add en,tth,eel models (#4439)

docs/module_usage/tutorials/ocr_modules/text_recognition.en.md

@@ -220,6 +220,16 @@ SVTRv2 is a server text recognition model developed by the OpenOCR team of Fudan
 <th>Introduction</th>
 </tr>
 <tr>
+<td>en_PP-OCRv5_mobile_rec</td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
+en_PP-OCRv5_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/en_PP-OCRv5_mobile_rec_pretrained.pdparams">Training Model</a></td>
+<td> 85.25</td>
+<td>-</td>
+<td>-</td>
+<td>7.5</td>
+<td>The ultra-lightweight English recognition model trained based on the PP-OCRv5 recognition model supports the recognition of English and numbers.</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="">Training Model</a></td>
 <td> 70.39</td>
@@ -280,6 +290,26 @@ eslav_PP-OCRv5_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle
 <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>th_PP-OCRv5_mobile_rec</td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
+th_PP-OCRv5_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/th_PP-OCRv5_mobile_rec_pretrained.pdparams">Training Model</a></td>
+<td>82.68</td>
+<td>-</td>
+<td>-</td>
+<td>7.5</td>
+<td>The Thai recognition model trained based on the PP-OCRv5 recognition model supports recognition of Thai, English, and numbers.</td>
+</tr>
+<tr>
+<td>el_PP-OCRv5_mobile_rec</td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
+el_PP-OCRv5_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/el_PP-OCRv5_mobile_rec_pretrained.pdparams">Training Model</a></td>
+<td>89.28</td>
+<td>-</td>
+<td>-</td>
+<td>7.5</td>
+<td>The Greek recognition model trained based on the PP-OCRv5 recognition model supports recognition of Greek, English, and numbers.</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="">Training Model</a></td>
 <td>60.21</td>

docs/module_usage/tutorials/ocr_modules/text_recognition.md

@@ -227,6 +227,16 @@ SVTRv2 是一种由复旦大学视觉与学习实验室（FVL）的OpenOCR团队
 <th>介绍</th>
 </tr>
 <tr>
+<td>en_PP-OCRv5_mobile_rec</td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
+en_PP-OCRv5_mobile_rec_infer.tar">推理模型</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/en_PP-OCRv5_mobile_rec_pretrained.pdparams">训练模型</a></td>
+<td> 85.25</td>
+<td>-</td>
+<td>-</td>
+<td>7.5</td>
+<td>基于PP-OCRv5识别模型训练得到的超轻量英文识别模型，支持英文、数字识别</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">推理模型</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/en_PP-OCRv4_mobile_rec_pretrained.pdparams">训练模型</a></td>
@@ -289,6 +299,25 @@ eslav_PP-OCRv5_mobile_rec_infer.tar">推理模型</a>/<a href="https://paddle-mo
 <td>14</td>
 <td>基于PP-OCRv5识别模型训练得到的东斯拉夫语言识别模型， 支持东斯拉夫语言、英文和数字识别</td>
 </tr>
+<td>th_PP-OCRv5_mobile_rec</td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
+th_PP-OCRv5_mobile_rec_infer.tar">推理模型</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/th_PP-OCRv5_mobile_rec_pretrained.pdparams">训练模型</a></td>
+<td>82.68</td>
+<td>-</td>
+<td>-</td>
+<td>7.5</td>
+<td>基于PP-OCRv5识别模型训练得到的泰语识别模型， 支持泰语、英文和数字识别</td>
+</tr>
+<tr>
+<td>el_PP-OCRv5_mobile_rec</td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
+el_PP-OCRv5_mobile_rec_infer.tar">推理模型</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/el_PP-OCRv5_mobile_rec_pretrained.pdparams">训练模型</a></td>
+<td>89.28</td>
+<td>-</td>
+<td>-</td>
+<td>7.5</td>
+<td>基于PP-OCRv5识别模型训练得到的希腊语识别模型， 支持希腊语、英文和数字识别</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/\

