whl.md 19.9 KB
Newer Older
WenmuZhou's avatar
WenmuZhou committed
1
2
# paddleocr package使用说明

WenmuZhou's avatar
WenmuZhou committed
3
## 1 快速上手
WenmuZhou's avatar
WenmuZhou committed
4

WenmuZhou's avatar
WenmuZhou committed
5
### 1.1 安装whl包
WenmuZhou's avatar
WenmuZhou committed
6
7
8

pip安装
```bash
WenmuZhou's avatar
WenmuZhou committed
9
pip install "paddleocr>=2.0.1" # 推荐使用2.0.1+版本
WenmuZhou's avatar
WenmuZhou committed
10
11
12
13
```

本地构建并安装
```bash
WenmuZhou's avatar
WenmuZhou committed
14
15
python3 setup.py bdist_wheel
pip3 install dist/paddleocr-x.x.x-py3-none-any.whl # x.x.x是paddleocr的版本号
WenmuZhou's avatar
WenmuZhou committed
16
17
```

WenmuZhou's avatar
WenmuZhou committed
18
19
20
21
22
## 2 使用
### 2.1 代码使用
paddleocr whl包会自动下载ppocr轻量级模型作为默认模型,可以根据第3节**自定义模型**进行自定义更换。

* 检测+方向分类器+识别全流程
WenmuZhou's avatar
WenmuZhou committed
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
```python
from paddleocr import PaddleOCR, draw_ocr
# Paddleocr目前支持中英文、英文、法语、德语、韩语、日语,可以通过修改lang参数进行切换
# 参数依次为`ch`, `en`, `french`, `german`, `korean`, `japan`。
ocr = PaddleOCR(use_angle_cls=True, lang="ch") # need to run only once to download and load model into memory
img_path = 'PaddleOCR/doc/imgs/11.jpg'
result = ocr.ocr(img_path, cls=True)
for line in result:
    print(line)

# 显示结果
from PIL import Image
image = Image.open(img_path).convert('RGB')
boxes = [line[0] for line in result]
txts = [line[1][0] for line in result]
scores = [line[1][1] for line in result]
im_show = draw_ocr(image, boxes, txts, scores, font_path='/path/to/PaddleOCR/doc/simfang.ttf')
im_show = Image.fromarray(im_show)
im_show.save('result.jpg')
```
结果是一个list,每个item包含了文本框,文字和识别置信度
```bash
[[[24.0, 36.0], [304.0, 34.0], [304.0, 72.0], [24.0, 74.0]], ['纯臻营养护发素', 0.964739]]
[[[24.0, 80.0], [172.0, 80.0], [172.0, 104.0], [24.0, 104.0]], ['产品信息/参数', 0.98069626]]
[[[24.0, 109.0], [333.0, 109.0], [333.0, 136.0], [24.0, 136.0]], ['(45元/每公斤,100公斤起订)', 0.9676722]]
......
```
结果可视化

<div align="center">
    <img src="../imgs_results/whl/11_det_rec.jpg" width="800">
</div>


* 检测+识别
WenmuZhou's avatar
WenmuZhou committed
58
59
```python
from paddleocr import PaddleOCR, draw_ocr
WenmuZhou's avatar
WenmuZhou committed
60
ocr = PaddleOCR() # need to run only once to download and load model into memory
WenmuZhou's avatar
WenmuZhou committed
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
img_path = 'PaddleOCR/doc/imgs/11.jpg'
result = ocr.ocr(img_path)
for line in result:
    print(line)

# 显示结果
from PIL import Image
image = Image.open(img_path).convert('RGB')
boxes = [line[0] for line in result]
txts = [line[1][0] for line in result]
scores = [line[1][1] for line in result]
im_show = draw_ocr(image, boxes, txts, scores, font_path='/path/to/PaddleOCR/doc/simfang.ttf')
im_show = Image.fromarray(im_show)
im_show.save('result.jpg')
```
结果是一个list,每个item包含了文本框,文字和识别置信度
```bash
[[[24.0, 36.0], [304.0, 34.0], [304.0, 72.0], [24.0, 74.0]], ['纯臻营养护发素', 0.964739]]
[[[24.0, 80.0], [172.0, 80.0], [172.0, 104.0], [24.0, 104.0]], ['产品信息/参数', 0.98069626]]
[[[24.0, 109.0], [333.0, 109.0], [333.0, 136.0], [24.0, 136.0]], ['(45元/每公斤,100公斤起订)', 0.9676722]]
WenmuZhou's avatar
WenmuZhou committed
81
......
WenmuZhou's avatar
WenmuZhou committed
82
83
84
85
86
87
88
```
结果可视化

