Documents/text_recognition.en.md

---
comments: true
---

# Text Recognition Module Tutorial

## 1. Overview

The text recognition module is the core part of the OCR (Optical Character Recognition) system, responsible for extracting text information from text regions in images. The performance of this module directly affects the accuracy and efficiency of the entire OCR system. The text recognition module usually receives the bounding boxes of text regions output by the text detection module as input, and then converts the text in the images into editable and searchable electronic text through complex image processing and deep learning algorithms. The accuracy of text recognition results is crucial for subsequent applications such as information extraction and data mining.

## 2. List of Supported Models

<table>
<tr>
<th>Model</th><th>Model Download Links</th>
<th>Recognition Avg Accuracy(%)</th>
<th>GPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
<th>CPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
<th>Model Storage Size (MB)</th>
<th>Introduction</th>
</tr>
<tr>
<td>PP-OCRv5_server_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
PP-OCRv5_server_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-OCRv5_server_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>86.38</td>
<td>8.46 / 2.36</td>
<td>31.21 / 31.21</td>
<td>81</td>
<td rowspan="2">PP-OCRv5_rec is a new generation text recognition model. It is designed to efficiently and accurately support the recognition of Simplified Chinese, Traditional Chinese, English, Japanese, as well as complex text scenarios such as handwriting, vertical text, pinyin, and rare characters with a single model. While maintaining recognition performance, it also balances inference speed and model robustness, providing efficient and accurate technical support for document understanding in various scenarios.</td>
</tr>
<tr>
<td>PP-OCRv5_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
PP-OCRv5_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-OCRv5_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>81.29</td>
<td>5.43 / 1.46</td>
<td>21.20 / 5.32</td>
<td>16</td>
</tr>
<tr>
<td>PP-OCRv4_server_rec_doc</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
PP-OCRv4_server_rec_doc_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-OCRv4_server_rec_doc_pretrained.pdparams">Pretrained Model</a></td>
<td>86.58</td>
<td>8.69 / 2.78</td>
<td>37.93 / 37.93</td>
<td>182</td>
<td>PP-OCRv4_server_rec_doc is trained on a mixed dataset of more Chinese document data and PP-OCR training data, building upon PP-OCRv4_server_rec. It enhances the recognition capabilities for some Traditional Chinese characters, Japanese characters, and special symbols, supporting over 15,000 characters. In addition to improving document-related text recognition, it also enhances general text recognition capabilities.</td>
</tr>
<tr>
<td>PP-OCRv4_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-OCRv4_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-OCRv4_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>78.74</td>
<td>5.26 / 1.12</td>
<td>17.48 / 3.61</td>
<td>10.5</td>
<td>A lightweight recognition model of PP-OCRv4 with high inference efficiency, suitable for deployment on various hardware devices, including edge devices.</td>
</tr>
<tr>
<td>PP-OCRv4_server_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-OCRv4_server_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-OCRv4_server_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>85.19</td>
<td>8.75 / 2.49</td>
<td>36.93 / 36.93</td>
<td>173</td>
<td>The server-side model of PP-OCRv4, offering high inference accuracy and deployable on various servers.</td>
</tr>
<tr>
<td>en_PP-OCRv4_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
en_PP-OCRv4_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/en_PP-OCRv4_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>70.39</td>
<td>4.81 / 1.23</td>
<td>17.20 / 4.18</td>
<td>7.5</td>
<td>An ultra-lightweight English recognition model trained based on the PP-OCRv4 recognition model, supporting English and numeric character recognition.</td>
</tr>
</table>

> ❗ The above lists the <b>4 core models</b> mainly supported by the text recognition module. The module supports a total of <b>20 full models</b>, including multiple multilingual text recognition models. The complete model list is as follows:

