1536, 1536],
T
tink2123 已提交
217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264
                    "conv2d_92.tmp_0": [1, 120, 400, 400],
                    "conv2d_91.tmp_0": [1, 24, 200, 200],
                    "conv2d_59.tmp_0": [1, 96, 400, 400],
                    "nearest_interp_v2_1.tmp_0": [1, 256, 200, 200],
                    "conv2d_124.tmp_0": [1, 256, 400, 400],
                    "nearest_interp_v2_2.tmp_0": [1, 256, 400, 400],
                    "nearest_interp_v2_3.tmp_0": [1, 64, 400, 400],
                    "nearest_interp_v2_4.tmp_0": [1, 64, 400, 400],
                    "nearest_interp_v2_5.tmp_0": [1, 64, 400, 400],
                    "elementwise_add_7": [1, 56, 400, 400],
                    "nearest_interp_v2_0.tmp_0": [1, 256, 400, 400]
                }
                opt_input_shape = {
                    "x": [1, 3, 640, 640],
                    "conv2d_92.tmp_0": [1, 120, 160, 160],
                    "conv2d_91.tmp_0": [1, 24, 80, 80],
                    "conv2d_59.tmp_0": [1, 96, 160, 160],
                    "nearest_interp_v2_1.tmp_0": [1, 256, 80, 80],
                    "nearest_interp_v2_2.tmp_0": [1, 256, 160, 160],
                    "conv2d_124.tmp_0": [1, 256, 160, 160],
                    "nearest_interp_v2_3.tmp_0": [1, 64, 160, 160],
                    "nearest_interp_v2_4.tmp_0": [1, 64, 160, 160],
                    "nearest_interp_v2_5.tmp_0": [1, 64, 160, 160],
                    "elementwise_add_7": [1, 56, 40, 40],
                    "nearest_interp_v2_0.tmp_0": [1, 256, 40, 40]
                }
                min_pact_shape = {
                    "nearest_interp_v2_26.tmp_0": [1, 256, 20, 20],
                    "nearest_interp_v2_27.tmp_0": [1, 64, 20, 20],
                    "nearest_interp_v2_28.tmp_0": [1, 64, 20, 20],
                    "nearest_interp_v2_29.tmp_0": [1, 64, 20, 20]
                }
                max_pact_shape = {
                    "nearest_interp_v2_26.tmp_0": [1, 256, 400, 400],
                    "nearest_interp_v2_27.tmp_0": [1, 64, 400, 400],
                    "nearest_interp_v2_28.tmp_0": [1, 64, 400, 400],
                    "nearest_interp_v2_29.tmp_0": [1, 64, 400, 400]
                }
                opt_pact_shape = {
                    "nearest_interp_v2_26.tmp_0": [1, 256, 160, 160],
                    "nearest_interp_v2_27.tmp_0": [1, 64, 160, 160],
                    "nearest_interp_v2_28.tmp_0": [1, 64, 160, 160],
                    "nearest_interp_v2_29.tmp_0": [1, 64, 160, 160]
                }
                min_input_shape.update(min_pact_shape)
                max_input_shape.update(max_pact_shape)
                opt_input_shape.update(opt_pact_shape)
            elif mode == "rec":
L
fix trt  
LDOUBLEV 已提交
265 266
                if args.rec_algorithm != "CRNN":
                    use_dynamic_shape = False
T
tink2123 已提交
267
                min_input_shape = {"x": [1, 3, 32, 10]}
L
LDOUBLEV 已提交
268
                max_input_shape = {"x": [args.rec_batch_num, 3, 32, 1536]}
T
tink2123 已提交
269 270 271
                opt_input_shape = {"x": [args.rec_batch_num, 3, 32, 320]}
            elif mode == "cls":
                min_input_shape = {"x": [1, 3, 48, 10]}
L
LDOUBLEV 已提交
272
                max_input_shape = {"x": [args.rec_batch_num, 3, 48, 1024]}
T
tink2123 已提交
273 274
                opt_input_shape = {"x": [args.rec_batch_num, 3, 48, 320]}
            else:
L
fix trt  
LDOUBLEV 已提交
275 276
                use_dynamic_shape = False
            if use_dynamic_shape:
A
andyjpaddle 已提交
277 278
                config.set_trt_dynamic_shape_info(
                    min_input_shape, max_input_shape, opt_input_shape)
L
LDOUBLEV 已提交
279