<div align="center">
    <img src="../imgs_results/whl/11_det_rec.jpg" width="800">
</div>

WenmuZhou's avatar
WenmuZhou committed
89

WenmuZhou's avatar
WenmuZhou committed
90
* 方向分类器+识别
WenmuZhou's avatar
WenmuZhou committed
91
92
93
94
95
96
97
98
99
100
101
102
103
```python
from paddleocr import PaddleOCR
ocr = PaddleOCR(use_angle_cls=True) # need to run only once to download and load model into memory
img_path = 'PaddleOCR/doc/imgs_words/ch/word_1.jpg'
result = ocr.ocr(img_path, det=False, cls=True)
for line in result:
    print(line)
```
结果是一个list,每个item只包含识别结果和识别置信度
```bash
['韩国小馆', 0.9907421]
```

WenmuZhou's avatar
WenmuZhou committed
104
105
106
* 单独执行检测
```python
from paddleocr import PaddleOCR, draw_ocr
WenmuZhou's avatar
WenmuZhou committed
107
ocr = PaddleOCR() # need to run only once to download and load model into memory
WenmuZhou's avatar
WenmuZhou committed
108
img_path = 'PaddleOCR/doc/imgs/11.jpg'
WenmuZhou's avatar
WenmuZhou committed
109
result = ocr.ocr(img_path, rec=False)
WenmuZhou's avatar
WenmuZhou committed
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
for line in result:
    print(line)

# 显示结果
from PIL import Image

image = Image.open(img_path).convert('RGB')
im_show = draw_ocr(image, result, txts=None, scores=None, font_path='/path/to/PaddleOCR/doc/simfang.ttf')
im_show = Image.fromarray(im_show)
im_show.save('result.jpg')
```
结果是一个list,每个item只包含文本框
```bash
[[26.0, 457.0], [137.0, 457.0], [137.0, 477.0], [26.0, 477.0]]
[[25.0, 425.0], [372.0, 425.0], [372.0, 448.0], [25.0, 448.0]]
[[128.0, 397.0], [273.0, 397.0], [273.0, 414.0], [128.0, 414.0]]
WenmuZhou's avatar
WenmuZhou committed
126
......
WenmuZhou's avatar
WenmuZhou committed
127
128
129
130
131
132
133
134
135
136
137
```
结果可视化


<div align="center">
    <img src="../imgs_results/whl/11_det.jpg" width="800">
</div>

* 单独执行识别
```python
from paddleocr import PaddleOCR
WenmuZhou's avatar
WenmuZhou committed
138
ocr = PaddleOCR() # need to run only once to download and load model into memory
WenmuZhou's avatar
WenmuZhou committed
139
img_path = 'PaddleOCR/doc/imgs_words/ch/word_1.jpg'
WenmuZhou's avatar
WenmuZhou committed
140
result = ocr.ocr(img_path, det=False)
WenmuZhou's avatar
WenmuZhou committed
141
142
143
144
145
146
147
148
for line in result:
    print(line)
```
结果是一个list,每个item只包含识别结果和识别置信度
```bash
['韩国小馆', 0.9907421]
```