<details><summary> 👉Model List Details</summary>

* <b>PP-OCRv5 Multi-Scenario Models</b>

<table>
<tr>
<th>Model</th><th>Model Download Links</th>
<th>Chinese Recognition Avg Accuracy(%)</th>
<th>English Recognition Avg Accuracy(%)</th>
<th>Traditional Chinese Recognition Avg Accuracy(%)</th>
<th>Japanese Recognition Avg Accuracy(%)</th>
<th>GPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
<th>CPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
<th>Model Storage Size (MB)</th>
<th>Introduction</th>
</tr>
<tr>
<td>PP-OCRv5_server_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
PP-OCRv5_server_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-OCRv5_server_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>86.38</td>
<td>64.70</td>
<td>93.29</td>
<td>60.35</td>
<td>8.46 / 2.36</td>
<td>31.21 / 31.21</td>
<td>81</td>
<td rowspan="2">PP-OCRv5_rec is a new generation text recognition model. It is designed to efficiently and accurately support the recognition of Simplified Chinese, Traditional Chinese, English, Japanese, as well as complex text scenarios such as handwriting, vertical text, pinyin, and rare characters with a single model. While maintaining recognition performance, it also balances inference speed and model robustness, providing efficient and accurate technical support for document understanding in various scenarios.</td>
</tr>
<tr>
<td>PP-OCRv5_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
PP-OCRv5_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-OCRv5_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>81.29</td>
<td>66.00</td>
<td>83.55</td>
<td>54.65</td>
<td>5.43 / 1.46</td>
<td>21.20 / 5.32</td>
<td>16</td>
</tr>
</table>

* <b>Chinese Recognition Models</b>
<table>
<tr>
<th>Model</th><th>Model Download Links</th>
<th>Recognition Avg Accuracy(%)</th>
<th>GPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
<th>CPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
<th>Model Storage Size (MB)</th>
<th>Introduction</th>
</tr>
<tr>
<td>PP-OCRv4_server_rec_doc</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
PP-OCRv4_server_rec_doc_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-OCRv4_server_rec_doc_pretrained.pdparams">Pretrained Model</a></td>
<td>86.58</td>
<td>8.69 / 2.78</td>
<td>37.93 / 37.93</td>
<td>182</td>
<td>PP-OCRv4_server_rec_doc is trained on a mixed dataset of more Chinese document data and PP-OCR training data, building upon PP-OCRv4_server_rec. It enhances the recognition capabilities for some Traditional Chinese characters, Japanese characters, and special symbols, supporting over 15,000 characters. In addition to improving document-related text recognition, it also enhances general text recognition capabilities.</td>
</tr>
<tr>
<td>PP-OCRv4_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-OCRv4_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-OCRv4_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>78.74</td>
<td>5.26 / 1.12</td>
<td>17.48 / 3.61</td>
<td>10.5</td>
<td>A lightweight recognition model of PP-OCRv4 with high inference efficiency, suitable for deployment on various hardware devices, including edge devices.</td>
</tr>
<tr>
<td>PP-OCRv4_server_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-OCRv4_server_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-OCRv4_server_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>85.19</td>
<td>8.75 / 2.49</td>
<td>36.93 / 36.93</td>
<td>173</td>
<td>The server-side model of PP-OCRv4, offering high inference accuracy and deployable on various servers.</td>
</tr>
<tr>
<td>PP-OCRv3_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
PP-OCRv3_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-OCRv3_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>72.96</td>
<td>3.89 / 1.16</td>
<td>8.72 / 3.56</td>
<td>10.3</td>
<td>A lightweight recognition model of PP-OCRv3 with high inference efficiency, suitable for deployment on various hardware devices, including edge devices.</td>
</tr>
</table>

<table>
<tr>
<th>Model</th><th>Model Download Links</th>
<th>Recognition Avg Accuracy(%)</th>
<th>GPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
<th>CPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
<th>Model Storage Size (MB)</th>
<th>Introduction</th>
</tr>
<tr>
<td>ch_SVTRv2_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/ch_SVTRv2_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/ch_SVTRv2_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>68.81</td>
<td>10.38 / 8.31</td>
<td>66.52 / 30.83</td>
<td>80.5</td>
<td rowspan="1">SVTRv2 is a server-side text recognition model developed by the OpenOCR team of the Vision and Learning Lab (FVL) at Fudan University. It won the first prize in the PaddleOCR Algorithm Model Challenge - Task 1: OCR End-to-End Recognition Task, with a 6% improvement in end-to-end recognition accuracy on Leaderboard A compared to PP-OCRv4.</td>
</tr>
</table>

