utility.py 22.6 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
L
LDOUBLEV 已提交
24 25
import time
from ppocr.utils.logging import get_logger
W
WenmuZhou 已提交
26

L
LDOUBLEV 已提交
27

28 29
def str2bool(v):
    return v.lower() in ("true", "t", "1")
L
LDOUBLEV 已提交
30 31


W
WenmuZhou 已提交
32
def init_args():
L
LDOUBLEV 已提交
33
    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("--min_subgraph_size", type=int, default=15)
L
LDOUBLEV 已提交
39
    parser.add_argument("--precision", type=str, default="fp32")
L
LDOUBLEV 已提交
40
    parser.add_argument("--gpu_mem", type=int, default=500)
L
LDOUBLEV 已提交
41

W
WenmuZhou 已提交
42
    # params for text detector
L
LDOUBLEV 已提交
43 44 45
    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
46 47
    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
48

W
WenmuZhou 已提交
49
    # DB parmas
L
LDOUBLEV 已提交
50
    parser.add_argument("--det_db_thresh", type=float, default=0.3)
L
LDOUBLEV 已提交
51 52
    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
53
    parser.add_argument("--max_batch_size", type=int, default=10)
littletomatodonkey's avatar
littletomatodonkey 已提交
54
    parser.add_argument("--use_dilation", type=str2bool, default=False)
littletomatodonkey's avatar
littletomatodonkey 已提交
55
    parser.add_argument("--det_db_score_mode", type=str, default="fast")
W
WenmuZhou 已提交
56
    # EAST parmas
L
LDOUBLEV 已提交
57 58 59 60
    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
61
    # SAST parmas
L
licx 已提交
62 63
    parser.add_argument("--det_sast_score_thresh", type=float, default=0.5)
    parser.add_argument("--det_sast_nms_thresh", type=float, default=0.2)
littletomatodonkey's avatar
littletomatodonkey 已提交
64
    parser.add_argument("--det_sast_polygon", type=str2bool, default=False)
L
licx 已提交
65

W
WenmuZhou 已提交
66 67 68 69
    # 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
70
    parser.add_argument("--det_pse_box_type", type=str, default='box')
W
WenmuZhou 已提交
71 72
    parser.add_argument("--det_pse_scale", type=int, default=1)

W
WenmuZhou 已提交
73
    # params for text recognizer
L
LDOUBLEV 已提交
74 75
    parser.add_argument("--rec_algorithm", type=str, default='CRNN')
    parser.add_argument("--rec_model_dir", type=str)
T
fix bug  
tink2123 已提交
76
    parser.add_argument("--rec_image_shape", type=str, default="3, 32, 320")
L
LDOUBLEV 已提交
77
    parser.add_argument("--rec_batch_num", type=int, default=6)
T
fix bug  
tink2123 已提交
78
    parser.add_argument("--max_text_length", type=int, default=25)
L
LDOUBLEV 已提交
79 80 81 82
    parser.add_argument(
        "--rec_char_dict_path",
        type=str,
        default="./ppocr/utils/ppocr_keys_v1.txt")
W
WenmuZhou 已提交
83 84
    parser.add_argument("--use_space_char", type=str2bool, default=True)
    parser.add_argument(
T
tink2123 已提交
85
        "--vis_font_path", type=str, default="./doc/fonts/simfang.ttf")
W
WenmuZhou 已提交
86
    parser.add_argument("--drop_score", type=float, default=0.5)
W
WenmuZhou 已提交
87

J
Jethong 已提交
88 89 90 91 92 93 94 95 96
    # 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
97
        "--e2e_char_dict_path", type=str, default="./ppocr/utils/ic15_dict.txt")
J
Jethong 已提交
98
    parser.add_argument("--e2e_pgnet_valid_set", type=str, default='totaltext')
littletomatodonkey's avatar
littletomatodonkey 已提交
99
    parser.add_argument("--e2e_pgnet_polygon", type=str2bool, default=True)
J
Jethong 已提交
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'])
L
LDOUBLEV 已提交
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)
L
LDOUBLEV 已提交
113
    parser.add_argument("--warmup", type=str2bool, default=True)
W
WenmuZhou 已提交
114