docs/pipeline_usage/tutorials/ocr_pipelines/OCR.en.md

@@ -379,6 +379,16 @@ SVTRv2 is a server text recognition model developed by the OpenOCR team of Fudan
 <th>Introduction</th>
 </tr>
 <tr>
+<td>en_PP-OCRv5_mobile_rec</td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
+en_PP-OCRv5_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/en_PP-OCRv5_mobile_rec_pretrained.pdparams">Training Model</a></td>
+<td> 85.25</td>
+<td>-</td>
+<td>-</td>
+<td>7.5</td>
+<td>The ultra-lightweight English recognition model trained based on the PP-OCRv5 recognition model supports the recognition of English and numbers.</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="">Training Model</a></td>
 <td> 70.39</td>
@@ -439,6 +449,26 @@ eslav_PP-OCRv5_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle
 <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>th_PP-OCRv5_mobile_rec</td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
+th_PP-OCRv5_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/th_PP-OCRv5_mobile_rec_pretrained.pdparams">Training Model</a></td>
+<td>82.68</td>
+<td>-</td>
+<td>-</td>
+<td>7.5</td>
+<td>The Thai recognition model trained based on the PP-OCRv5 recognition model supports recognition of Thai, English, and numbers.</td>
+</tr>
+<tr>
+<td>el_PP-OCRv5_mobile_rec</td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
+el_PP-OCRv5_mobile_rec_infer.tar">Inference Model</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/el_PP-OCRv5_mobile_rec_pretrained.pdparams">Training Model</a></td>
+<td>89.28</td>
+<td>-</td>
+<td>-</td>
+<td>7.5</td>
+<td>The Greek recognition model trained based on the PP-OCRv5 recognition model supports recognition of Greek, English, and numbers.</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="">Training Model</a></td>
 <td>60.21</td>
@@ -948,6 +978,18 @@ In the above Python script, the following steps are executed:
 <td>The file path for saving, supporting both directory and file paths</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>return_word_box</code></td>
+<td>Whether to return the position coordinates of each character</td>
+<td><code>bool|None</code></td>
+<td>
+<ul>
+<li><b>bool</b>：<code>True</code> 或者 <code>False</code>；</li>
+<li><b>None</b>：If set to<code>None</code>, the default value initialized by the pipeline will be used, which is initialized as<code>False</code>；</li>
+</ul>
+</td>
+<td><code>None</code></td>
+</tr>
 </table>
 
 - Calling the `print()` method will print the result to the terminal. The printed content is explained as follows:
@@ -992,6 +1034,10 @@ In the above Python script, the following steps are executed:
 
     - `rec_boxes`: `(numpy.ndarray)` An array of rectangular bounding boxes for detection boxes, with a shape of (n, 4) and dtype int16. Each row represents the [x_min, y_min, x_max, y_max] coordinates of a rectangle, where (x_min, y_min) is the top-left corner and (x_max, y_max) is the bottom-right corner
 
+    - `text_word`: `(List[str])` When `return_word_box` is set to `True`, returns a list of the recognized text for each character.
+
+    - `text_word_boxes`: `(List[numpy.ndarray])` When `return_word_box` is set to `True`, returns a list of bounding box coordinates for each recognized character.
+
 - Calling the `save_to_json()` method will save the above content to the specified `save_path`. If a directory is specified, the saved path will be `save_path/{your_img_basename}_res.json`. If a file is specified, it will be saved directly to that file. Since JSON files do not support saving numpy arrays, the `numpy.array` type will be converted to a list format.
 
 - Calling the `save_to_img()` method will save the visualization results to the specified `save_path`. If a directory is specified, the saved path will be `save_path/{your_img_basename}_ocr_res_img.{your_img_extension}`. If a file is specified, it will be saved directly to that file. (Since the pipeline usually contains multiple result images, it is not recommended to specify a specific file path directly, as multiple images will be overwritten and only the last image will be retained)