<table>
<tr>
<th>Model</th><th>Model Download Links</th>
<th>Recognition Avg Accuracy(%)</th>
<th>GPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
<th>CPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
<th>Model Storage Size (MB)</th>
<th>Introduction</th>
</tr>
<tr>
<td>ch_RepSVTR_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/ch_RepSVTR_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/ch_RepSVTR_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>65.07</td>
<td>6.29 / 1.57</td>
<td>20.64 / 5.40</td>
<td>22.1</td>
<td rowspan="1">RepSVTR is a mobile-side text recognition model based on SVTRv2. It won the first prize in the PaddleOCR Algorithm Model Challenge - Task 1: OCR End-to-End Recognition Task, with a 2.5% improvement in end-to-end recognition accuracy on Leaderboard B compared to PP-OCRv4, while maintaining similar inference speed.</td>
</tr>
</table>

* <b>English Recognition Models</b>
<table>
<tr>
<th>Model</th><th>Model Download Links</th>
<th>Recognition Avg Accuracy(%)</th>
<th>GPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
<th>CPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
<th>Model Storage Size (MB)</th>
<th>Introduction</th>
</tr>
<tr>
<td>en_PP-OCRv4_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
en_PP-OCRv4_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/en_PP-OCRv4_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td> 70.39</td>
<td>4.81 / 1.23</td>
<td>17.20 / 4.18</td>
<td>7.5</td>
<td>An ultra-lightweight English recognition model trained based on the PP-OCRv4 recognition model, supporting English and numeric character recognition.</td>
</tr>
<tr>
<td>en_PP-OCRv3_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
en_PP-OCRv3_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/en_PP-OCRv3_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>70.69</td>
<td>3.56 / 0.78</td>
<td>8.44 / 5.78</td>
<td>17.3</td>
<td>An ultra-lightweight English recognition model trained based on the PP-OCRv3 recognition model, supporting English and numeric character recognition.</td>
</tr>
</table>

