File size: 9,111 Bytes
2c67080
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
<figure markdown>

[中文](examples.md) | English

</figure>

# Examples
## Recognize PDF Files and Return Markdown Format

For PDF files, you can use the `.recognize_pdf()` function to recognize the entire file or specific pages and output the results as a Markdown file. For example, for the following PDF file ([examples/test-doc.pdf](examples/test-doc.pdf)), 
you can call the function like this:

```python
from pix2text import Pix2Text

img_fp = './examples/test-doc.pdf'
p2t = Pix2Text.from_config()
doc = p2t.recognize_pdf(img_fp, page_numbers=[0, 1])
doc.to_markdown('output-md')  # The exported Markdown information is saved in the output-md directory
```

You can also achieve the same functionality using the command line. Below is a command that uses the premium models (MFD + MFR + CnOCR) for recognition:

```bash
p2t predict -l en,ch_sim --mfd-config '{"model_name": "mfd-pro-1.5", "model_backend": "onnx"}' --formula-ocr-config '{"model_name":"mfr-pro-1.5","model_backend":"onnx"}' --text-ocr-config '{"rec_model_name": "doc-densenet_lite_666-gru_large"}' --rec-kwargs '{"page_numbers": [0, 1]}' --resized-shape 768 --file-type pdf -i docs/examples/test-doc.pdf -o output-md --save-debug-res output-debug
```

The recognition result can be found in [output-md/output.md](output-md/output.md).

<br/>

> If you wish to export formats other than Markdown, such as Word, HTML, PDF, etc., it is recommended to use the tool [Pandoc](https://pandoc.org) to convert the Markdown result.

## Recognize Images with Complex Layout

You can use the `.recognize_page()` function to recognize text and mathematical formulas in images. For example, for the following image ([examples/page2.png](examples/page2.png)):

<figure markdown>
![Page-image](examples/page2.png){: style="width:600px"}
</figure>

You can call the function like this:

```python
from pix2text import Pix2Text

img_fp = './examples/test-doc.pdf'
p2t = Pix2Text.from_config()
page = p2t.recognize_page(img_fp)
page.to_markdown('output-page')  # The exported Markdown information is saved in the output-page directory
```

You can also achieve the same functionality using the command line. Below is a command that uses the premium models (MFD + MFR + CnOCR) for recognition:

```bash
p2t predict -l en,ch_sim --mfd-config '{"model_name": "mfd-pro-1.5", "model_backend": "onnx"}' --formula-ocr-config '{"model_name":"mfr-pro-1.5","model_backend":"onnx"}' --text-ocr-config '{"rec_model_name": "doc-densenet_lite_666-gru_large"}' --resized-shape 768 --file-type page -i docs/examples/page2.png -o output-page --save-debug-res output-debug-page
```

The recognition result is similar to [output-md/output.md](output-md/output.md).


## Recognize Paragraph Images with Both Formulas and Texts

For paragraph images containing both formulas and texts, you don't need to use the layout analysis model. You can use the `.recognize_text_formula()` function to recognize both texts and mathematical formulas in the image. For example, for the following image ([examples/en1.jpg](examples/en1.jpg)):

<figure markdown>
![English-mixed-image](examples/en1.jpg){: style="width:600px"}
</figure>

You can call the function like this:

```python
from pix2text import Pix2Text, merge_line_texts

img_fp = './examples/en1.jpg'
p2t = Pix2Text.from_config()
outs = p2t.recognize_text_formula(img_fp, resized_shape=768, return_text=True)
print(outs)
```

The returned result `outs` is a dictionary, where the key `position` represents the box position information, `type` represents the category information, and `text` represents the recognition result. For detailed explanations, see [API Documentation](#api-documentation).

You can also achieve the same functionality using the command line. Below is a command that uses the premium models (MFD + MFR + CnOCR) for recognition:

```bash
p2t predict -l en,ch_sim --mfd-config '{"model_name": "mfd-pro-1.5", "model_backend": "onnx"}' --formula-ocr-config '{"model_name":"mfr-pro-1.5","model_backend":"onnx"}' --text-ocr-config '{"rec_model_name": "doc-densenet_lite_666-gru_large"}' --resized-shape 768 --file-type text_formula -i docs/examples/en1.jpg --save-debug-res out-debug-en1.jpg
```

Or use the free open-source models for recognition:

```bash
p2t predict -l en,ch_sim --resized-shape 768 --file-type text_formula -i docs/examples/en1.jpg --save-debug-res out-debug-en1.jpg
```

## Recognize Pure Formula Images

For images containing only mathematical formulas, you can use the `.recognize_formula()` function to recognize the formulas as LaTeX expressions. For example, for the following image ([examples/math-formula-42.png](examples/math-formula-42.png)):

<figure markdown>
![Pure-Math-Formula-image](examples/math-formula-42.png){: style="width:300px"}
</figure>

You can call the function like this:

```python
from pix2text import Pix2Text

img_fp = './examples/math-formula-42.png'
p2t = Pix2Text.from_config()
outs = p2t.recognize_formula(img_fp)
print(outs)
```

The returned result is a string representing the corresponding LaTeX expression. For detailed explanations, see [Usage](usage.md).