L
LDOUBLEV 已提交
115
    # multi-process
littletomatodonkey's avatar
littletomatodonkey 已提交
116
    parser.add_argument("--use_mp", type=str2bool, default=False)
117 118
    parser.add_argument("--total_process_num", type=int, default=1)
    parser.add_argument("--process_id", type=int, default=0)
W
WenmuZhou 已提交
119

littletomatodonkey's avatar
littletomatodonkey 已提交
120
    parser.add_argument("--benchmark", type=str2bool, default=False)
L
LDOUBLEV 已提交
121
    parser.add_argument("--save_log_path", type=str, default="./log_output/")
D
Double_V 已提交
122

W
WenmuZhou 已提交
123
    parser.add_argument("--show_log", type=str2bool, default=True)
W
WenmuZhou 已提交
124
    return parser
W
WenmuZhou 已提交
125

126

127
def parse_args():
W
WenmuZhou 已提交
128
    parser = init_args()
L
LDOUBLEV 已提交
129 130 131
    return parser.parse_args()


W
WenmuZhou 已提交
132 133 134 135 136
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
137
    elif mode == 'rec':
W
WenmuZhou 已提交
138
        model_dir = args.rec_model_dir
W
WenmuZhou 已提交
139 140
    elif mode == 'table':
        model_dir = args.table_model_dir
J
Jethong 已提交
141 142
    else:
        model_dir = args.e2e_model_dir
W
WenmuZhou 已提交
143 144 145 146

    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
147 148
    model_file_path = model_dir + "/inference.pdmodel"
    params_file_path = model_dir + "/inference.pdiparams"
W
WenmuZhou 已提交
149
    if not os.path.exists(model_file_path):
L
LDOUBLEV 已提交
150
        raise ValueError("not find model file path {}".format(model_file_path))
W
WenmuZhou 已提交
151
    if not os.path.exists(params_file_path):
L
LDOUBLEV 已提交
152 153
        raise ValueError("not find params file path {}".format(
            params_file_path))
W
WenmuZhou 已提交
154

W
WenmuZhou 已提交
155
    config = inference.Config(model_file_path, params_file_path)
W
WenmuZhou 已提交
156

L
LDOUBLEV 已提交
157 158 159 160 161 162 163 164 165 166
    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
    else:
        precision = inference.PrecisionType.Float32