* <b>Multilingual Recognition Models</b>
<table>
<tr>
<th>Model</th><th>Model Download Links</th>
<th>Recognition Avg Accuracy(%)</th>
<th>GPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
<th>CPU Inference Time (ms)<br/>[Normal Mode / High-Performance Mode]</th>
<th>Model Storage Size (MB)</th>
<th>Introduction</th>
</tr>
<tr>
<td>korean_PP-OCRv5_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
korean_PP-OCRv5_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/korean_PP-OCRv5_mobile_rec_pretrained.pdparams">Pre-trained Model</a></td>
<td>88.0</td>
<td>5.43 / 1.46</td>
<td>21.20 / 5.32</td>
<td>14</td>
<td>An ultra-lightweight Korean text recognition model trained based on the PP-OCRv5 recognition framework. Supports Korean, English and numeric text recognition.</td>
</tr>
<tr>
<td>latin_PP-OCRv5_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
latin_PP-OCRv5_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/latin_PP-OCRv5_mobile_rec_pretrained.pdparams">Pre-trained Model</a></td>
<td>84.7</td>
<td>5.43 / 1.46</td>
<td>21.20 / 5.32</td>
<td>14</td>
<td>A Latin-script text recognition model trained based on the PP-OCRv5 recognition framework. Supports most Latin alphabet languages and numeric text recognition.</td>
</tr>
<tr>
<td>eslav_PP-OCRv5_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
eslav_PP-OCRv5_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/eslav_PP-OCRv5_mobile_rec_pretrained.pdparams">Pre-trained Model</a></td>
<td>81.6</td>
<td>5.43 / 1.46</td>
<td>21.20 / 5.32</td>
<td>14</td>
<td>An East Slavic language recognition model trained based on the PP-OCRv5 recognition framework. Supports East Slavic languages, English and numeric text recognition.</td>
</tr>
<tr>
<td>korean_PP-OCRv3_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
korean_PP-OCRv3_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/korean_PP-OCRv3_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>60.21</td>
<td>3.73 / 0.98</td>
<td>8.76 / 2.91</td>
<td>9.6</td>
<td>An ultra-lightweight Korean recognition model trained based on the PP-OCRv3 recognition model, supporting Korean and numeric character recognition.</td>
</tr>
<tr>
<td>japan_PP-OCRv3_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
japan_PP-OCRv3_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/japan_PP-OCRv3_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>45.69</td>
<td>3.86 / 1.01</td>
<td>8.62 / 2.92</td>
<td>9.8</td>
<td>An ultra-lightweight Japanese recognition model trained based on the PP-OCRv3 recognition model, supporting Japanese and numeric character recognition.</td>
</tr>
<tr>
<td>chinese_cht_PP-OCRv3_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
chinese_cht_PP-OCRv3_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/chinese_cht_PP-OCRv3_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>82.06</td>
<td>3.90 / 1.16</td>
<td>9.24 / 3.18</td>
<td>10.8</td>
<td>An ultra-lightweight Traditional Chinese recognition model trained based on the PP-OCRv3 recognition model, supporting Traditional Chinese and numeric character recognition.</td>
</tr>
<tr>
<td>te_PP-OCRv3_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
te_PP-OCRv3_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/te_PP-OCRv3_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>95.88</td>
<td>3.59 / 0.81</td>
<td>8.28 / 6.21</td>
<td>8.7</td>
<td>An ultra-lightweight Telugu recognition model trained based on the PP-OCRv3 recognition model, supporting Telugu and numeric character recognition.</td>
</tr>
<tr>
<td>ka_PP-OCRv3_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
ka_PP-OCRv3_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/ka_PP-OCRv3_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>96.96</td>
<td>3.49 / 0.89</td>
<td>8.63 / 2.77</td>
<td>17.4</td>
<td>An ultra-lightweight Kannada recognition model trained based on the PP-OCRv3 recognition model, supporting Kannada and numeric character recognition.</td>
</tr>
<tr>
<td>ta_PP-OCRv3_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
ta_PP-OCRv3_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/ta_PP-OCRv3_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>76.83</td>
<td>3.49 / 0.86</td>
<td>8.35 / 3.41</td>
<td>8.7</td>
<td>An ultra-lightweight Tamil recognition model trained based on the PP-OCRv3 recognition model, supporting Tamil and numeric character recognition.</td>
</tr>
<tr>
<td>latin_PP-OCRv3_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
latin_PP-OCRv3_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/latin_PP-OCRv3_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>76.93</td>
<td>3.53 / 0.78</td>
<td>8.50 / 6.83</td>
<td>8.7</td>
<td>An ultra-lightweight Latin recognition model trained based on the PP-OCRv3 recognition model, supporting Latin and numeric character recognition.</td>
</tr>
<tr>
<td>arabic_PP-OCRv3_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
arabic_PP-OCRv3_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/arabic_PP-OCRv3_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>73.55</td>
<td>3.60 / 0.83</td>
<td>8.44 / 4.69</td>
<td>17.3</td>
<td>An ultra-lightweight Arabic alphabet recognition model trained based on the PP-OCRv3 recognition model, supporting Arabic alphabet and numeric character recognition.</td>
</tr>
<tr>
<td>cyrillic_PP-OCRv3_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
cyrillic_PP-OCRv3_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/cyrillic_PP-OCRv3_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>94.28</td>
<td>3.56 / 0.79</td>
<td>8.22 / 2.76</td>
<td>8.7</td>
<td>An ultra-lightweight Cyrillic alphabet recognition model trained based on the PP-OCRv3 recognition model, supporting Cyrillic alphabet and numeric character recognition.</td>
</tr>
<tr>
<td>devanagari_PP-OCRv3_mobile_rec</td>
<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
devanagari_PP-OCRv3_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/devanagari_PP-OCRv3_mobile_rec_pretrained.pdparams">Pretrained Model</a></td>
<td>96.44</td>
<td>3.60 / 0.78</td>
<td>6.95 / 2.87</td>
<td>8.7</td>
<td>An ultra-lightweight Devanagari alphabet recognition model trained based on the PP-OCRv3 recognition model, supporting Devanagari alphabet and numeric character recognition.</td>
</tr>
</table>

