File size: 6,861 Bytes
0464f00
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
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
# coding: utf-8
# Copyright (C) 2021, [Breezedeus](https://github.com/breezedeus).
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements.  See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership.  The ASF licenses this file
# to you 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
from collections import OrderedDict

import cv2
import numpy as np
from PIL import Image
import streamlit as st
from cnstd.utils import pil_to_numpy, imsave

from cnocr import CnOcr, DET_AVAILABLE_MODELS, REC_AVAILABLE_MODELS
from cnocr.utils import set_logger, draw_ocr_results, download


logger = set_logger()
st.set_page_config(layout="wide")


def plot_for_debugging(rotated_img, one_out, box_score_thresh, crop_ncols, prefix_fp):
    import matplotlib.pyplot as plt
    import math

    rotated_img = rotated_img.copy()
    crops = [info['cropped_img'] for info in one_out]
    print('%d boxes are found' % len(crops))
    ncols = crop_ncols
    nrows = math.ceil(len(crops) / ncols)
    fig, ax = plt.subplots(nrows=nrows, ncols=ncols)
    for i, axi in enumerate(ax.flat):
        if i >= len(crops):
            break
        axi.imshow(crops[i])
    crop_fp = '%s-crops.png' % prefix_fp
    plt.savefig(crop_fp)
    print('cropped results are save to file %s' % crop_fp)

    for info in one_out:
        box, score = info.get('position'), info['score']
        if score < box_score_thresh:  # score < 0.5
            continue
        if box is not None:
            box = box.astype(int).reshape(-1, 2)
            cv2.polylines(rotated_img, [box], True, color=(255, 0, 0), thickness=2)
    result_fp = '%s-result.png' % prefix_fp
    imsave(rotated_img, result_fp, normalized=False)
    print('boxes results are save to file %s' % result_fp)


@st.cache(allow_output_mutation=True)
def get_ocr_model(det_model_name, rec_model_name, det_more_configs):
    det_model_name, det_model_backend = det_model_name
    rec_model_name, rec_model_backend = rec_model_name
    return CnOcr(
        det_model_name=det_model_name,
        det_model_backend=det_model_backend,
        rec_model_name=rec_model_name,
        rec_model_backend=rec_model_backend,
        det_more_configs=det_more_configs,
    )


def visualize_naive_result(img, det_model_name, std_out, box_score_thresh):
    img = pil_to_numpy(img).transpose((1, 2, 0)).astype(np.uint8)

    plot_for_debugging(img, std_out, box_score_thresh, 2, './streamlit-app')
    st.subheader('Detection Result')
    if det_model_name == 'default_det':
        st.warning('⚠️ Warning: "default_det" 检测模型不返回文本框位置!')
    cols = st.columns([1, 7, 1])
    cols[1].image('./streamlit-app-result.png')

    st.subheader('Recognition Result')
    cols = st.columns([1, 7, 1])
    cols[1].image('./streamlit-app-crops.png')

    _visualize_ocr(std_out)


def _visualize_ocr(ocr_outs):
    st.empty()
    ocr_res = OrderedDict({'文本': []})
    ocr_res['得分'] = []
    for out in ocr_outs:
        # cropped_img = out['cropped_img']  # 检测出的文本框
        ocr_res['得分'].append(out['score'])
        ocr_res['文本'].append(out['text'])
    st.table(ocr_res)


def visualize_result(img, ocr_outs):
    out_draw_fp = './streamlit-app-det-result.png'
    font_path = 'docs/fonts/simfang.ttf'
    if not os.path.exists(font_path):
        url = 'https://huggingface.co/datasets/breezedeus/cnocr-wx-qr-code/resolve/main/fonts/simfang.ttf'
        os.makedirs(os.path.dirname(font_path), exist_ok=True)
        download(url, path=font_path, overwrite=True)
    draw_ocr_results(img, ocr_outs, out_draw_fp, font_path)
    st.image(out_draw_fp)


def main():
    st.sidebar.header('模型设置')
    det_models = list(DET_AVAILABLE_MODELS.all_models())
    det_models.append(('naive_det', 'onnx'))
    det_models.sort()
    det_model_name = st.sidebar.selectbox(
        '选择检测模型', det_models, index=det_models.index(('ch_PP-OCRv3_det', 'onnx'))
    )

    all_models = list(REC_AVAILABLE_MODELS.all_models())
    all_models.sort()
    idx = all_models.index(('densenet_lite_136-fc', 'onnx'))
    rec_model_name = st.sidebar.selectbox('选择识别模型', all_models, index=idx)

    st.sidebar.subheader('检测参数')
    rotated_bbox = st.sidebar.checkbox('是否检测带角度文本框', value=True)
    use_angle_clf = st.sidebar.checkbox('是否使用角度预测模型校正文本框', value=False)
    new_size = st.sidebar.slider(
        'resize 后图片(长边)大小', min_value=124, max_value=4096, value=768
    )
    box_score_thresh = st.sidebar.slider(
        '得分阈值(低于阈值的结果会被过滤掉)', min_value=0.05, max_value=0.95, value=0.3
    )
    min_box_size = st.sidebar.slider(
        '框大小阈值(更小的文本框会被过滤掉)', min_value=4, max_value=50, value=10
    )
    # std = get_std_model(det_model_name, rotated_bbox, use_angle_clf)

    # st.sidebar.markdown("""---""")
    # st.sidebar.header('CnOcr 设置')
    det_more_configs = dict(rotated_bbox=rotated_bbox, use_angle_clf=use_angle_clf)
    ocr = get_ocr_model(det_model_name, rec_model_name, det_more_configs)

    st.markdown('# 开源Python OCR工具 ' '[CnOCR](https://github.com/breezedeus/cnocr)')
    st.markdown('> 详细说明参见:[CnOCR 文档](https://cnocr.readthedocs.io/) ;'
                '欢迎加入 [交流群](https://cnocr.readthedocs.io/zh/latest/contact/) ;'
                '作者:[breezedeus](https://github.com/breezedeus) 。')
    st.markdown('')
    st.subheader('选择待检测图片')
    content_file = st.file_uploader('', type=["png", "jpg", "jpeg", "webp"])
    if content_file is None:
        st.stop()

    try:
        img = Image.open(content_file).convert('RGB')

        ocr_out = ocr.ocr(
            img,
            return_cropped_image=True,
            resized_shape=new_size,
            preserve_aspect_ratio=True,
            box_score_thresh=box_score_thresh,
            min_box_size=min_box_size,
        )
        if det_model_name[0] == 'naive_det':
            visualize_naive_result(img, det_model_name[0], ocr_out, box_score_thresh)
        else:
            visualize_result(img, ocr_out)

    except Exception as e:
        st.error(e)


if __name__ == '__main__':
    main()