WenmuZhou's avatar
WenmuZhou committed
149
* 单独执行方向分类器
WenmuZhou's avatar
WenmuZhou committed
150
151
152
153
154
155
156
157
158
159
160
161
162
```python
from paddleocr import PaddleOCR
ocr = PaddleOCR(use_angle_cls=True) # need to run only once to download and load model into memory
img_path = 'PaddleOCR/doc/imgs_words/ch/word_1.jpg'
result = ocr.ocr(img_path, det=False, rec=False, cls=True)
for line in result:
    print(line)
```
结果是一个list,每个item只包含分类结果和分类置信度
```bash
['0', 0.9999924]
```

WenmuZhou's avatar
WenmuZhou committed
163
### 2.2 通过命令行使用
WenmuZhou's avatar
WenmuZhou committed
164
165
166
167
168
169

查看帮助信息
```bash
paddleocr -h
```

WenmuZhou's avatar
WenmuZhou committed
170
* 检测+方向分类器+识别全流程
WenmuZhou's avatar
WenmuZhou committed
171
```bash
WenmuZhou's avatar
WenmuZhou committed
172
paddleocr --image_dir PaddleOCR/doc/imgs/11.jpg --use_angle_cls true
WenmuZhou's avatar
WenmuZhou committed
173
174
175
176
177
178
179
180
181
182
```
结果是一个list,每个item包含了文本框,文字和识别置信度
```bash
[[[24.0, 36.0], [304.0, 34.0], [304.0, 72.0], [24.0, 74.0]], ['纯臻营养护发素', 0.964739]]
[[[24.0, 80.0], [172.0, 80.0], [172.0, 104.0], [24.0, 104.0]], ['产品信息/参数', 0.98069626]]
[[[24.0, 109.0], [333.0, 109.0], [333.0, 136.0], [24.0, 136.0]], ['(45元/每公斤,100公斤起订)', 0.9676722]]
......
```

* 检测+识别
WenmuZhou's avatar
WenmuZhou committed
183
184
185
186
187
188
189
190
```bash
paddleocr --image_dir PaddleOCR/doc/imgs/11.jpg
```
结果是一个list,每个item包含了文本框,文字和识别置信度
```bash
[[[24.0, 36.0], [304.0, 34.0], [304.0, 72.0], [24.0, 74.0]], ['纯臻营养护发素', 0.964739]]
[[[24.0, 80.0], [172.0, 80.0], [172.0, 104.0], [24.0, 104.0]], ['产品信息/参数', 0.98069626]]
[[[24.0, 109.0], [333.0, 109.0], [333.0, 136.0], [24.0, 136.0]], ['(45元/每公斤,100公斤起订)', 0.9676722]]
WenmuZhou's avatar
WenmuZhou committed
191
......
WenmuZhou's avatar
WenmuZhou committed
192
193
```

WenmuZhou's avatar
WenmuZhou committed
194
* 方向分类器+识别
WenmuZhou's avatar
WenmuZhou committed
195
```bash
WenmuZhou's avatar
WenmuZhou committed
196
paddleocr --image_dir PaddleOCR/doc/imgs_words/ch/word_1.jpg --use_angle_cls true --det false
WenmuZhou's avatar
WenmuZhou committed
197
198
199
200
201
202
203
```

结果是一个list,每个item只包含识别结果和识别置信度
```bash
['韩国小馆', 0.9907421]
```

WenmuZhou's avatar
WenmuZhou committed
204
205
206
207
208
209
210
211
212
* 单独执行检测
```bash
paddleocr --image_dir PaddleOCR/doc/imgs/11.jpg --rec false
```
结果是一个list,每个item只包含文本框
```bash
[[26.0, 457.0], [137.0, 457.0], [137.0, 477.0], [26.0, 477.0]]
[[25.0, 425.0], [372.0, 425.0], [372.0, 448.0], [25.0, 448.0]]
[[128.0, 397.0], [273.0, 397.0], [273.0, 414.0], [128.0, 414.0]]
WenmuZhou's avatar
WenmuZhou committed
213
......
WenmuZhou's avatar
WenmuZhou committed
214
215
216
217
218
219
220
221
222
223
224
225
```

* 单独执行识别
```bash
paddleocr --image_dir PaddleOCR/doc/imgs_words/ch/word_1.jpg --det false
```

