Spaces:
Running
Running
Commit
·
255fbb0
1
Parent(s):
21d5ed5
Update to v2.3.2. Refactor detection model handling in app.py; update requirements for cnocr and numpy
Browse files- app.py +16 -17
- requirements.txt +3 -272
app.py
CHANGED
@@ -34,6 +34,12 @@ from cnocr.utils import set_logger, draw_ocr_results, download
|
|
34 |
logger = set_logger()
|
35 |
MODELS = {}
|
36 |
|
|
|
|
|
|
|
|
|
|
|
|
|
37 |
|
38 |
def plot_for_debugging(rotated_img, one_out, box_score_thresh, crop_ncols, prefix_fp):
|
39 |
import matplotlib.pyplot as plt
|
@@ -73,8 +79,7 @@ def get_ocr_model(det_model_name, rec_model_name, det_more_configs):
|
|
73 |
if (det_model_name, rec_model_name, config_str) in MODELS:
|
74 |
return MODELS[(det_model_name, rec_model_name, config_str)]
|
75 |
|
76 |
-
|
77 |
-
# rec_model_name, rec_model_backend = rec_model_name.split('::')
|
78 |
rec_model_backend = 'onnx'
|
79 |
model = CnOcr(
|
80 |
det_model_name=det_model_name,
|
@@ -157,7 +162,6 @@ def recognize(
|
|
157 |
min_box_size=min_box_size,
|
158 |
)
|
159 |
|
160 |
-
det_model_name, det_model_backend = det_model_name.split('::')
|
161 |
if is_single_line or det_model_name == 'naive_det':
|
162 |
out_texts = visualize_naive_result(
|
163 |
img, det_model_name, ocr_out, box_score_thresh
|
@@ -183,11 +187,6 @@ def recognize(
|
|
183 |
|
184 |
|
185 |
def main():
|
186 |
-
det_models = list(DET_AVAILABLE_MODELS.all_models())
|
187 |
-
det_models.append(('naive_det', 'onnx'))
|
188 |
-
det_models.sort()
|
189 |
-
det_models = [f'{m}::{b}' for m, b in det_models]
|
190 |
-
|
191 |
all_models = list(REC_AVAILABLE_MODELS.all_models())
|
192 |
all_models = [f'{m}' for m, b in all_models if b == 'onnx']
|
193 |
cnocr_models = [name for name in all_models if 'densenet' in name]
|
@@ -206,7 +205,7 @@ def main():
|
|
206 |
)
|
207 |
example_func = functools.partial(
|
208 |
recognize,
|
209 |
-
# det_model_name='ch_PP-OCRv3_det
|
210 |
rotated_bbox=True,
|
211 |
# use_angle_clf=False,
|
212 |
new_size=768,
|
@@ -215,49 +214,49 @@ def main():
|
|
215 |
)
|
216 |
examples = [
|
217 |
[
|
218 |
-
'ch_PP-OCRv3_det
|
219 |
True,
|
220 |
'number-densenet_lite_136-fc',
|
221 |
False,
|
222 |
'docs/examples/card1-s.jpg',
|
223 |
],
|
224 |
[
|
225 |
-
'ch_PP-OCRv3_det
|
226 |
True,
|
227 |
'number-densenet_lite_136-fc',
|
228 |
False,
|
229 |
'docs/examples/card2-s.jpg',
|
230 |
],
|
231 |
[
|
232 |
-
'ch_PP-OCRv3_det
|
233 |
True,
|
234 |
'number-densenet_lite_136-fc',
|
235 |
False,
|
236 |
'docs/examples/cy1-s.jpg',
|
237 |
],
|
238 |
[
|
239 |
-
'ch_PP-OCRv3_det
|
240 |
False,
|
241 |
'densenet_lite_136-gru',
|
242 |
False,
|
243 |