L
LDOUBLEV 已提交
280
        else:
T
tink2123 已提交
281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313
            config.disable_gpu()
            if hasattr(args, "cpu_threads"):
                config.set_cpu_math_library_num_threads(args.cpu_threads)
            else:
                # default cpu threads as 10
                config.set_cpu_math_library_num_threads(10)
            if args.enable_mkldnn:
                # cache 10 different shapes for mkldnn to avoid memory leak
                config.set_mkldnn_cache_capacity(10)
                config.enable_mkldnn()
                if args.precision == "fp16":
                    config.enable_mkldnn_bfloat16()
        # enable memory optim
        config.enable_memory_optim()
        config.disable_glog_info()

        config.delete_pass("conv_transpose_eltwiseadd_bn_fuse_pass")
        if mode == 'table':
            config.delete_pass("fc_fuse_pass")  # not supported for table
        config.switch_use_feed_fetch_ops(False)
        config.switch_ir_optim(True)

        # create predictor
        predictor = inference.create_predictor(config)
        input_names = predictor.get_input_names()
        for name in input_names:
            input_tensor = predictor.get_input_handle(name)
        output_names = predictor.get_output_names()
        output_tensors = []
        for output_name in output_names:
            output_tensor = predictor.get_output_handle(output_name)
            output_tensors.append(output_tensor)
        return predictor, input_tensor, output_tensors, config
314 315


316
def get_infer_gpuid():
文幕地方's avatar
文幕地方 已提交
317 318 319 320
    sysstr = platform.system()
    if sysstr == "Windows":
        return 0

R
ronny1996 已提交
321 322 323 324
    if not paddle.fluid.core.is_compiled_with_rocm():
        cmd = "env | grep CUDA_VISIBLE_DEVICES"
    else:
        cmd = "env | grep HIP_VISIBLE_DEVICES"
325 326 327 328 329 330 331 332
    env_cuda = os.popen(cmd).readlines()
    if len(env_cuda) == 0:
        return 0
    else:
        gpu_id = env_cuda[0].strip().split("=")[1]
        return int(gpu_id[0])


J
Jethong 已提交
333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348
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


349
def draw_text_det_res(dt_boxes, img_path):
L
LDOUBLEV 已提交
350 351 352 353
    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)
354
    return src_im
L
LDOUBLEV 已提交
355 356


357 358
def resize_img(img, input_size=600):
    """
359
    resize img and limit the longest side of the image to input_size
360 361 362 363 364
    """
    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
365 366
    img = cv2.resize(img, None, None, fx=im_scale, fy=im_scale)
    return img
367 368


W
WenmuZhou 已提交
369 370 371 372 373
def draw_ocr(image,
             boxes,
             txts=None,
             scores=None,
             drop_score=0.5,
L
LDOUBLEV 已提交
374
             font_path="./doc/fonts/simfang.ttf"):
375 376 377
    """
    Visualize the results of OCR detection and recognition
    args:
378
        image(Image|array): RGB image
379 380 381 382
        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
383
        font_path: the path of font which is used to draw text
384 385 386
    return(array):
        the visualized img
    """
L
LDOUBLEV 已提交
387 388
    if scores is None:
        scores = [1] * len(boxes)
W
WenmuZhou 已提交
389 390 391 392
    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])):
393
            continue
W
WenmuZhou 已提交
394
        box = np.reshape(np.array(boxes[i]), [-1, 1, 2]).astype(np.int64)
L
LDOUBLEV 已提交
395
        image = cv2.polylines(np.array(image), [box], True, (255, 0, 0), 2)
W
WenmuZhou 已提交
396
    if txts is not None:
L
LDOUBLEV 已提交
397
        img = np.array(resize_img(image, input_size=600))
398
        txt_img = text_visual(
W
WenmuZhou 已提交
399 400 401 402 403 404
            txts,
            scores,
            img_h=img.shape[0],
            img_w=600,
            threshold=drop_score,
            font_path=font_path)
405
        img = np.concatenate([np.array(img), np.array(txt_img)], axis=1)
L
LDOUBLEV 已提交
406 407
        return img
    return image
408 409


W
WenmuZhou 已提交
410 411 412 413 414 415
def draw_ocr_box_txt(image,
                     boxes,
                     txts,
                     scores=None,
                     drop_score=0.5,
                     font_path="./doc/simfang.ttf"):
416 417 418
    h, w = image.height, image.width
    img_left = image.copy()
    img_right = Image.new('RGB', (w, h), (255, 255, 255))
419 420