结果是一个list,每个item只包含识别结果和识别置信度
```bash
['韩国小馆', 0.9907421]
```

WenmuZhou's avatar
WenmuZhou committed
226
* 单独执行方向分类器
WenmuZhou's avatar
WenmuZhou committed
227
```bash
WenmuZhou's avatar
WenmuZhou committed
228
paddleocr --image_dir PaddleOCR/doc/imgs_words/ch/word_1.jpg --use_angle_cls true --det false --rec false
WenmuZhou's avatar
WenmuZhou committed
229
230
231
232
233
234
235
```

结果是一个list,每个item只包含分类结果和分类置信度
```bash
['0', 0.9999924]
```

WenmuZhou's avatar
WenmuZhou committed
236
## 3 自定义模型
WenmuZhou's avatar
WenmuZhou committed
237
当内置模型无法满足需求时,需要使用到自己训练的模型。
WenmuZhou's avatar
WenmuZhou committed
238
首先,参照[inference.md](./inference.md) 第一节转换将检测、分类和识别模型转换为inference模型,然后按照如下方式使用
WenmuZhou's avatar
WenmuZhou committed
239

WenmuZhou's avatar
WenmuZhou committed
240
### 3.1 代码使用
WenmuZhou's avatar
WenmuZhou committed
241
242
```python
from paddleocr import PaddleOCR, draw_ocr
WenmuZhou's avatar
WenmuZhou committed
243
244
# 模型路径下必须含有model和params文件
ocr = PaddleOCR(det_model_dir='{your_det_model_dir}', rec_model_dir='{your_rec_model_dir}', rec_char_dict_path='{your_rec_char_dict_path}', cls_model_dir='{your_cls_model_dir}', use_angle_cls=True)
WenmuZhou's avatar
WenmuZhou committed
245
img_path = 'PaddleOCR/doc/imgs/11.jpg'
WenmuZhou's avatar
WenmuZhou committed
246
result = ocr.ocr(img_path, cls=True)
WenmuZhou's avatar
WenmuZhou committed
247
248
249
250
251
252
253
254
255
256
257
258
259
260
for line in result:
    print(line)

# 显示结果
from PIL import Image
image = Image.open(img_path).convert('RGB')
boxes = [line[0] for line in result]
txts = [line[1][0] for line in result]
scores = [line[1][1] for line in result]
im_show = draw_ocr(image, boxes, txts, scores, font_path='/path/to/PaddleOCR/doc/simfang.ttf')
im_show = Image.fromarray(im_show)
im_show.save('result.jpg')
```

WenmuZhou's avatar
WenmuZhou committed
261
### 3.2 通过命令行使用
WenmuZhou's avatar
WenmuZhou committed
262
263

```bash
WenmuZhou's avatar
WenmuZhou committed
264
paddleocr --image_dir PaddleOCR/doc/imgs/11.jpg --det_model_dir {your_det_model_dir} --rec_model_dir {your_rec_model_dir} --rec_char_dict_path {your_rec_char_dict_path} --cls_model_dir {your_cls_model_dir} --use_angle_cls true
WenmuZhou's avatar
WenmuZhou committed
265
266
```

WenmuZhou's avatar
WenmuZhou committed
267
## 4 使用网络图片或者numpy数组作为输入
WenmuZhou's avatar
WenmuZhou committed
268

WenmuZhou's avatar
WenmuZhou committed
269
### 4.1 网络图片
WenmuZhou's avatar
WenmuZhou committed
270

