paddleocr.py 14.0 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 21 22 23 24 25 26 27 28
# 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 os
import sys

__dir__ = os.path.dirname(__file__)
sys.path.append(os.path.join(__dir__, ''))

import cv2
import numpy as np
from pathlib import Path
import tarfile
import requests
from tqdm import tqdm

from tools.infer import predict_system
W
WenmuZhou 已提交
29
from ppocr.utils.logging import get_logger
W
WenmuZhou 已提交
30

W
WenmuZhou 已提交
31
logger = get_logger()
32
from ppocr.utils.utility import check_and_read_gif, get_image_file_list
W
WenmuZhou 已提交
33 34 35

__all__ = ['PaddleOCR']

W
WenmuZhou 已提交
36 37
model_urls = {
    'det':
W
WenmuZhou 已提交
38
    'https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar',
W
WenmuZhou 已提交
39 40 41
    'rec': {
        'ch': {
            'url':
W
WenmuZhou 已提交
42
            'https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_infer.tar',
W
WenmuZhou 已提交
43 44 45 46
            'dict_path': './ppocr/utils/ppocr_keys_v1.txt'
        },
        'en': {
            'url':
W
WenmuZhou 已提交
47 48
            'https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/en_number_mobile_v2.0_rec_infer.tar',
            'dict_path': './ppocr/utils/dict/en_dict.txt'
W
WenmuZhou 已提交
49 50 51
        },
        'french': {
            'url':
W
WenmuZhou 已提交
52
            'https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/french_mobile_v2.0_rec_infer.tar',
W
WenmuZhou 已提交
53 54 55 56
            'dict_path': './ppocr/utils/dict/french_dict.txt'
        },
        'german': {
            'url':
W
WenmuZhou 已提交
57
            'https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/german_mobile_v2.0_rec_infer.tar',
W
WenmuZhou 已提交
58 59 60 61
            'dict_path': './ppocr/utils/dict/german_dict.txt'
        },
        'korean': {
            'url':
W
WenmuZhou 已提交
62
            'https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/korean_mobile_v2.0_rec_infer.tar',
W
WenmuZhou 已提交
63 64 65 66
            'dict_path': './ppocr/utils/dict/korean_dict.txt'
        },
        'japan': {
            'url':
W
WenmuZhou 已提交
67
            'https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/japan_mobile_v2.0_rec_infer.tar',
W
WenmuZhou 已提交
68 69 70 71
            'dict_path': './ppocr/utils/dict/japan_dict.txt'
        }
    },
    'cls':
W
WenmuZhou 已提交
72
    'https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar'
W
WenmuZhou 已提交
73 74 75
}

SUPPORT_DET_MODEL = ['DB']
W
WenmuZhou 已提交
76
VERSION = 2.0
77 78
SUPPORT_REC_MODEL = ['CRNN']
BASE_DIR = os.path.expanduser("~/.paddleocr/")
W
WenmuZhou 已提交
79 80 81 82 83 84 85 86 87 88 89 90


def download_with_progressbar(url, save_path):
    response = requests.get(url, stream=True)
    total_size_in_bytes = int(response.headers.get('content-length', 0))
    block_size = 1024  # 1 Kibibyte
    progress_bar = tqdm(total=total_size_in_bytes, unit='iB', unit_scale=True)
    with open(save_path, 'wb') as file:
        for data in response.iter_content(block_size):
            progress_bar.update(len(data))
            file.write(data)
    progress_bar.close()
W
WenmuZhou 已提交
91 92
    if total_size_in_bytes == 0 or progress_bar.n != total_size_in_bytes:
        logger.error("Something went wrong while downloading models")
W
WenmuZhou 已提交
93 94 95
        sys.exit(0)