docs/pipeline_usage/tutorials/ocr_pipelines/OCR.md

@@ -354,6 +354,16 @@ SVTRv2 是一种由复旦大学视觉与学习实验室（FVL）的OpenOCR团队
 <th>介绍</th>
 </tr>
 <tr>
+<td>en_PP-OCRv5_mobile_rec</td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
+en_PP-OCRv5_mobile_rec_infer.tar">推理模型</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/en_PP-OCRv5_mobile_rec_pretrained.pdparams">训练模型</a></td>
+<td> 85.25</td>
+<td>-</td>
+<td>-</td>
+<td>7.5</td>
+<td>基于PP-OCRv5识别模型训练得到的超轻量英文识别模型，支持英文、数字识别</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">推理模型</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/en_PP-OCRv4_mobile_rec_pretrained.pdparams">训练模型</a></td>
@@ -416,6 +426,26 @@ eslav_PP-OCRv5_mobile_rec_infer.tar">推理模型</a>/<a href="https://paddle-mo
 <td>基于PP-OCRv5识别模型训练得到的东斯拉夫语言识别模型， 支持东斯拉夫语言、英文和数字识别</td>
 </tr>
 <tr>
+<td>th_PP-OCRv5_mobile_rec</td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
+th_PP-OCRv5_mobile_rec_infer.tar">推理模型</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/th_PP-OCRv5_mobile_rec_pretrained.pdparams">训练模型</a></td>
+<td>82.68</td>
+<td>-</td>
+<td>-</td>
+<td>7.5</td>
+<td>基于PP-OCRv5识别模型训练得到的泰语识别模型， 支持泰语、英文和数字识别</td>
+</tr>
+<tr>
+<td>el_PP-OCRv5_mobile_rec</td>
+<td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/\
+el_PP-OCRv5_mobile_rec_infer.tar">推理模型</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/el_PP-OCRv5_mobile_rec_pretrained.pdparams">训练模型</a></td>
+<td>89.28</td>
+<td>-</td>
+<td>-</td>
+<td>7.5</td>
+<td>基于PP-OCRv5识别模型训练得到的希腊语识别模型， 支持希腊语、英文和数字识别</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">推理模型</a>/<a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/korean_PP-OCRv3_mobile_rec_pretrained.pdparams">训练模型</a></td>
@@ -924,6 +954,18 @@ for res in output:
 <td>保存的文件路径，支持目录或文件路径</td>
 <td>无</td>
 </tr>
+<tr>
+<td><code>return_word_box</code></td>
+<td>是否返回每个文字的位置坐标</td>
+<td><code>bool|None</code></td>
+<td>
+<ul>
+<li><b>bool</b>：<code>True</code> 或者 <code>False</code>；</li>
+<li><b>None</b>：如果设置为<code>None</code>, 将默认使用产线初始化的该参数值，初始化为<code>False</code>；</li>
+</ul>
+</td>
+<td><code>None</code></td>
+</tr>
 </table>
 
 - 调用`print()` 方法会将结果打印到终端，打印到终端的内容解释如下：
@@ -969,6 +1011,10 @@ for res in output:
     - `rec_boxes`: `(numpy.ndarray)` 检测框的矩形边界框数组，shape为(n, 4)，dtype为int16。每一行表示一个矩形框的[x_min, y_min, x_max, y_max]坐标
     ，其中(x_min, y_min)为左上角坐标，(x_max, y_max)为右下角坐标
 
+    - `text_word`: `(List[str])` 当 `return_word_box` 设置为 `True` 时，返回每个识别文字的文本列表。
+
+    - `text_word_boxes`: `(List[numpy.ndarray])` 当 `return_word_box` 设置为 `True` 时，返回每个识别文字的边界框坐标列表。
+
 - 调用`save_to_json()` 方法会将上述内容保存到指定的`save_path`中，如果指定为目录，则保存的路径为`save_path/{your_img_basename}_res.json`，如果指定为文件，则直接保存到该文件中。由于json文件不支持保存numpy数组，因此会将其中的`numpy.array`类型转换为列表形式。
 - 调用`save_to_img()` 方法会将可视化结果保存到指定的`save_path`中，如果指定为目录，则保存的路径为`save_path/{your_img_basename}_ocr_res_img.{your_img_extension}`，如果指定为文件，则直接保存到该文件中。(产线通常包含较多结果图片，不建议直接指定为具体的文件路径，否则多张图会被覆盖，仅保留最后一张图)
 