'docs/examples/huochepiao.jpeg',
|
244 |
],
|
245 |
[
|
246 |
-
'ch_PP-OCRv3_det
|
247 |
False,
|
248 |
'densenet_lite_136-gru',
|
249 |
False,
|
250 |
'docs/examples/1_res.jpg',
|
251 |
],
|
252 |
[
|
253 |
-
'db_shufflenet_v2
|
254 |
False,
|
255 |
'en_number_mobile_v2.0',
|
256 |
False,
|
257 |
'docs/examples/en_book1.jpeg',
|
258 |
],
|
259 |
[
|
260 |
-
'db_shufflenet_v2
|
261 |
False,
|
262 |
'densenet_lite_136-gru',
|
263 |
True,
|
@@ -291,7 +290,7 @@ def main():
|
|
291 |
with gr.Column(min_width=200, variant='panel', scale=3):
|
292 |
gr.Markdown('### 模型设置')
|
293 |
det_model_name = gr.Dropdown(
|
294 |
-
label='选择检测模型', choices=
|
295 |
)
|
296 |
is_single_line = gr.Checkbox(label='单行文字模式(不使用检测模型)', value=False)
|
297 |
|
|
|
34 |
logger = set_logger()
|
35 |
MODELS = {}
|
36 |
|
37 |
+
DET_MODELS = list(DET_AVAILABLE_MODELS.all_models())
|
38 |
+
DET_MODELS.append(('naive_det', 'onnx'))
|
39 |
+
DET_MODELS.sort()
|
40 |
+
DET_MODELS_NAME = [f'{m}' for m, _ in DET_MODELS]
|
41 |
+
DET_MODELS_DICT = {f'{m}': f'{b}' for m, b in DET_MODELS}
|
42 |
+
|
43 |
|
44 |
def plot_for_debugging(rotated_img, one_out, box_score_thresh, crop_ncols, prefix_fp):
|
45 |
import matplotlib.pyplot as plt
|
|
|
79 |
if (det_model_name, rec_model_name, config_str) in MODELS:
|
80 |
return MODELS[(det_model_name, rec_model_name, config_str)]
|
81 |
|
82 |
+
det_model_backend = DET_MODELS_DICT[det_model_name]
|
|
|
83 |
rec_model_backend = 'onnx'
|
84 |
model = CnOcr(
|
85 |
det_model_name=det_model_name,
|
|
|
162 |
min_box_size=min_box_size,
|
163 |
)
|
164 |
|
|
|
165 |
if is_single_line or det_model_name == 'naive_det':
|
166 |
out_texts = visualize_naive_result(
|
167 |
img, det_model_name, ocr_out, box_score_thresh
|
|
|
187 |
|
188 |
|
189 |
def main():
|
|
|
|
|
|
|
|
|
|
|
190 |
all_models = list(REC_AVAILABLE_MODELS.all_models())
|
191 |
all_models = [f'{m}' for m, b in all_models if b == 'onnx']
|
192 |
cnocr_models = [name for name in all_models if 'densenet' in name]
|
|
|
205 |
)
|
206 |
example_func = functools.partial(
|
207 |
recognize,
|
208 |
+
# det_model_name='ch_PP-OCRv3_det',
|
209 |
rotated_bbox=True,
|
210 |
# use_angle_clf=False,
|
211 |
new_size=768,
|
|
|
214 |
)
|
215 |
examples = [
|
216 |
[
|
217 |
+
'ch_PP-OCRv3_det',
|
218 |
True,
|
219 |
'number-densenet_lite_136-fc',
|
220 |
False,
|
221 |
'docs/examples/card1-s.jpg',
|
222 |
],
|
223 |
[
|
224 |
+
'ch_PP-OCRv3_det',
|
225 |
True,
|
226 |
'number-densenet_lite_136-fc',
|
227 |
False,
|
228 |
'docs/examples/card2-s.jpg',
|
229 |
],
|
230 |
[
|
231 |
+
'ch_PP-OCRv3_det',
|
232 |
True,
|
233 |
'number-densenet_lite_136-fc',
|