96
def maybe_download(model_storage_directory, url):
W
WenmuZhou 已提交
97
    # using custom model
W
WenmuZhou 已提交
98 99 100 101 102 103 104
    tar_file_name_list = [
        'inference.pdiparams', 'inference.pdiparams.info', 'inference.pdmodel'
    ]
    if not os.path.exists(
            os.path.join(model_storage_directory, 'inference.pdiparams')
    ) or not os.path.exists(
            os.path.join(model_storage_directory, 'inference.pdmodel')):
105 106 107 108 109 110
        tmp_path = os.path.join(model_storage_directory, url.split('/')[-1])
        print('download {} to {}'.format(url, tmp_path))
        os.makedirs(model_storage_directory, exist_ok=True)
        download_with_progressbar(url, tmp_path)
        with tarfile.open(tmp_path, 'r') as tarObj:
            for member in tarObj.getmembers():
W
WenmuZhou 已提交
111 112 113 114 115
                filename = None
                for tar_file_name in tar_file_name_list:
                    if tar_file_name in member.name:
                        filename = tar_file_name
                if filename is None:
116 117 118 119 120 121 122
                    continue
                file = tarObj.extractfile(member)
                with open(
                        os.path.join(model_storage_directory, filename),
                        'wb') as f:
                    f.write(file.read())
        os.remove(tmp_path)
W
WenmuZhou 已提交
123 124


W
WenmuZhou 已提交
125
def parse_args(mMain=True, add_help=True):
W
WenmuZhou 已提交
126 127 128 129 130
    import argparse

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

W
WenmuZhou 已提交
131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148
    if mMain:
        parser = argparse.ArgumentParser(add_help=add_help)
        # params for prediction engine
        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)
        parser.add_argument("--gpu_mem", type=int, default=8000)

        # params for text detector
        parser.add_argument("--image_dir", type=str)
        parser.add_argument("--det_algorithm", type=str, default='DB')
        parser.add_argument("--det_model_dir", type=str, default=None)
        parser.add_argument("--det_limit_side_len", type=float, default=960)
        parser.add_argument("--det_limit_type", type=str, default='max')

        # DB parmas
        parser.add_argument("--det_db_thresh", type=float, default=0.3)
        parser.add_argument("--det_db_box_thresh", type=float, default=0.5)
L
LDOUBLEV 已提交
149 150
        parser.add_argument("--det_db_unclip_ratio", type=float, default=1.6)
        parser.add_argument("--use_dilation", type=bool, default=False)
W
WenmuZhou 已提交
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

        # EAST parmas
        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)

        # params for text recognizer
        parser.add_argument("--rec_algorithm", type=str, default='CRNN')
        parser.add_argument("--rec_model_dir", type=str, default=None)
        parser.add_argument("--rec_image_shape", type=str, default="3, 32, 320")
        parser.add_argument("--rec_char_type", type=str, default='ch')
        parser.add_argument("--rec_batch_num", type=int, default=30)
        parser.add_argument("--max_text_length", type=int, default=25)
        parser.add_argument("--rec_char_dict_path", type=str, default=None)
        parser.add_argument("--use_space_char", type=bool, default=True)
        parser.add_argument("--drop_score", type=float, default=0.5)

        # params for text classifier
        parser.add_argument("--cls_model_dir", type=str, default=None)
        parser.add_argument("--cls_image_shape", type=str, default="3, 48, 192")
        parser.add_argument("--label_list", type=list, default=['0', '180'])
        parser.add_argument("--cls_batch_num", type=int, default=30)
        parser.add_argument("--cls_thresh", type=float, default=0.9)

        parser.add_argument("--enable_mkldnn", type=bool, default=False)
        parser.add_argument("--use_zero_copy_run", type=bool, default=False)
        parser.add_argument("--use_pdserving", type=str2bool, default=False)