paddlex/configs/modules/text_recognition/el_PP-OCRv5_mobile_rec.yaml

@@ -0,0 +1,39 @@
+Global:
+  model: el_PP-OCRv5_mobile_rec
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/ocr_rec/ocr_rec_dataset_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert: 
+    enable: False
+    src_dataset_type: null
+  split: 
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  epochs_iters: 20
+  batch_size: 8
+  learning_rate: 0.001
+  pretrain_weight_path: https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/el_PP-OCRv5_mobile_rec_pretrained.pdparams
+  resume_path: null
+  log_interval: 20
+  eval_interval: 1
+  save_interval: 1
+
+Evaluate:
+  weight_path: "output/best_accuracy/best_accuracy.pdparams"
+  log_interval: 1
+
+Export:
+  weight_path: https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/el_PP-OCRv5_mobile_rec_pretrained.pdparams
+
+Predict:
+  batch_size: 1
+  model_dir: "output/best_accuracy/inference"
+  input: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_ocr_rec_el.png"
+  kernel_option:
+    run_mode: paddle

paddlex/configs/modules/text_recognition/en_PP-OCRv5_mobile_rec.yaml

@@ -0,0 +1,39 @@
+Global:
+  model: en_PP-OCRv5_mobile_rec
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/ocr_rec/ocr_rec_dataset_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert: 
+    enable: False
+    src_dataset_type: null
+  split: 
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  epochs_iters: 20
+  batch_size: 8
+  learning_rate: 0.001
+  pretrain_weight_path: https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/en_PP-OCRv5_mobile_rec_pretrained.pdparams
+  resume_path: null
+  log_interval: 20
+  eval_interval: 1
+  save_interval: 1
+
+Evaluate:
+  weight_path: "output/best_accuracy/best_accuracy.pdparams"
+  log_interval: 1
+
+Export:
+  weight_path: https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/en_PP-OCRv5_mobile_rec_pretrained.pdparams
+
+Predict:
+  batch_size: 1
+  model_dir: "output/best_accuracy/inference"
+  input: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_ocr_rec_en.png"
+  kernel_option:
+    run_mode: paddle

paddlex/configs/modules/text_recognition/th_PP-OCRv5_mobile_rec.yaml

@@ -0,0 +1,39 @@
+Global:
+  model: th_PP-OCRv5_mobile_rec
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/ocr_rec/ocr_rec_dataset_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert: 
+    enable: False
+    src_dataset_type: null
+  split: 
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  epochs_iters: 20
+  batch_size: 8
+  learning_rate: 0.001
+  pretrain_weight_path: https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/th_PP-OCRv5_mobile_rec_pretrained.pdparams
+  resume_path: null
+  log_interval: 20
+  eval_interval: 1
+  save_interval: 1
+
+Evaluate:
+  weight_path: "output/best_accuracy/best_accuracy.pdparams"
+  log_interval: 1
+
+Export:
+  weight_path: https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/th_PP-OCRv5_mobile_rec_pretrained.pdparams
+
+Predict:
+  batch_size: 1
+  model_dir: "output/best_accuracy/inference"
+  input: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_ocr_rec_th.png"
+  kernel_option:
+    run_mode: paddle

paddlex/inference/models/text_recognition/predictor.py

@@ -19,12 +19,14 @@ from ....utils.fonts import (
     ARABIC_FONT,
     CYRILLIC_FONT,
     DEVANAGARI_FONT,
+    EL_FONT,
     KANNADA_FONT,
     KOREAN_FONT,
     LATIN_FONT,
     SIMFANG_FONT,
     TAMIL_FONT,
     TELUGU_FONT,
+    TH_FONT,
 )
 from ....utils.func_register import FuncRegister
 from ...common.batch_sampler import ImageBatchSampler
@@ -160,6 +162,12 @@ class TextRecPredictor(BasePredictor):
         ):
             return KOREAN_FONT
 