<strong>Test Environment Description:</strong>

  <ul>
      <li><b>Performance Test Environment</b>
          <ul>
              <li><strong>Test Dataset:</strong>
                 <ul>
                    <li>
                    Chinese Recognition Models: A self-built Chinese dataset by PaddleOCR, covering street views, online images, documents, handwriting, with 11,000 images for text recognition.
                    </li>
                    <li>
                      ch_SVTRv2_rec: <a href="https://aistudio.baidu.com/competition/detail/1131/0/introduction">PaddleOCR Algorithm Model Challenge - Task 1: OCR End-to-End Recognition Task</a> Leaderboard A evaluation set.
                    </li>
                    <li>
                      ch_RepSVTR_rec: <a href="https://aistudio.baidu.com/competition/detail/1131/0/introduction">PaddleOCR Algorithm Model Challenge - Task 1: OCR End-to-End Recognition Task</a> Leaderboard B evaluation set.
                    </li>
                    <li>
                      English Recognition Models: A self-built English dataset by PaddleOCR.
                    </li>
                    <li>
                      Multilingual Recognition Models: A self-built multilingual dataset by PaddleOCR.
                    </li>
                 </ul>
              </li>
              <li><strong>Hardware Configuration:</strong>
                  <ul>
                      <li>GPU: NVIDIA Tesla T4</li>
                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
                  </ul>
              </li>
              <li><strong>Software Environment:</strong>
                  <ul>
                      <li>Ubuntu 20.04 / CUDA 11.8 / cuDNN 8.9 / TensorRT 8.6.1.6</li>
                      <li>paddlepaddle 3.0.0 / paddleocr 3.0.3</li>
                  </ul>
              </li>
          </ul>
      </li>
      <li><b>Explanation of Inference Modes</b></li>
  </ul>

<table border="1">
    <thead>
        <tr>
            <th>Mode</th>
            <th>GPU Configuration</th>
            <th>CPU Configuration</th>
            <th>Acceleration Technology Combination</th>
        </tr>
    </thead>
    <tbody>
        <tr>
            <td>Normal Mode</td>
            <td>FP32 Precision / No TRT Acceleration</td>
            <td>FP32 Precision / 8 Threads</td>
            <td>PaddleInference</td>
        </tr>
        <tr>
            <td>High-Performance Mode</td>
            <td>Optimal combination of precision type and acceleration strategy</td>
            <td>FP32 Precision / 8 Threads</td>
            <td>Selection of the optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
        </tr>
    </tbody>
</table>

</details>

## 3. Quick Start

> ❗ Before starting, please install the PaddleOCR wheel package. For details, please refer to the [Installation Guide](../installation.en.md).

You can quickly experience it with one command:

```bash
paddleocr text_recognition -i https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_ocr_rec_001.png
```

<b>Note:</b> The official PaddleOCR models are downloaded from HuggingFace by default. If you cannot access HuggingFace, you can change the model source to BOS by setting the environment variable `PADDLE_PDX_MODEL_SOURCE="BOS"`. More mainstream model sources will be supported in the future.