        parser.add_argument("--lang", type=str, default='ch')
        parser.add_argument("--det", type=str2bool, default=True)
        parser.add_argument("--rec", type=str2bool, default=True)
        parser.add_argument("--use_angle_cls", type=str2bool, default=False)
        return parser.parse_args()
    else:
W
WenmuZhou 已提交
185 186 187 188 189 190 191 192 193 194 195 196
        return argparse.Namespace(
            use_gpu=True,
            ir_optim=True,
            use_tensorrt=False,
            gpu_mem=8000,
            image_dir='',
            det_algorithm='DB',
            det_model_dir=None,
            det_limit_side_len=960,
            det_limit_type='max',
            det_db_thresh=0.3,
            det_db_box_thresh=0.5,
L
LDOUBLEV 已提交
197 198
            det_db_unclip_ratio=1.6,
            use_dilation=False,
W
WenmuZhou 已提交
199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222
            det_east_score_thresh=0.8,
            det_east_cover_thresh=0.1,
            det_east_nms_thresh=0.2,
            rec_algorithm='CRNN',
            rec_model_dir=None,
            rec_image_shape="3, 32, 320",
            rec_char_type='ch',
            rec_batch_num=30,
            max_text_length=25,
            rec_char_dict_path=None,
            use_space_char=True,
            drop_score=0.5,
            cls_model_dir=None,
            cls_image_shape="3, 48, 192",
            label_list=['0', '180'],
            cls_batch_num=30,
            cls_thresh=0.9,
            enable_mkldnn=False,
            use_zero_copy_run=False,
            use_pdserving=False,
            lang='ch',
            det=True,
            rec=True,
            use_angle_cls=False)
W
WenmuZhou 已提交
223 224 225


class PaddleOCR(predict_system.TextSystem):
226
    def __init__(self, **kwargs):
W
WenmuZhou 已提交
227 228 229 230 231
        """
        paddleocr package
        args:
            **kwargs: other params show in paddleocr --help
        """
W
WenmuZhou 已提交
232
        postprocess_params = parse_args(mMain=False, add_help=False)
233
        postprocess_params.__dict__.update(**kwargs)
W
WenmuZhou 已提交
234 235 236 237
        self.use_angle_cls = postprocess_params.use_angle_cls
        lang = postprocess_params.lang
        assert lang in model_urls[
            'rec'], 'param lang must in {}, but got {}'.format(
W
WenmuZhou 已提交
238
                model_urls['rec'].keys(), lang)
W
WenmuZhou 已提交
239
        use_inner_dict = False
W
WenmuZhou 已提交
240
        if postprocess_params.rec_char_dict_path is None:
W
WenmuZhou 已提交
241
            use_inner_dict = True
W
WenmuZhou 已提交
242 243
            postprocess_params.rec_char_dict_path = model_urls['rec'][lang][
                'dict_path']
W
WenmuZhou 已提交
244

245 246
        # init model dir
        if postprocess_params.det_model_dir is None:
W
WenmuZhou 已提交
247 248
            postprocess_params.det_model_dir = os.path.join(
                BASE_DIR, '{}/det'.format(VERSION))
249
        if postprocess_params.rec_model_dir is None:
W
WenmuZhou 已提交
250
            postprocess_params.rec_model_dir = os.path.join(
W
WenmuZhou 已提交
251
                BASE_DIR, '{}/rec/{}'.format(VERSION, lang))
W
WenmuZhou 已提交
252
        if postprocess_params.cls_model_dir is None:
W
WenmuZhou 已提交
253 254
            postprocess_params.cls_model_dir = os.path.join(
                BASE_DIR, '{}/cls'.format(VERSION))
255
        print(postprocess_params)
W
WenmuZhou 已提交
256
        # download model
W
WenmuZhou 已提交
257 258 259 260
        maybe_download(postprocess_params.det_model_dir, model_urls['det'])
        maybe_download(postprocess_params.rec_model_dir,
                       model_urls['rec'][lang]['url'])
        maybe_download(postprocess_params.cls_model_dir, model_urls['cls'])
W
WenmuZhou 已提交
261 262 263 264 265 266 267

        if postprocess_params.det_algorithm not in SUPPORT_DET_MODEL:
            logger.error('det_algorithm must in {}'.format(SUPPORT_DET_MODEL))
            sys.exit(0)
        if postprocess_params.rec_algorithm not in SUPPORT_REC_MODEL:
            logger.error('rec_algorithm must in {}'.format(SUPPORT_REC_MODEL))
            sys.exit(0)
W
WenmuZhou 已提交
268 269 270
        if use_inner_dict:
            postprocess_params.rec_char_dict_path = str(
                Path(__file__).parent / postprocess_params.rec_char_dict_path)
W
WenmuZhou 已提交
271 272 273 274