+        if self.model_name == "th_PP-OCRv5_mobile_rec":
+            return TH_FONT
+
+        if self.model_name == "el_PP-OCRv5_mobile_rec":
+            return EL_FONT
+
         if self.model_name == "arabic_PP-OCRv3_mobile_rec":
             return ARABIC_FONT
 

paddlex/inference/utils/official_models.py

@@ -312,6 +312,9 @@ ALL_MODELS = [
     "PP-DocBee2-3B",
     "latin_PP-OCRv5_mobile_rec",
     "korean_PP-OCRv5_mobile_rec",
+    "th_PP-OCRv5_mobile_rec",
+    "el_PP-OCRv5_mobile_rec",
+    "en_PP-OCRv5_mobile_rec",
 ]
 
 
@@ -332,6 +335,9 @@ OCR_MODELS = [
     "LaTeX_OCR_rec",
     "latin_PP-OCRv3_mobile_rec",
     "latin_PP-OCRv5_mobile_rec",
+    "en_PP-OCRv5_mobile_rec",
+    "th_PP-OCRv5_mobile_rec",
+    "el_PP-OCRv5_mobile_rec",
     "PicoDet_layout_1x",
     "PicoDet_layout_1x_table",
     "PicoDet-L_layout_17cls",

paddlex/modules/text_recognition/model_list.py

@@ -36,4 +36,7 @@ MODELS = [
     "latin_PP-OCRv5_mobile_rec",
     "eslav_PP-OCRv5_mobile_rec",
     "korean_PP-OCRv5_mobile_rec",
+    "en_PP-OCRv5_mobile_rec",
+    "el_PP-OCRv5_mobile_rec",
+    "th_PP-OCRv5_mobile_rec",
 ]

paddlex/repo_apis/PaddleOCR_api/configs/el_PP-OCRv5_mobile_rec.yaml

@@ -0,0 +1,141 @@
+Global:
+  model_name: el_PP-OCRv5_mobile_rec # To use static model for inference.
+  debug: false
+  use_gpu: true
+  epoch_num: 75
+  log_smooth_window: 20
+  print_batch_step: 10
+  save_model_dir: ./output/el_rec_ppocr_v5
+  save_epoch_step: 10
+  eval_batch_step: [0, 1000]
+  cal_metric_during_train: true
+  pretrained_model:
+  checkpoints:
+  save_inference_dir:
+  use_visualdl: false
+  infer_img:
+  character_dict_path: ./ppocr/utils/dict/ppocrv5_el_dict.txt
+  max_text_length: &max_text_length 25
+  infer_mode: false
+  use_space_char: true
+  distributed: true
+  save_res_path: ./output/rec/predicts_el_ppocrv5.txt
+  d2s_train_image_shape: [3, 48, 320]
+
+
+Optimizer:
+  name: Adam
+  beta1: 0.9
+  beta2: 0.999
+  lr:
+    name: Cosine
+    learning_rate: 0.0005
+    warmup_epoch: 5
+  regularizer:
+    name: L2
+    factor: 3.0e-05
+
+
+Architecture:
+  model_type: rec
+  algorithm: SVTR_LCNet
+  Transform:
+  Backbone:
+    name: PPLCNetV3
+    scale: 0.95
+  Head:
+    name: MultiHead
+    head_list:
+      - CTCHead:
+          Neck:
+            name: svtr
+            dims: 120
+            depth: 2
+            hidden_dims: 120
+            kernel_size: [1, 3]
+            use_guide: True
+          Head:
+            fc_decay: 0.00001
+      - NRTRHead:
+          nrtr_dim: 384
+          max_text_length: *max_text_length
+
+Loss:
+  name: MultiLoss
+  loss_config_list:
+    - CTCLoss:
+    - NRTRLoss:
+
+PostProcess:  
+  name: CTCLabelDecode
+
+Metric:
+  name: RecMetric
+  main_indicator: acc
+  ignore_space: False
+
+Train:
+  dataset:
+    name: MultiScaleDataSet
+    ds_width: false
+    data_dir: ./train_data/
+    ext_op_transform_idx: 1
+    label_file_list:
+    - ./train_data/train_list.txt
+    transforms:
+    - DecodeImage:
+        img_mode: BGR
+        channel_first: false
+    - RecConAug:
+        prob: 0.5
+        ext_data_num: 2
+        image_shape: [48, 320, 3]
+        max_text_length: *max_text_length
+    - RecAug:
+    - MultiLabelEncode:
+        gtc_encode: NRTRLabelEncode
+    - KeepKeys:
+        keep_keys:
+        - image
+        - label_ctc
+        - label_gtc
+        - length
+        - valid_ratio
+  sampler:
+    name: MultiScaleSampler
+    scales: [[320, 32], [320, 48], [320, 64]]
+    first_bs: &bs 128
+    fix_bs: false
+    divided_factor: [8, 16] # w, h
+    is_training: True
+  loader:
+    shuffle: true
+    batch_size_per_card: *bs
+    drop_last: true
+    num_workers: 8
+Eval:
+  dataset:
+    name: SimpleDataSet
+    data_dir: ./train_data/
+    label_file_list:
+    - ./train_data/val_list.txt
+    transforms:
+    - DecodeImage:
+        img_mode: BGR
+        channel_first: false
+    - MultiLabelEncode:
+        gtc_encode: NRTRLabelEncode
+    - RecResizeImg:
+        image_shape: [3, 48, 320]
+    - KeepKeys:
+        keep_keys:
+        - image
+        - label_ctc
+        - label_gtc
+        - length
+        - valid_ratio
+  loader:
+    shuffle: true
+    drop_last: false
+    batch_size_per_card: 128
+    num_workers: 4