234 |
False,
|
235 |
'docs/examples/cy1-s.jpg',
|
236 |
],
|
237 |
[
|
238 |
+
'ch_PP-OCRv3_det',
|
239 |
False,
|
240 |
'densenet_lite_136-gru',
|
241 |
False,
|
242 |
'docs/examples/huochepiao.jpeg',
|
243 |
],
|
244 |
[
|
245 |
+
'ch_PP-OCRv3_det',
|
246 |
False,
|
247 |
'densenet_lite_136-gru',
|
248 |
False,
|
249 |
'docs/examples/1_res.jpg',
|
250 |
],
|
251 |
[
|
252 |
+
'db_shufflenet_v2',
|
253 |
False,
|
254 |
'en_number_mobile_v2.0',
|
255 |
False,
|
256 |
'docs/examples/en_book1.jpeg',
|
257 |
],
|
258 |
[
|
259 |
+
'db_shufflenet_v2',
|
260 |
False,
|
261 |
'densenet_lite_136-gru',
|
262 |
True,
|
|
|
290 |
with gr.Column(min_width=200, variant='panel', scale=3):
|
291 |
gr.Markdown('### 模型设置')
|
292 |
det_model_name = gr.Dropdown(
|
293 |
+
label='选择检测模型', choices=DET_MODELS_NAME, value='ch_PP-OCRv3_det',
|
294 |
)
|
295 |
is_single_line = gr.Checkbox(label='单行文字模式(不使用检测模型)', value=False)
|
296 |
|
requirements.txt
CHANGED
@@ -6,278 +6,9 @@
|
|
6 |
#
|
7 |
--extra-index-url https://pypi.org/simple
|
8 |
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
# via aiohttp
|
13 |
-
albumentations==1.3.1
|
14 |
-
# via -r requirements.in
|
15 |
-
appdirs==1.4.4
|
16 |
-
# via wandb
|
17 |
-
async-timeout==4.0.2
|
18 |
-
# via aiohttp
|
19 |
-
attrs==23.1.0
|
20 |
-
# via aiohttp
|
21 |
-
certifi==2023.5.7
|
22 |
-
# via
|
23 |
-
# requests
|
24 |
-
# sentry-sdk
|
25 |
-
charset-normalizer==3.1.0
|
26 |
-
# via
|
27 |
-
# aiohttp
|
28 |
-
# requests
|
29 |
-
click==8.1.3
|
30 |
-
# via
|
31 |
-
# -r requirements.in
|
32 |
-
# cnstd
|
33 |
-
# wandb
|
34 |
-
cnstd>=1.2.3.4
|
35 |
-
# via -r requirements.in
|
36 |
-
coloredlogs==15.0.1
|
37 |
-
# via onnxruntime
|
38 |
-
contourpy==1.1.0
|
39 |
-
# via matplotlib
|
40 |
-
cycler==0.11.0
|
41 |
-
# via matplotlib
|
42 |
-
docker-pycreds==0.4.0
|
43 |
-
# via wandb
|
44 |
-
filelock==3.12.2
|
45 |
-
# via
|
46 |
-
# huggingface-hub
|
47 |
-
# torch
|
48 |
-
flatbuffers==23.5.26
|
49 |
-
# via onnxruntime
|
50 |
-
fonttools==4.40.0
|
51 |
-
# via matplotlib
|
52 |
-
frozenlist==1.3.3
|
53 |
-
# via
|
54 |
-
# aiohttp
|
55 |
-
# aiosignal
|
56 |
-
fsspec[http]==2023.6.0
|
57 |
-
# via
|
58 |
-
# huggingface-hub
|
59 |
-
# pytorch-lightning
|
60 |
-
gitdb==4.0.10
|
61 |
-
# via gitpython
|
62 |
-
gitpython==3.1.34
|
63 |
-
# via wandb
|
64 |
-
huggingface-hub==0.15.1
|
65 |
-
# via cnstd
|
66 |
-
humanfriendly==10.0
|
67 |
-
# via coloredlogs
|
68 |
-
idna==3.4
|
69 |
-
# via
|
70 |
-
# requests
|
71 |
-
# yarl
|
72 |
-
imageio==2.31.3
|
73 |
-
# via scikit-image
|
74 |
-
importlib-resources==5.12.0
|
75 |
-
# via matplotlib
|
76 |
-