W
WenmuZhou 已提交
167
    if args.use_gpu:
168 169 170 171 172
        gpu_id = get_infer_gpuid()
        if gpu_id is None:
            raise ValueError(
                "Not found GPU in current device. Please check your device or set args.use_gpu as False"
            )
W
WenmuZhou 已提交
173
        config.enable_use_gpu(args.gpu_mem, 0)
L
LDOUBLEV 已提交
174 175
        if args.use_tensorrt:
            config.enable_tensorrt_engine(
L
LDOUBLEV 已提交
176
                workspace_size=1 << 30,
D
Double_V 已提交
177
                precision_mode=precision,
L
LDOUBLEV 已提交
178
                max_batch_size=args.max_batch_size,
L
LDOUBLEV 已提交
179 180
                min_subgraph_size=args.min_subgraph_size)
            # skip the minmum trt subgraph
L
LDOUBLEV 已提交
181
        if mode == "det":
L
LDOUBLEV 已提交
182 183
            min_input_shape = {
                "x": [1, 3, 50, 50],
F
fengshuai03 已提交
184 185
                "conv2d_92.tmp_0": [1, 120, 20, 20],
                "conv2d_91.tmp_0": [1, 24, 10, 10],
L
LDOUBLEV 已提交
186
                "conv2d_59.tmp_0": [1, 96, 20, 20],
F
fengshuai03 已提交
187 188 189 190 191 192
                "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],
L
LDOUBLEV 已提交
193
                "elementwise_add_7": [1, 56, 2, 2],
F
fengshuai03 已提交
194
                "nearest_interp_v2_0.tmp_0": [1, 256, 2, 2]
L
LDOUBLEV 已提交
195 196 197
            }
            max_input_shape = {
                "x": [1, 3, 2000, 2000],
F
fengshuai03 已提交
198 199
                "conv2d_92.tmp_0": [1, 120, 400, 400],
                "conv2d_91.tmp_0": [1, 24, 200, 200],
L
LDOUBLEV 已提交
200
                "conv2d_59.tmp_0": [1, 96, 400, 400],
F
fengshuai03 已提交
201
                "nearest_interp_v2_1.tmp_0": [1, 256, 200, 200],
L
LDOUBLEV 已提交
202
                "conv2d_124.tmp_0": [1, 256, 400, 400],
F
fengshuai03 已提交
203 204 205 206
                "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],
L
LDOUBLEV 已提交
207
                "elementwise_add_7": [1, 56, 400, 400],
F
fengshuai03 已提交
208
                "nearest_interp_v2_0.tmp_0": [1, 256, 400, 400]
L
LDOUBLEV 已提交
209 210 211
            }
            opt_input_shape = {
                "x": [1, 3, 640, 640],
F
fengshuai03 已提交
212 213
                "conv2d_92.tmp_0": [1, 120, 160, 160],
                "conv2d_91.tmp_0": [1, 24, 80, 80],
L
LDOUBLEV 已提交
214
                "conv2d_59.tmp_0": [1, 96, 160, 160],
F
fengshuai03 已提交
215 216
                "nearest_interp_v2_1.tmp_0": [1, 256, 80, 80],
                "nearest_interp_v2_2.tmp_0": [1, 256, 160, 160],
L
LDOUBLEV 已提交
217
                "conv2d_124.tmp_0": [1, 256, 160, 160],
F
fengshuai03 已提交
218 219 220
                "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],
L
LDOUBLEV 已提交
221
                "elementwise_add_7": [1, 56, 40, 40],
F
fengshuai03 已提交
222
                "nearest_interp_v2_0.tmp_0": [1, 256, 40, 40]
L
LDOUBLEV 已提交
223
            }
F
fengshuai03 已提交
224
            min_pact_shape = {
littletomatodonkey's avatar
littletomatodonkey 已提交
225 226 227 228
                "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]
F
fengshuai03 已提交
229 230
            }
            max_pact_shape = {
littletomatodonkey's avatar
littletomatodonkey 已提交
231 232 233 234
                "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]
F
fengshuai03 已提交
235 236
            }
            opt_pact_shape = {
littletomatodonkey's avatar
littletomatodonkey 已提交
237 238 239 240
                "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]
F
fengshuai03 已提交
241 242 243 244
            }
            min_input_shape.update(min_pact_shape)
            max_input_shape.update(max_pact_shape)
            opt_input_shape.update(opt_pact_shape)
L
LDOUBLEV 已提交
245
        elif mode == "rec":
L
LDOUBLEV 已提交
246
            min_input_shape = {"x": [1, 3, 32, 10]}
L
LDOUBLEV 已提交
247 248 249
            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":
L
LDOUBLEV 已提交
250
            min_input_shape = {"x": [1, 3, 48, 10]}
L
LDOUBLEV 已提交
251 252
            max_input_shape = {"x": [args.rec_batch_num, 3, 48, 2000]}
            opt_input_shape = {"x": [args.rec_batch_num, 3, 48, 320]}
L
LDOUBLEV 已提交
253 254 255 256
        else:
            min_input_shape = {"x": [1, 3, 10, 10]}
            max_input_shape = {"x": [1, 3, 1000, 1000]}
            opt_input_shape = {"x": [1, 3, 500, 500]}
L
LDOUBLEV 已提交
257 258 259
        config.set_trt_dynamic_shape_info(min_input_shape, max_input_shape,
                                          opt_input_shape)

W
WenmuZhou 已提交
260 261
    else:
        config.disable_gpu()
L
LDOUBLEV 已提交
262 263 264
        if hasattr(args, "cpu_threads"):
            config.set_cpu_math_library_num_threads(args.cpu_threads)
        else:
W
WenmuZhou 已提交
265
            # default cpu threads as 10
L
LDOUBLEV 已提交
266
            config.set_cpu_math_library_num_threads(10)
W
WenmuZhou 已提交
267 268 269 270
        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
271 272
            if args.precision == "fp16":
                config.enable_mkldnn_bfloat16()
L
LDOUBLEV 已提交
273 274
    # enable memory optim
    config.enable_memory_optim()
L
LDOUBLEV 已提交
275
    config.disable_glog_info()
W
WenmuZhou 已提交
276