paddlex/repo_apis/PaddleOCR_api/configs/en_PP-OCRv5_mobile_rec.yaml

@@ -0,0 +1,141 @@
+Global:
+  model_name: en_PP-OCRv5_mobile_rec # To use static model for inference.
+  debug: false
+  use_gpu: true
+  epoch_num: 75
+  log_smooth_window: 20
+  print_batch_step: 10
+  save_model_dir: ./output/en_rec_ppocr_v5
+  save_epoch_step: 10
+  eval_batch_step: [0, 1000]
+  cal_metric_during_train: true
+  pretrained_model:
+  checkpoints:
+  save_inference_dir:
+  use_visualdl: false
+  infer_img:
+  character_dict_path: ./ppocr/utils/dict/ppocrv5_en_dict.txt
+  max_text_length: &max_text_length 25
+  infer_mode: false
+  use_space_char: true
+  distributed: true
+  save_res_path: ./output/rec/predicts_en_ppocrv5.txt
+  d2s_train_image_shape: [3, 48, 320]
+
+
+Optimizer:
+  name: Adam
+  beta1: 0.9
+  beta2: 0.999
+  lr:
+    name: Cosine
+    learning_rate: 0.0005
+    warmup_epoch: 5
+  regularizer:
+    name: L2
+    factor: 3.0e-05
+
+
+Architecture:
+  model_type: rec
+  algorithm: SVTR_LCNet
+  Transform:
+  Backbone:
+    name: PPLCNetV3
+    scale: 0.95
+  Head:
+    name: MultiHead
+    head_list:
+      - CTCHead:
+          Neck:
+            name: svtr
+            dims: 120
+            depth: 2
+            hidden_dims: 120
+            kernel_size: [1, 3]
+            use_guide: True
+          Head:
+            fc_decay: 0.00001
+      - NRTRHead:
+          nrtr_dim: 384
+          max_text_length: *max_text_length
+
+Loss:
+  name: MultiLoss
+  loss_config_list:
+    - CTCLoss:
+    - NRTRLoss:
+
+PostProcess:  
+  name: CTCLabelDecode
+
+Metric:
+  name: RecMetric
+  main_indicator: acc
+  ignore_space: False
+
+Train:
+  dataset:
+    name: MultiScaleDataSet
+    ds_width: false
+    data_dir: ./train_data/
+    ext_op_transform_idx: 1
+    label_file_list:
+    - ./train_data/train_list.txt
+    transforms:
+    - DecodeImage:
+        img_mode: BGR
+        channel_first: false
+    - RecConAug:
+        prob: 0.5
+        ext_data_num: 2
+        image_shape: [48, 320, 3]
+        max_text_length: *max_text_length
+    - RecAug:
+    - MultiLabelEncode:
+        gtc_encode: NRTRLabelEncode
+    - KeepKeys:
+        keep_keys:
+        - image
+        - label_ctc
+        - label_gtc
+        - length
+        - valid_ratio
+  sampler:
+    name: MultiScaleSampler
+    scales: [[320, 32], [320, 48], [320, 64]]
+    first_bs: &bs 128
+    fix_bs: false
+    divided_factor: [8, 16] # w, h
+    is_training: True
+  loader:
+    shuffle: true
+    batch_size_per_card: *bs
+    drop_last: true
+    num_workers: 8
+Eval:
+  dataset:
+    name: SimpleDataSet
+    data_dir: ./train_data/
+    label_file_list:
+    - ./train_data/val_list.txt
+    transforms:
+    - DecodeImage:
+        img_mode: BGR
+        channel_first: false
+    - MultiLabelEncode:
+        gtc_encode: NRTRLabelEncode
+    - RecResizeImg:
+        image_shape: [3, 48, 320]
+    - KeepKeys:
+        keep_keys:
+        - image
+        - label_ctc
+        - label_gtc
+        - length
+        - valid_ratio
+  loader:
+    shuffle: true
+    drop_last: false
+    batch_size_per_card: 128
+    num_workers: 4