You can also integrate the model inference of the text recognition module into your project. Before running the following code, please download the [sample image](https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_ocr_rec_001.png) to your local machine.

```python
from paddleocr import TextRecognition
model = TextRecognition(model_name="PP-OCRv5_server_rec")
output = model.predict(input="general_ocr_rec_001.png", batch_size=1)
for res in output:
    res.print()
    res.save_to_img(save_path="./output/")
    res.save_to_json(save_path="./output/res.json")
```

After running, the result is as follows:
```bash
{'res': {'input_path': 'general_ocr_rec_001.png', 'page_index': None, 'rec_text': '绿洲仕格维花园公寓', 'rec_score': 0.9823867082595825}}
```

The meanings of the parameters in the result are as follows:
- `input_path`: The path of the input text line image to be predicted
- `page_index`: If the input is a PDF file, it indicates which page of the PDF the current text line is from; otherwise, it is `None`
- `rec_text`: The predicted text of the text line image
- `rec_score`: The confidence score of the predicted text for the text line image

The visualized image is as follows:

<img src="https://raw.githubusercontent.com/cuicheng01/PaddleX_doc_images/refs/heads/main/images/modules/text_recog/general_ocr_rec_001.png"/>

Descriptions of related methods and parameters are as follows:

* Instantiate the text recognition model using `TextRecognition` (using `PP-OCRv5_server_rec` as an example), as follows:
<table>
<thead>
<tr>
<th>Parameter</th>
<th>Description</th>
<th>Type</th>
<th>Default</th>
</tr>
</thead>
<tbody>
<tr>
<td><code>model_name</code></td>
<td>If set to <code>None</code>, <code>PP-OCRv5_server_rec</code> is used.</td>
<td><code>str|None</code></td>
<td><code>None</code></td>
</tr>
<tr>
<td><code>model_dir</code></td>
<td>Model storage path.</td>
<td><code>str|None</code></td>
<td><code>None</code></td>
</tr>
<tr>
<td><code>device</code></td>
<td>Device for inference.<br/>
<b>Examples:</b> <code>"cpu"</code>, <code>"gpu"</code>, <code>"npu"</code>, <code>"gpu:0"</code>, <code>"gpu:0,1"</code>.<br/>
If multiple devices are specified, inference will be performed in parallel.<br/>
By default, GPU 0 is used; if unavailable, CPU is used.
</td>
<td><code>str|None</code></td>
<td><code>None</code></td>
</tr>
<tr>
<td><code>enable_hpi</code></td>
<td>Whether to enable high performance inference.</td>
<td><code>bool</code></td>
<td><code>False</code></td>
</tr>
<tr>
<td><code>use_tensorrt</code></td>
<td>Whether to enable the TensorRT subgraph engine of Paddle Inference.<br/>
For Paddle with CUDA 11.8, the compatible TensorRT version is 8.x (x>=6), recommended 8.6.1.6.<br/>

</td>
<td><code>bool</code></td>
<td><code>False</code></td>
</tr>
<tr>
<td><code>precision</code></td>
<td>Precision for TensorRT when using the Paddle Inference TensorRT subgraph engine.<br/><b>Options:</b> <code>fp32</code>, <code>fp16</code>.</td>
<td><code>str</code></td>
<td><code>"fp32"</code></td>
</tr>
<tr>
<td><code>enable_mkldnn</code></td>
<td>Whether to enable MKL-DNN acceleration for inference. If MKL-DNN is unavailable or the model does not support it, acceleration will not be used even if this flag is set.</td>
<td><code>bool</code></td>
<td><code>True</code></td>
</tr>
<tr>
<td><code>mkldnn_cache_capacity</code></td>
<td>MKL-DNN cache capacity.</td>
<td><code>int</code></td>
<td><code>10</code></td>
</tr>
<tr>
<td><code>cpu_threads</code></td>
<td>Number of threads to use for inference on CPUs.</td>
<td><code>int</code></td>
<td><code>10</code></td>
</tr>
<tr>
<td><code>input_shape</code></td>
<td>Input image size for the model in the format <code>(C, H, W)</code>.</td>
<td><code>tuple|None</code></td>
<td><code>None</code></td>
</tr>
</tbody>
</table>

* Call the `predict()` method of the text recognition model for inference. This method returns a list of results. In addition, this module also provides the `predict_iter()` method. The two methods are completely consistent in terms of parameter acceptance and result return. The difference is that `predict_iter()` returns a `generator`, which can process and obtain prediction results step by step. It is suitable for scenarios where large datasets need to be processed or memory savings are desired. You can choose either of these two methods according to your actual needs. The parameters of the `predict()` method include `input` and `batch_size`, with specific descriptions as follows:

<table>
<thead>
<tr>
<th>Parameter</th>
<th>Description</th>
<th>Type</th>
<th>Default</th>
</tr>
</thead>
<tr>
<td><code>input</code></td>
<td>Data to be predicted, supporting multiple input types, required.
<ul>
<li><b>Python Var</b>: Image data represented by <code>numpy.ndarray</code></li>
<li><b>str</b>: Local path of image file or PDF file: <code>/root/data/img.jpg</code>; <b>URL link</b>: Network URL of image file or PDF file: <a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_ocr_rec_001.png">Example</a>; <b>Local directory</b>: The directory should contain the images to be predicted, such as <code>/root/data/</code> (currently, prediction of PDF files in the directory is not supported, PDF files need to be specified to a specific file path)</li>
<li><b>list</b>: The elements of the list should be data of the above types, such as <code>[numpy.ndarray, numpy.ndarray]</code>, <code>["/root/data/img1.jpg", "/root/data/img2.jpg"]</code>, <code>["/root/data1", "/root/data2"]</code></li>
</ul>
</td>
<td><code>Python Var|str|list</code></td>
<td></td>
</tr>
<tr>
<td><code>batch_size</code></td>
<td>Batch size, can be set to any positive integer.</td>
<td><code>int</code></td>
<td>1</td>
</tr>
</table>

* Process the prediction results. The prediction result for each sample is a corresponding Result object, which supports operations such as printing, saving as an image, and saving as a `json` file:

<table>
<thead>
<tr>
<th>Method</th>
<th>Description</th>
<th>Parameter</th>
<th>Type</th>
<th>Description</th>
<th>Default</th>
</tr>
</thead>
<tr>
<td rowspan="3"><code>print()</code></td>
<td rowspan="3">Print the result to the terminal</td>
<td><code>format_json</code></td>
<td><code>bool</code></td>
<td>Whether to format the output content using <code>JSON</code> indentation</td>
<td><code>True</code></td>
</tr>
<tr>
<td><code>indent</code></td>
<td><code>int</code></td>
<td>Specifies the indentation level to beautify the output <code>JSON</code> data, making it more readable. Only effective when <code>format_json</code> is <code>True</code>.</td>
<td>4</td>
</tr>
<tr>
<td><code>ensure_ascii</code></td>
<td><code>bool</code></td>
<td>Controls whether to escape non-<code>ASCII</code> characters as <code>Unicode</code>. When set to <code>True</code>, all non-<code>ASCII</code> characters will be escaped; <code>False</code> retains the original characters. Only effective when <code>format_json</code> is <code>True</code>.</td>
<td><code>False</code></td>
</tr>
<tr>
<td rowspan="3"><code>save_to_json()</code></td>
<td rowspan="3">Save the result as a file in <code>json</code> format</td>
<td><code>save_path</code></td>
<td><code>str</code></td>
<td>The file path to save the result. When it is a directory, the saved file name is consistent with the naming of the input file type.</td>
<td>None</td>
</tr>
<tr>
<td><code>indent</code></td>
<td><code>int</code></td>
<td>Specifies the indentation level to beautify the output <code>JSON</code> data, making it more readable. Only effective when <code>format_json</code> is <code>True</code>.</td>
<td>4</td>
</tr>
<tr>
<td><code>ensure_ascii</code></td>
<td><code>bool</code></td>
<td>Controls whether to escape non-<code>ASCII</code> characters as <code>Unicode</code>. When set to <code>True</code>, all non-<code>ASCII</code> characters will be escaped; <code>False</code> retains the original characters. Only effective when <code>format_json</code> is <code>True</code>.</td>
<td><code>False</code></td>
</tr>
<tr>
<td><code>save_to_img()</code></td>
<td>Save the result as a file in image format</td>
<td><code>save_path</code></td>
<td><code>str</code></td>
<td>The file path to save the result. When it is a directory, the saved file name is consistent with the naming of the input file type.</td>
<td>None</td>
</tr>
</table>