WenmuZhou's avatar
WenmuZhou committed
271
- 代码使用
WenmuZhou's avatar
WenmuZhou committed
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
```python
from paddleocr import PaddleOCR, draw_ocr
# Paddleocr目前支持中英文、英文、法语、德语、韩语、日语,可以通过修改lang参数进行切换
# 参数依次为`ch`, `en`, `french`, `german`, `korean`, `japan`。
ocr = PaddleOCR(use_angle_cls=True, lang="ch") # need to run only once to download and load model into memory
img_path = 'http://n.sinaimg.cn/ent/transform/w630h933/20171222/o111-fypvuqf1838418.jpg'
result = ocr.ocr(img_path, cls=True)
for line in result:
    print(line)

# 显示结果
from PIL import Image
image = Image.open(img_path).convert('RGB')
boxes = [line[0] for line in result]
txts = [line[1][0] for line in result]
scores = [line[1][1] for line in result]
im_show = draw_ocr(image, boxes, txts, scores, font_path='/path/to/PaddleOCR/doc/simfang.ttf')
im_show = Image.fromarray(im_show)
im_show.save('result.jpg')
```
WenmuZhou's avatar
WenmuZhou committed
292
- 命令行模式
WenmuZhou's avatar
WenmuZhou committed
293
294
295
296
```bash
paddleocr --image_dir http://n.sinaimg.cn/ent/transform/w630h933/20171222/o111-fypvuqf1838418.jpg --use_angle_cls=true
```

WenmuZhou's avatar
WenmuZhou committed
297
### 4.2 numpy数组
WenmuZhou's avatar
WenmuZhou committed
298
299
300
301
302
303
304
305
306
仅通过代码使用时支持numpy数组作为输入
```python
from paddleocr import PaddleOCR, draw_ocr
# Paddleocr目前支持中英文、英文、法语、德语、韩语、日语,可以通过修改lang参数进行切换
# 参数依次为`ch`, `en`, `french`, `german`, `korean`, `japan`。
ocr = PaddleOCR(use_angle_cls=True, lang="ch") # need to run only once to download and load model into memory
img_path = 'PaddleOCR/doc/imgs/11.jpg'
img = cv2.imread(img_path)
# img = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY), 如果你自己训练的模型支持灰度图,可以将这句话的注释取消
WenmuZhou's avatar
WenmuZhou committed
307
result = ocr.ocr(img, cls=True)
WenmuZhou's avatar
WenmuZhou committed
308
309
310
311
312
313
314
315
316
317
318
319
320
321
for line in result:
    print(line)

# 显示结果
from PIL import Image
image = Image.open(img_path).convert('RGB')
boxes = [line[0] for line in result]
txts = [line[1][0] for line in result]
scores = [line[1][1] for line in result]
im_show = draw_ocr(image, boxes, txts, scores, font_path='/path/to/PaddleOCR/doc/simfang.ttf')
im_show = Image.fromarray(im_show)
im_show.save('result.jpg')
```

WenmuZhou's avatar
WenmuZhou committed
322
## 5 参数说明
WenmuZhou's avatar
WenmuZhou committed
323
324
325
326
327
328
329