You can also achieve the same functionality using the command line. Below is a command that uses the premium model (MFR) for recognition:

```bash
p2t predict -l en,ch_sim --formula-ocr-config '{"model_name":"mfr-pro-1.5","model_backend":"onnx"}' --file-type formula -i docs/examples/math-formula-42.png
```

Or use the free open-source model for recognition:

```bash
p2t predict -l en,ch_sim --file-type formula -i docs/examples/math-formula-42.png
```

## Recognize Pure Text Images

For images containing only text without mathematical formulas, you can use the `.recognize_text()` function to recognize the text in the image. In this case, Pix2Text acts as a general text OCR engine. For example, for the following image ([examples/general.jpg](examples/general.jpg)):

<figure markdown>
![Scene-Text](examples/general.jpg){: style="width:400px"}
</figure>

You can call the function like this:

```python
from pix2text import Pix2Text

img_fp = './examples/general.jpg'
p2t = Pix2Text.from_config()
outs = p2t.recognize_text(img_fp)
print(outs)
```

The returned result is a string representing the corresponding text sequence. For detailed explanations, see [API Documentation](https://pix2text.readthedocs.io/zh-cn/latest/pix2text/pix_to_text/).

You can also achieve the same functionality using the command line. Below is a command that uses the premium model (CnOCR) for recognition:

```bash
p2t predict -l en,ch_sim --text-ocr-config '{"rec_model_name": "doc-densenet_lite_666-gru_large"}' --file-type text --no-return-text -i docs/examples/general.jpg --save-debug-res out-debug-general.jpg
```

Or use the free open-source model for recognition:

```bash
p2t predict -l en,ch_sim --file-type text --no-return-text -i docs/examples/general.jpg --save-debug-res out-debug-general.jpg
```

## For Different Languages

### English

**Recognition Result**:

![Pix2Text Recognizing English](figs/output-en.jpg)

**Recognition Command**:

```bash
p2t predict -l en --mfd-config '{"model_name": "mfd-pro-1.5", "model_backend": "onnx"}' --formula-ocr-config '{"model_name":"mfr-pro-1.5","model_backend":"onnx"}' --text-ocr-config '{"rec_model_name": "doc-densenet_lite_666-gru_large"}' --resized-shape 768 --file-type text_formula -i docs/examples/en1.jpg
```

### Simplified Chinese

**Recognition Result**:

![Pix2Text Recognizing Simplified Chinese](figs/output-ch_sim.jpg)

**Recognition Command**:

```bash
p2t predict -l en,ch_sim --mfd-config '{"model_name": "mfd-pro-1.5", "model_backend": "onnx"}' --formula-ocr-config '{"model_name":"mfr-pro-1.5","model_backend":"onnx"}' --text-ocr-config '{"rec_model_name": "doc-densenet_lite_666-gru_large"}' --resized-shape 768 --auto-line-break --file-type text_formula -i docs/examples/mixed.jpg --save-debug-res out-debug-mixed.jpg
```

### Traditional Chinese

**Recognition Result**:

![Pix2Text Recognizing Traditional Chinese](figs/output-ch_tra.jpg)

**Recognition Command**:

```bash
p2t predict -l en,ch_tra --mfd-config '{"model_name": "mfd-pro-1.5", "model_backend": "onnx"}' --formula-ocr-config '{"model_name":"mfr-pro-1.5","model_backend":"onnx"}' --resized-shape 768 --auto-line-break --file-type text_formula -i docs/examples/ch_tra.jpg --save-debug-res out-debug-tra.jpg
```

> Note ⚠️: Please install the multilingual version of pix2text using the following command:
> ```bash
> pip install pix2text[multilingual]
> ```

### Vietnamese

**Recognition Result**:

![Pix2Text Recognizing Vietnamese](figs/output-vietnamese.jpg)

**Recognition Command**:

```bash
p2t predict -l en,vi --mfd-config '{"model_name": "mfd-pro-1.5", "model_backend": "onnx"}' --formula-ocr-config '{"model_name":"mfr-pro-1.5","model_backend":"onnx"}' --resized-shape 608 --no-auto-line-break --file-type text_formula -i docs/examples/vietnamese.jpg --save-debug-res out-debug-vi.jpg
```

> Note ⚠️: Please install the multilingual version of pix2text using the following command:
> ```bash
> pip install pix2text[multilingual]
> ```