        # init det_model and rec_model
        super().__init__(postprocess_params)

W
WenmuZhou 已提交
275
    def ocr(self, img, det=True, rec=True, cls=False):
W
WenmuZhou 已提交
276 277 278 279 280 281 282 283
        """
        ocr with paddleocr
        args:
            img: img for ocr, support ndarray, img_path and list or ndarray
            det: use text detection or not, if false, only rec will be exec. default is True
            rec: use text recognition or not, if false, only det will be exec. default is True
        """
        assert isinstance(img, (np.ndarray, list, str))
W
WenmuZhou 已提交
284 285 286
        if isinstance(img, list) and det == True:
            logger.error('When input a list of images, det must be false')
            exit(0)
W
WenmuZhou 已提交
287 288 289 290 291 292
        if cls == False:
            self.use_angle_cls = False
        elif cls == True and self.use_angle_cls == False:
            logger.warning(
                'Since the angle classifier is not initialized, the angle classifier will not be uesd during the forward process'
            )
W
WenmuZhou 已提交
293

W
WenmuZhou 已提交
294
        if isinstance(img, str):
W
WenmuZhou 已提交
295 296 297 298
            # download net image
            if img.startswith('http'):
                download_with_progressbar(img, 'tmp.jpg')
                img = 'tmp.jpg'
W
WenmuZhou 已提交
299 300 301
            image_file = img
            img, flag = check_and_read_gif(image_file)
            if not flag:
302 303 304
                with open(image_file, 'rb') as f:
                    np_arr = np.frombuffer(f.read(), dtype=np.uint8)
                    img = cv2.imdecode(np_arr, cv2.IMREAD_COLOR)
W
WenmuZhou 已提交
305 306 307
            if img is None:
                logger.error("error in loading image:{}".format(image_file))
                return None
W
WenmuZhou 已提交
308 309
        if isinstance(img, np.ndarray) and len(img.shape) == 2:
            img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
W
WenmuZhou 已提交
310 311 312 313 314 315 316 317 318 319 320
        if det and rec:
            dt_boxes, rec_res = self.__call__(img)
            return [[box.tolist(), res] for box, res in zip(dt_boxes, rec_res)]
        elif det and not rec:
            dt_boxes, elapse = self.text_detector(img)
            if dt_boxes is None:
                return None
            return [box.tolist() for box in dt_boxes]
        else:
            if not isinstance(img, list):
                img = [img]
W
WenmuZhou 已提交
321 322 323 324
            if self.use_angle_cls:
                img, cls_res, elapse = self.text_classifier(img)
                if not rec:
                    return cls_res
W
WenmuZhou 已提交
325 326
            rec_res, elapse = self.text_recognizer(img)
            return rec_res
327 328 329


def main():
W
WenmuZhou 已提交
330 331 332 333 334 335 336 337
    # for cmd
    args = parse_args(mMain=True)
    image_dir = args.image_dir
    if image_dir.startswith('http'):
        download_with_progressbar(image_dir, 'tmp.jpg')
        image_file_list = ['tmp.jpg']
    else:
        image_file_list = get_image_file_list(args.image_dir)
338 339 340
    if len(image_file_list) == 0:
        logger.error('no images find in {}'.format(args.image_dir))
        return
W
WenmuZhou 已提交
341 342

    ocr_engine = PaddleOCR(**(args.__dict__))
343
    for img_path in image_file_list:
W
WenmuZhou 已提交
344 345 346 347 348 349 350 351
        logger.info('{}{}{}'.format('*' * 10, img_path, '*' * 10))
        result = ocr_engine.ocr(img_path,
                                det=args.det,
                                rec=args.rec,
                                cls=args.use_angle_cls)
        if result is not None:
            for line in result:
                logger.info(line)