infer_rec.py 5.6 KB
Newer Older
L
LDOUBLEV 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
# 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.

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

import numpy as np
W
WenmuZhou 已提交
20

L
LDOUBLEV 已提交
21 22
import os
import sys
littletomatodonkey's avatar
littletomatodonkey 已提交
23
import json
W
WenmuZhou 已提交
24

25
__dir__ = os.path.dirname(os.path.abspath(__file__))
L
LDOUBLEV 已提交
26
sys.path.append(__dir__)
27
sys.path.append(os.path.abspath(os.path.join(__dir__, '..')))
L
LDOUBLEV 已提交
28

L
LDOUBLEV 已提交
29 30
os.environ["FLAGS_allocator_strategy"] = 'auto_growth'

W
WenmuZhou 已提交
31
import paddle
T
tink2123 已提交
32

W
WenmuZhou 已提交
33
from ppocr.data import create_operators, transform
W
WenmuZhou 已提交
34
from ppocr.modeling.architectures import build_model
W
WenmuZhou 已提交
35
from ppocr.postprocess import build_post_process
36
from ppocr.utils.save_load import load_model
W
WenmuZhou 已提交
37
from ppocr.utils.utility import get_image_file_list
W
WenmuZhou 已提交
38
import tools.program as program
L
LDOUBLEV 已提交
39 40 41


def main():
W
WenmuZhou 已提交
42 43 44 45 46 47 48 49
    global_config = config['Global']

    # build post process
    post_process_class = build_post_process(config['PostProcess'],
                                            global_config)

    # build model
    if hasattr(post_process_class, 'character'):
littletomatodonkey's avatar
littletomatodonkey 已提交
50 51 52 53 54 55 56 57
        char_num = len(getattr(post_process_class, 'character'))
        if config['Architecture']["algorithm"] in ["Distillation",
                                                   ]:  # distillation model
            for key in config['Architecture']["Models"]:
                config['Architecture']["Models"][key]["Head"][
                    'out_channels'] = char_num
        else:  # base rec model
            config['Architecture']["Head"]['out_channels'] = char_num
W
WenmuZhou 已提交
58 59 60

    model = build_model(config['Architecture'])

61
    load_model(config, model)
W
WenmuZhou 已提交
62 63 64

    # create data ops
    transforms = []
W
WenmuZhou 已提交
65
    for op in config['Eval']['dataset']['transforms']:
W
WenmuZhou 已提交
66 67 68 69 70
        op_name = list(op)[0]
        if 'Label' in op_name:
            continue
        elif op_name in ['RecResizeImg']:
            op[op_name]['infer_mode'] = True
W
WenmuZhou 已提交
71
        elif op_name == 'KeepKeys':
T
tink2123 已提交
72 73 74 75 76
            if config['Architecture']['algorithm'] == "SRN":
                op[op_name]['keep_keys'] = [
                    'image', 'encoder_word_pos', 'gsrm_word_pos',
                    'gsrm_slf_attn_bias1', 'gsrm_slf_attn_bias2'
                ]
A
andyjpaddle 已提交
77
            elif config['Architecture']['algorithm'] == "SAR":
78
                op[op_name]['keep_keys'] = ['image', 'valid_ratio']
T
tink2123 已提交
79 80
            else:
                op[op_name]['keep_keys'] = ['image']
W
WenmuZhou 已提交
81 82 83 84
        transforms.append(op)
    global_config['infer_mode'] = True
    ops = create_operators(transforms, global_config)

littletomatodonkey's avatar
littletomatodonkey 已提交
85 86 87 88 89
    save_res_path = config['Global'].get('save_res_path',
                                         "./output/rec/predicts_rec.txt")
    if not os.path.exists(os.path.dirname(save_res_path)):
        os.makedirs(os.path.dirname(save_res_path))

W
WenmuZhou 已提交
90
    model.eval()
littletomatodonkey's avatar
littletomatodonkey 已提交
91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110

    with open(save_res_path, "w") as fout:
        for file in get_image_file_list(config['Global']['infer_img']):
            logger.info("infer_img: {}".format(file))
            with open(file, 'rb') as f:
                img = f.read()
                data = {'image': img}
            batch = transform(data, ops)
            if config['Architecture']['algorithm'] == "SRN":
                encoder_word_pos_list = np.expand_dims(batch[1], axis=0)
                gsrm_word_pos_list = np.expand_dims(batch[2], axis=0)
                gsrm_slf_attn_bias1_list = np.expand_dims(batch[3], axis=0)
                gsrm_slf_attn_bias2_list = np.expand_dims(batch[4], axis=0)

                others = [
                    paddle.to_tensor(encoder_word_pos_list),
                    paddle.to_tensor(gsrm_word_pos_list),
                    paddle.to_tensor(gsrm_slf_attn_bias1_list),
                    paddle.to_tensor(gsrm_slf_attn_bias2_list)
                ]
A
andyjpaddle 已提交
111 112 113
            if config['Architecture']['algorithm'] == "SAR":
                valid_ratio = np.expand_dims(batch[-1], axis=0)
                img_metas = [paddle.to_tensor(valid_ratio)]
littletomatodonkey's avatar
littletomatodonkey 已提交
114 115 116 117 118

            images = np.expand_dims(batch[0], axis=0)
            images = paddle.to_tensor(images)
            if config['Architecture']['algorithm'] == "SRN":
                preds = model(images, others)
A
andyjpaddle 已提交
119 120
            elif config['Architecture']['algorithm'] == "SAR":
                preds = model(images, img_metas)
littletomatodonkey's avatar
littletomatodonkey 已提交
121 122 123
            else:
                preds = model(images)
            post_result = post_process_class(preds)
littletomatodonkey's avatar
littletomatodonkey 已提交
124 125 126 127 128 129 130
            info = None
            if isinstance(post_result, dict):
                rec_info = dict()
                for key in post_result:
                    if len(post_result[key][0]) >= 2:
                        rec_info[key] = {
                            "label": post_result[key][0][0],
131
                            "score": float(post_result[key][0][1]),
littletomatodonkey's avatar
littletomatodonkey 已提交
132 133 134 135 136 137 138 139
                        }
                info = json.dumps(rec_info)
            else:
                if len(post_result[0]) >= 2:
                    info = post_result[0][0] + "\t" + str(post_result[0][1])

            if info is not None:
                logger.info("\t result: {}".format(info))
livingbody's avatar
livingbody 已提交
140
                fout.write(os.path.basename(file) + "\t" + info + "\n")
W
WenmuZhou 已提交
141 142
    logger.info("success!")

L
LDOUBLEV 已提交
143 144

if __name__ == '__main__':
W
WenmuZhou 已提交
145
    config, device, logger, vdl_writer = program.preprocess()
L
LDOUBLEV 已提交
146
    main()