    import random
L
LDOUBLEV 已提交
421

422 423 424
    random.seed(0)
    draw_left = ImageDraw.Draw(img_left)
    draw_right = ImageDraw.Draw(img_right)
W
WenmuZhou 已提交
425 426 427
    for idx, (box, txt) in enumerate(zip(boxes, txts)):
        if scores is not None and scores[idx] < drop_score:
            continue
428 429
        color = (random.randint(0, 255), random.randint(0, 255),
                 random.randint(0, 255))
430
        draw_left.polygon(box, fill=color)
431 432 433 434 435 436 437 438 439 440
        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)
441 442
        if box_height > 2 * box_width:
            font_size = max(int(box_width * 0.9), 10)
W
WenmuZhou 已提交
443
            font = ImageFont.truetype(font_path, font_size, encoding="utf-8")
444 445 446
            cur_y = box[0][1]
            for c in txt:
                char_size = font.getsize(c)
447 448
                draw_right.text(
                    (box[0][0] + 3, cur_y), c, fill=(0, 0, 0), font=font)
449 450 451
                cur_y += char_size[1]
        else:
            font_size = max(int(box_height * 0.8), 10)
W
WenmuZhou 已提交
452
            font = ImageFont.truetype(font_path, font_size, encoding="utf-8")
453 454
            draw_right.text(
                [box[0][0], box[0][1]], txt, fill=(0, 0, 0), font=font)
455 456 457 458
    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))
459 460 461
    return np.array(img_show)


462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485
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
486 487 488 489 490 491
def text_visual(texts,
                scores,
                img_h=400,
                img_w=600,
                threshold=0.,
                font_path="./doc/simfang.ttf"):
492 493 494 495 496 497 498
    """
    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
499
        font_path: the path of font which is used to draw text
500 501 502 503 504 505 506 507 508
    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
509 510
        blank_img = Image.fromarray(blank_img).convert("RGB")
        draw_txt = ImageDraw.Draw(blank_img)
511
        return blank_img, draw_txt
L
LDOUBLEV 已提交
512

513 514 515 516
    blank_img, draw_txt = create_blank_img()

    font_size = 20
    txt_color = (0, 0, 0)
W
WenmuZhou 已提交
517
    font = ImageFont.truetype(font_path, font_size, encoding="utf-8")
518 519 520

    gap = font_size + 5
    txt_img_list = []
L
LDOUBLEV 已提交
521
    count, index = 1, 0
522 523
    for idx, txt in enumerate(texts):
        index += 1
524
        if scores[idx] < threshold or math.isnan(scores[idx]):
525 526 527 528 529 530 531 532 533 534 535
            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
536
            draw_txt.text((0, gap * count), new_txt, txt_color, font=font)
537 538 539 540 541
            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
542
            count += 1
543 544 545
        if first_line:
            new_txt = str(index) + ': ' + txt + '   ' + '%.3f' % (scores[idx])
        else:
546
            new_txt = "  " + txt + "  " + '%.3f' % (scores[idx])
L
LDOUBLEV 已提交
547
        draw_txt.text((0, gap * count), new_txt, txt_color, font=font)
548
        # whether add new blank img or not
L
LDOUBLEV 已提交
549
        if count >= img_h // gap - 1 and idx + 1 < len(texts):
550 551 552
            txt_img_list.append(np.array(blank_img))
            blank_img, draw_txt = create_blank_img()
            count = 0
L
LDOUBLEV 已提交
553
        count += 1
554 555 556 557 558 559
    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
560 561


D
dyning 已提交
562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580
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


W
WenmuZhou 已提交
581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615
def get_rotate_crop_image(img, points):
    '''
    img_height, img_width = img.shape[0:2]
    left = int(np.min(points[:, 0]))
    right = int(np.max(points[:, 0]))
    top = int(np.min(points[:, 1]))
    bottom = int(np.max(points[:, 1]))
    img_crop = img[top:bottom, left:right, :].copy()
    points[:, 0] = points[:, 0] - left
    points[:, 1] = points[:, 1] - top
    '''
    assert len(points) == 4, "shape of points must be 4*2"
    img_crop_width = int(
        max(
            np.linalg.norm(points[0] - points[1]),
            np.linalg.norm(points[2] - points[3])))
    img_crop_height = int(
        max(
            np.linalg.norm(points[0] - points[3]),
            np.linalg.norm(points[1] - points[2])))
    pts_std = np.float32([[0, 0], [img_crop_width, 0],
                          [img_crop_width, img_crop_height],
                          [0, img_crop_height]])
    M = cv2.getPerspectiveTransform(points, pts_std)
    dst_img = cv2.warpPerspective(
        img,
        M, (img_crop_width, img_crop_height),
        borderMode=cv2.BORDER_REPLICATE,
        flags=cv2.INTER_CUBIC)
    dst_img_height, dst_img_width = dst_img.shape[0:2]
    if dst_img_height * 1.0 / dst_img_width >= 1.5:
        dst_img = np.rot90(dst_img)
    return dst_img


616 617 618 619 620 621 622
def check_gpu(use_gpu):
    if use_gpu and not paddle.is_compiled_with_cuda():

        use_gpu = False
    return use_gpu


L
LDOUBLEV 已提交
623
if __name__ == '__main__':
L
LDOUBLEV 已提交
624
    pass
新手
引导
客服 返回
顶部