* In addition, it also supports obtaining the visualized image with results and the prediction results through attributes, as follows:

<table>
<thead>
<tr>
<th>Attribute</th>
<th>Description</th>
</tr>
</thead>
<tr>
<td rowspan="1"><code>json</code></td>
<td rowspan="1">Obtain the prediction result in <code>json</code> format</td>
</tr>
<tr>
<td rowspan="1"><code>img</code></td>
<td rowspan="1">Obtain the visualized image in <code>dict</code> format</td>
</tr>
</table>

## 4. Secondary Development

If the above models do not perform well in your scenario, you can try the following steps for secondary development. Here, we take training `PP-OCRv5_server_rec` as an example. For other models, just replace the corresponding configuration file. First, you need to prepare a dataset for text recognition. You can refer to the format of the [Text Recognition Demo Data](https://paddle-model-ecology.bj.bcebos.com/paddlex/data/ocr_rec_dataset_examples.tar) for preparation. After preparation, you can train and export the model as follows. After export, the model can be quickly integrated into the above API. This example uses the Text Recognition Demo Data. Before training the model, please make sure you have installed the dependencies required by PaddleOCR as described in the [Installation Guide](../installation.md).

### 4.1 Dataset and Pre-trained Model Preparation

#### 4.1.1 Prepare the Dataset

```shell
# Download the example dataset
wget https://paddle-model-ecology.bj.bcebos.com/paddlex/data/ocr_rec_dataset_examples.tar
tar -xf ocr_rec_dataset_examples.tar
```

#### 4.1.2 Download the Pre-trained Model

```shell
# Download the PP-OCRv5_server_rec pre-trained model
wget https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-OCRv5_server_rec_pretrained.pdparams
```

### 4.2 Model Training

PaddleOCR modularizes its code. To train the `PP-OCRv5_server_rec` recognition model, you need to use its [configuration file](https://github.com/PaddlePaddle/PaddleOCR/blob/main/configs/rec/PP-OCRv5/PP-OCRv5_server_rec.yml).

The training commands are as follows:

```bash
# Single-GPU training (default training method)
python3 tools/train.py -c configs/rec/PP-OCRv5/PP-OCRv5_server_rec.yml \
   -o Global.pretrained_model=./PP-OCRv5_server_rec_pretrained.pdparams

# Multi-GPU training, specify GPU IDs via the --gpus parameter
python3 -m paddle.distributed.launch --gpus '0,1,2,3'  tools/train.py -c configs/rec/PP-OCRv5/PP-OCRv5_server_rec.yml \
        -o Global.pretrained_model=./PP-OCRv5_server_rec_pretrained.pdparams
```

### 4.3 Model Evaluation

You can evaluate the trained weights, such as `output/xxx/xxx.pdparams`, using the following command:

```bash
# Note: Set the path of pretrained_model to a local path. If you use a model you trained and saved yourself, please modify the path and file name to {path/to/weights}/{model_name}.
# Demo test set evaluation
python3 tools/eval.py -c configs/rec/PP-OCRv5/PP-OCRv5_server_rec.yml -o \
Global.pretrained_model=output/xxx/xxx.pdparams
```

### 4.4 Model Export

```bash
python3 tools/export_model.py -c configs/rec/PP-OCRv5/PP-OCRv5_server_rec.yml -o \
Global.pretrained_model=output/xxx/xxx.pdparams \
Global.save_inference_dir="./PP-OCRv5_server_rec_infer/"
```

After exporting the model, the static graph model will be stored in `./PP-OCRv5_server_rec_infer/` in the current directory. In this directory, you will see the following files:
```
./PP-OCRv5_server_rec_infer/
├── inference.json
├── inference.pdiparams
├── inference.yml
```
At this point, the secondary development is complete. This static graph model can be directly integrated into the PaddleOCR API.