W
WenmuZhou 已提交
277
    config.delete_pass("conv_transpose_eltwiseadd_bn_fuse_pass")
W
WenmuZhou 已提交
278
    if mode == 'table':
W
WenmuZhou 已提交
279
        config.delete_pass("fc_fuse_pass")  # not supported for table
W
WenmuZhou 已提交
280
    config.switch_use_feed_fetch_ops(False)
W
WenmuZhou 已提交
281
    config.switch_ir_optim(True)
282

W
WenmuZhou 已提交
283 284
    # create predictor
    predictor = inference.create_predictor(config)
W
WenmuZhou 已提交
285 286
    input_names = predictor.get_input_names()
    for name in input_names:
W
WenmuZhou 已提交
287
        input_tensor = predictor.get_input_handle(name)
W
WenmuZhou 已提交
288 289 290
    output_names = predictor.get_output_names()
    output_tensors = []
    for output_name in output_names:
W
WenmuZhou 已提交
291
        output_tensor = predictor.get_output_handle(output_name)
W
WenmuZhou 已提交
292
        output_tensors.append(output_tensor)
L
LDOUBLEV 已提交
293
    return predictor, input_tensor, output_tensors, config
W
WenmuZhou 已提交
294 295


L
LDOUBLEV 已提交
296 297
def get_infer_gpuid():
    cmd = "nvidia-smi"
L
LDOUBLEV 已提交
298 299 300 301
    try:
        res = os.popen(cmd).readlines()
    except:
        res = None
L
LDOUBLEV 已提交
302 303 304 305 306 307 308 309 310 311 312
    if len(res) == 0:
        return None
    cmd = "env | grep CUDA_VISIBLE_DEVICES"
    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
313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328
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
329
def draw_text_det_res(dt_boxes, img_path):
L
LDOUBLEV 已提交
330 331 332 333
    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
334
    return src_im
L
LDOUBLEV 已提交
335 336


L
LDOUBLEV 已提交
337 338
def resize_img(img, input_size=600):
    """
L
LDOUBLEV 已提交
339
    resize img and limit the longest side of the image to input_size
L
LDOUBLEV 已提交
340 341 342 343 344
    """
    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
345 346
    img = cv2.resize(img, None, None, fx=im_scale, fy=im_scale)
    return img
L
LDOUBLEV 已提交
347 348


W
WenmuZhou 已提交
349 350 351 352 353
def draw_ocr(image,
             boxes,
             txts=None,
             scores=None,
             drop_score=0.5,
L
LDOUBLEV 已提交
354
             font_path="./doc/fonts/simfang.ttf"):
355 356 357
    """
    Visualize the results of OCR detection and recognition
    args:
L
LDOUBLEV 已提交
358
        image(Image|array): RGB image
359 360 361 362
        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
363
        font_path: the path of font which is used to draw text
364 365 366
    return(array):
        the visualized img
    """
L
LDOUBLEV 已提交
367 368
    if scores is None:
        scores = [1] * len(boxes)
W
WenmuZhou 已提交
369 370 371 372
    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
373
            continue
W
WenmuZhou 已提交
374
        box = np.reshape(np.array(boxes[i]), [-1, 1, 2]).astype(np.int64)
L
LDOUBLEV 已提交
375
        image = cv2.polylines(np.array(image), [box], True, (255, 0, 0), 2)
W
WenmuZhou 已提交
376
    if txts is not None:
L
LDOUBLEV 已提交
377
        img = np.array(resize_img(image, input_size=600))
378
        txt_img = text_visual(
W
WenmuZhou 已提交
379 380 381 382 383 384
            txts,
            scores,
            img_h=img.shape[0],
            img_w=600,
            threshold=drop_score,
            font_path=font_path)
385
        img = np.concatenate([np.array(img), np.array(txt_img)], axis=1)
L
LDOUBLEV 已提交
386 387
        return img
    return image
388 389