jinja2==3.1.2
|
77 |
-
# via torch
|
78 |
-
joblib==1.3.2
|
79 |
-
# via scikit-learn
|
80 |
-
kiwisolver==1.4.4
|
81 |
-
# via matplotlib
|
82 |
-
lazy-loader==0.3
|
83 |
-
# via scikit-image
|
84 |
-
lightning-utilities==0.9.0
|
85 |
-
# via pytorch-lightning
|
86 |
-
markupsafe==2.1.3
|
87 |
-
# via jinja2
|
88 |
-
matplotlib==3.7.1
|
89 |
-
# via
|
90 |
-
# cnstd
|
91 |
-
# seaborn
|
92 |
-
mpmath==1.3.0
|
93 |
-
# via sympy
|
94 |
-
multidict==6.0.4
|
95 |
-
# via
|
96 |
-
# aiohttp
|
97 |
-
# yarl
|
98 |
-
networkx==3.1
|
99 |
-
# via
|
100 |
-
# scikit-image
|
101 |
-
# torch
|
102 |
-
numpy==1.25.0
|
103 |
-
# via
|
104 |
-
# -r requirements.in
|
105 |
-
# albumentations
|
106 |
-
# cnstd
|
107 |
-
# contourpy
|
108 |
-
# imageio
|
109 |
-
# matplotlib
|
110 |
-
# onnx
|
111 |
-
# onnxruntime
|
112 |
-
# opencv-python
|
113 |
-
# opencv-python-headless
|
114 |
-
# pandas
|
115 |
-
# pytorch-lightning
|
116 |
-
# pywavelets
|
117 |
-
# qudida
|
118 |
-
# scikit-image
|
119 |
-
# scikit-learn
|
120 |
-
# scipy
|
121 |
-
# seaborn
|
122 |
-
# shapely
|
123 |
-
# tifffile
|
124 |
-
# torchmetrics
|
125 |
-
# torchvision
|
126 |
-
onnx==1.14.0
|
127 |
-
# via
|
128 |
-
# -r requirements.in
|
129 |
-
# cnstd
|
130 |
-
onnxruntime==1.15.1
|
131 |
-
# via
|
132 |
-
# -r requirements.in
|
133 |
-
# cnstd
|
134 |
-
opencv-python==4.7.0.72
|
135 |
-
# via cnstd
|
136 |
-
opencv-python-headless==4.8.0.76
|
137 |
-
# via
|
138 |
-
# albumentations
|
139 |
-
# qudida
|
140 |
-
packaging==23.1
|
141 |
-
# via
|
142 |
-
# huggingface-hub
|
143 |
-
# lightning-utilities
|
144 |
-
# matplotlib
|
145 |
-
# onnxruntime
|
146 |
-
# pytorch-lightning
|
147 |
-
# scikit-image
|
148 |
-
# torchmetrics
|
149 |
-
pandas==2.0.3
|
150 |
-
# via
|
151 |
-
# cnstd
|
152 |
-
# seaborn
|
153 |
-
pathtools==0.1.2
|
154 |
-
# via wandb
|
155 |
-
pillow==9.5.0
|
156 |
-
# via
|
157 |
-
# -r requirements.in
|
158 |
-
# cnstd
|
159 |
-
# imageio
|
160 |
-
# matplotlib
|
161 |
-
# scikit-image
|
162 |
-
# torchvision
|
163 |
-
polygon3==3.0.9.1
|
164 |
-
# via cnstd
|
165 |
-
protobuf==4.23.3
|
166 |
-
# via
|
167 |
-
# onnx
|
168 |
-
# onnxruntime
|
169 |
-
# wandb
|
170 |
-
psutil==5.9.5
|
171 |
-
# via wandb
|
172 |
-
pyclipper==1.3.0.post4
|
173 |
-
# via cnstd
|
174 |
-
pyparsing==3.1.0
|
175 |
-
# via matplotlib
|
176 |
-
python-dateutil==2.8.2
|
177 |
-
# via
|
178 |
-
# matplotlib
|
179 |
-
# pandas
|
180 |
-
pytorch-lightning==2.0.8
|
181 |
-
# via
|
182 |
-
# -r requirements.in
|
183 |
-
# cnstd
|
184 |
-
pytz==2023.3
|
185 |
-
# via pandas
|
186 |
-
pywavelets==1.4.1
|
187 |
-
# via scikit-image
|
188 |
-