| 字段                    | 说明                                                                                                                                                                                                                 | 默认值                  |
|-------------------------|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-------------------------|
| use_gpu                 | 是否使用GPU                                                                                                                                                                                                          | TRUE                    |
| gpu_mem                 | 初始化占用的GPU内存大小                                                                                                                                                                                              | 8000M                   |
| image_dir               | 通过命令行调用时执行预测的图片或文件夹路径                                                                                                                                                                           |                         |
| det_algorithm           | 使用的检测算法类型                                                                                                                                                                                                   | DB                      |
WenmuZhou's avatar
WenmuZhou committed
330
| det_model_dir          |  检测模型所在文件夹。传参方式有两种,1. None: 自动下载内置模型到 `~/.paddleocr/det`;2.自己转换好的inference模型路径,模型路径下必须包含model和params文件 |   None        |
WenmuZhou's avatar
WenmuZhou committed
331
332
333
334
335
336
337
338
| det_max_side_len        | 检测算法前向时图片长边的最大尺寸,当长边超出这个值时会将长边resize到这个大小,短边等比例缩放                                                                                                                         | 960                     |
| det_db_thresh           | DB模型输出预测图的二值化阈值                                                                                                                                                                                         | 0.3                     |
| det_db_box_thresh       | DB模型输出框的阈值,低于此值的预测框会被丢弃                                                                                                                                                                           | 0.5                     |
| det_db_unclip_ratio     | DB模型输出框扩大的比例                                                                                                                                                                                               | 2                       |
| det_east_score_thresh   | EAST模型输出预测图的二值化阈值                                                                                                                                                                                       | 0.8                     |
| det_east_cover_thresh   | EAST模型输出框的阈值,低于此值的预测框会被丢弃                                                                                                                                                                         | 0.1                     |
| det_east_nms_thresh     | EAST模型输出框NMS的阈值                                                                                                                                                                                              | 0.2                     |
| rec_algorithm           | 使用的识别算法类型                                                                                                                                                                                                   | CRNN                    |
WenmuZhou's avatar
WenmuZhou committed
339
| rec_model_dir          | 识别模型所在文件夹。传参方式有两种,1. None: 自动下载内置模型到 `~/.paddleocr/rec`;2.自己转换好的inference模型路径,模型路径下必须包含model和params文件 | None |
WenmuZhou's avatar
WenmuZhou committed
340
| rec_image_shape         | 识别算法的输入图片尺寸                                                                                                                                                                                             | "3,32,320"              |
WenmuZhou's avatar
WenmuZhou committed
341
| rec_char_type           | 识别算法的字符类型,中英文(ch)、英文(en)、法语(french)、德语(german)、韩语(korean)、日语(japan)                                                                                                                                                                               | ch                      |
WenmuZhou's avatar
WenmuZhou committed
342
| rec_batch_num           | 进行识别时,同时前向的图片数                                                                                                                                                                                         | 30                      |
WenmuZhou's avatar
WenmuZhou committed
343
344
| max_text_length         | 识别算法能识别的最大文字长度                                                                                                                                                                                         | 25                      |
| rec_char_dict_path      | 识别模型字典路径,当rec_model_dir使用方式2传参时需要修改为自己的字典路径                                                                                                                                                | ./ppocr/utils/ppocr_keys_v1.txt                        |
WenmuZhou's avatar
WenmuZhou committed
345
| use_space_char          | 是否识别空格                                                                                                                                                                                                         | TRUE                    |
WenmuZhou's avatar
WenmuZhou committed
346
| drop_score          | 对输出按照分数(来自于识别模型)进行过滤,低于此分数的不返回                                                                                                                                                                                                         | 0.5                    |
WenmuZhou's avatar
WenmuZhou committed
347
348
349
350
351
| use_angle_cls          | 是否加载分类模型                                                                                                                                                                                                         | FALSE                    |
| cls_model_dir          | 分类模型所在文件夹。传参方式有两种,1. None: 自动下载内置模型到 `~/.paddleocr/cls`;2.自己转换好的inference模型路径,模型路径下必须包含model和params文件                                                                                 | None                    |
| cls_image_shape          | 分类算法的输入图片尺寸                                                                           | "3, 48, 192"                    |
| label_list          | 分类算法的标签列表                                                                           | ['0', '180']                  |
| cls_batch_num          | 进行分类时,同时前向的图片数                                                                          |30                 |
WenmuZhou's avatar
WenmuZhou committed
352
| enable_mkldnn           | 是否启用mkldnn                                                                                                                                                                                                       | FALSE                   |
WenmuZhou's avatar
WenmuZhou committed
353
| use_zero_copy_run           | 是否通过zero_copy_run的方式进行前向                                                                                                                                                                               | FALSE                   |
WenmuZhou's avatar
WenmuZhou committed
354
| lang                     | 模型语言类型,目前支持 目前支持中英文(ch)、英文(en)、法语(french)、德语(german)、韩语(korean)、日语(japan)                                                                                                                                                                                               | ch                    |
WenmuZhou's avatar
WenmuZhou committed
355
356
| det                     | 前向时使用启动检测                                                                                                                                                                                                   | TRUE                    |
| rec                     | 前向时是否启动识别                                                                                                                                                                                                   | TRUE                    |
WenmuZhou's avatar
WenmuZhou committed
357
| cls                     | 前向时是否启动分类 (命令行模式下使用use_angle_cls控制前向是否启动分类)                                                                                                                                                                                                | FALSE                    |