W
WenmuZhou 已提交
390 391 392 393 394 395
def draw_ocr_box_txt(image,
                     boxes,
                     txts,
                     scores=None,
                     drop_score=0.5,
                     font_path="./doc/simfang.ttf"):
396 397 398
    h, w = image.height, image.width
    img_left = image.copy()
    img_right = Image.new('RGB', (w, h), (255, 255, 255))
399 400

    import random
L
LDOUBLEV 已提交
401

402 403 404
    random.seed(0)
    draw_left = ImageDraw.Draw(img_left)
    draw_right = ImageDraw.Draw(img_right)
W
WenmuZhou 已提交
405 406 407
    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
408 409
        color = (random.randint(0, 255), random.randint(0, 255),
                 random.randint(0, 255))
410
        draw_left.polygon(box, fill=color)
T
tink2123 已提交
411 412 413 414 415 416 417 418 419 420
        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)
421 422
        if box_height > 2 * box_width:
            font_size = max(int(box_width * 0.9), 10)
W
WenmuZhou 已提交
423
            font = ImageFont.truetype(font_path, font_size, encoding="utf-8")
424 425 426
            cur_y = box[0][1]
            for c in txt:
                char_size = font.getsize(c)
T
tink2123 已提交
427 428
                draw_right.text(
                    (box[0][0] + 3, cur_y), c, fill=(0, 0, 0), font=font)
429 430 431
                cur_y += char_size[1]
        else:
            font_size = max(int(box_height * 0.8), 10)
W
WenmuZhou 已提交
432
            font = ImageFont.truetype(font_path, font_size, encoding="utf-8")
T
tink2123 已提交
433 434
            draw_right.text(
                [box[0][0], box[0][1]], txt, fill=(0, 0, 0), font=font)
435 436 437 438
    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))
439 440 441
    return np.array(img_show)


442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465
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
466 467 468 469 470 471
def text_visual(texts,
                scores,
                img_h=400,
                img_w=600,
                threshold=0.,
                font_path="./doc/simfang.ttf"):
472 473 474 475 476 477 478
    """
    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
479
        font_path: the path of font which is used to draw text
480 481 482 483 484 485 486 487 488
    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
489 490
        blank_img = Image.fromarray(blank_img).convert("RGB")
        draw_txt = ImageDraw.Draw(blank_img)
491
        return blank_img, draw_txt
L
LDOUBLEV 已提交
492

493 494 495 496
    blank_img, draw_txt = create_blank_img()

    font_size = 20
    txt_color = (0, 0, 0)
W
WenmuZhou 已提交
497
    font = ImageFont.truetype(font_path, font_size, encoding="utf-8")
498 499 500

    gap = font_size + 5
    txt_img_list = []
L
LDOUBLEV 已提交
501
    count, index = 1, 0
502 503
    for idx, txt in enumerate(texts):
        index += 1
L
LDOUBLEV 已提交
504
        if scores[idx] < threshold or math.isnan(scores[idx]):
505 506 507 508 509 510 511 512 513 514 515
            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
516
            draw_txt.text((0, gap * count), new_txt, txt_color, font=font)
517 518 519 520 521
            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
522
            count += 1
523 524 525
        if first_line:
            new_txt = str(index) + ': ' + txt + '   ' + '%.3f' % (scores[idx])
        else:
L
LDOUBLEV 已提交
526
            new_txt = "  " + txt + "  " + '%.3f' % (scores[idx])
L
LDOUBLEV 已提交
527
        draw_txt.text((0, gap * count), new_txt, txt_color, font=font)
528
        # whether add new blank img or not
L
LDOUBLEV 已提交
529
        if count >= img_h // gap - 1 and idx + 1 < len(texts):
530 531 532
            txt_img_list.append(np.array(blank_img))
            blank_img, draw_txt = create_blank_img()
            count = 0
L
LDOUBLEV 已提交
533
        count += 1
534 535 536 537 538 539
    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
540 541


D
dyning 已提交
542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560
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
561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595
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


L
LDOUBLEV 已提交
596
if __name__ == '__main__':
L
LDOUBLEV 已提交
597
    pass