pyyaml==6.0
|
189 |
-
# via
|
190 |
-
# albumentations
|
191 |
-
# cnstd
|
192 |
-
# huggingface-hub
|
193 |
-
# pytorch-lightning
|
194 |
-
# wandb
|
195 |
-
qudida==0.0.4
|
196 |
-
# via albumentations
|
197 |
-
requests==2.31.0
|
198 |
-
# via
|
199 |
-
# fsspec
|
200 |
-
# huggingface-hub
|
201 |
-
# torchvision
|
202 |
-
# wandb
|
203 |
-
scikit-image==0.21.0
|
204 |
-
# via albumentations
|
205 |
-
scikit-learn==1.3.0
|
206 |
-
# via qudida
|
207 |
-
scipy==1.11.1
|
208 |
-
# via
|
209 |
-
# albumentations
|
210 |
-
# cnstd
|
211 |
-
# scikit-image
|
212 |
-
# scikit-learn
|
213 |
-
seaborn==0.12.2
|
214 |
-
# via cnstd
|
215 |
-
sentry-sdk==1.30.0
|
216 |
-
# via wandb
|
217 |
-
setproctitle==1.3.2
|
218 |
-
# via wandb
|
219 |
-
shapely==2.0.1
|
220 |
-
# via cnstd
|
221 |
-
six==1.16.0
|
222 |
-
# via
|
223 |
-
# docker-pycreds
|
224 |
-
# python-dateutil
|
225 |
-
smmap==5.0.0
|
226 |
-
# via gitdb
|
227 |
-
sympy==1.12
|
228 |
-
# via
|
229 |
-
# onnxruntime
|
230 |
-
# torch
|
231 |
-
threadpoolctl==3.2.0
|
232 |
-
# via scikit-learn
|
233 |
-
tifffile==2023.8.30
|
234 |
-
# via scikit-image
|
235 |
-
torch==2.0.1
|
236 |
-
# via
|
237 |
-
# -r requirements.in
|
238 |
-
# cnstd
|
239 |
-
# pytorch-lightning
|
240 |
-
# torchmetrics
|
241 |
-
# torchvision
|
242 |
-
torchmetrics==0.11.4
|
243 |
-
# via
|
244 |
-
# -r requirements.in
|
245 |
-
# pytorch-lightning
|
246 |
-
torchvision==0.15.2
|
247 |
-
# via
|
248 |
-
# -r requirements.in
|
249 |
-
# cnstd
|
250 |
-
tqdm==4.65.0
|
251 |
-
# via
|
252 |
-
# -r requirements.in
|
253 |
-
# cnstd
|
254 |
-
# huggingface-hub
|
255 |
-
# pytorch-lightning
|
256 |
-
typing-extensions==4.7.0
|
257 |
-
# via
|
258 |
-
# huggingface-hub
|
259 |
-
# lightning-utilities
|
260 |
-
# onnx
|
261 |
-
# pytorch-lightning
|
262 |
-
# qudida
|
263 |
-
# torch
|
264 |
-
# wandb
|
265 |
-
tzdata==2023.3
|
266 |
-
# via pandas
|
267 |
-
unidecode==1.3.6
|
268 |
-
# via cnstd
|
269 |
-
urllib3==2.0.3
|
270 |
-
# via
|
271 |
-
# requests
|
272 |
-
# sentry-sdk
|
273 |
-
wandb==0.15.10
|
274 |
-
# via -r requirements.in
|
275 |
-
yarl==1.9.2
|
276 |
-
# via aiohttp
|
277 |
-
zipp==3.15.0
|
278 |
-
# via importlib-resources
|
279 |
-
|
280 |
-
cnocr==2.3.0.1
|
281 |
|
282 |
# The following packages are considered to be unsafe in a requirements file:
|
283 |
# setuptools
|
|
|
6 |
#
|
7 |
--extra-index-url https://pypi.org/simple
|
8 |
|
9 |
+
cnocr>=2.3.2.1,<2.4
|
10 |
+
opencv-python
|
11 |
+
numpy
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
|
13 |
# The following packages are considered to be unsafe in a requirements file:
|
14 |
# setuptools
|