paddlex/repo_apis/PaddleOCR_api/configs/th_PP-OCRv5_mobile_rec.yaml

@@ -0,0 +1,141 @@
+Global:
+  model_name: th_PP-OCRv5_mobile_rec # To use static model for inference.
+  debug: false
+  use_gpu: true
+  epoch_num: 75
+  log_smooth_window: 20
+  print_batch_step: 10
+  save_model_dir: ./output/th_rec_ppocr_v5
+  save_epoch_step: 10
+  eval_batch_step: [0, 1000]
+  cal_metric_during_train: true
+  pretrained_model:
+  checkpoints:
+  save_inference_dir:
+  use_visualdl: false
+  infer_img:
+  character_dict_path: ./ppocr/utils/dict/ppocrv5_th_dict.txt
+  max_text_length: &max_text_length 25
+  infer_mode: false
+  use_space_char: true
+  distributed: true
+  save_res_path: ./output/rec/predicts_th_ppocrv5.txt
+  d2s_train_image_shape: [3, 48, 320]
+
+
+Optimizer:
+  name: Adam
+  beta1: 0.9
+  beta2: 0.999
+  lr:
+    name: Cosine
+    learning_rate: 0.0005
+    warmup_epoch: 5
+  regularizer:
+    name: L2
+    factor: 3.0e-05
+
+
+Architecture:
+  model_type: rec
+  algorithm: SVTR_LCNet
+  Transform:
+  Backbone:
+    name: PPLCNetV3
+    scale: 0.95
+  Head:
+    name: MultiHead
+    head_list:
+      - CTCHead:
+          Neck:
+            name: svtr
+            dims: 120
+            depth: 2
+            hidden_dims: 120
+            kernel_size: [1, 3]
+            use_guide: True
+          Head:
+            fc_decay: 0.00001
+      - NRTRHead:
+          nrtr_dim: 384
+          max_text_length: *max_text_length
+
+Loss:
+  name: MultiLoss
+  loss_config_list:
+    - CTCLoss:
+    - NRTRLoss:
+
+PostProcess:  
+  name: CTCLabelDecode
+
+Metric:
+  name: RecMetric
+  main_indicator: acc
+  ignore_space: False
+
+Train:
+  dataset:
+    name: MultiScaleDataSet
+    ds_width: false
+    data_dir: ./train_data/
+    ext_op_transform_idx: 1
+    label_file_list:
+    - ./train_data/train_list.txt
+    transforms:
+    - DecodeImage:
+        img_mode: BGR
+        channel_first: false
+    - RecConAug:
+        prob: 0.5
+        ext_data_num: 2
+        image_shape: [48, 320, 3]
+        max_text_length: *max_text_length
+    - RecAug:
+    - MultiLabelEncode:
+        gtc_encode: NRTRLabelEncode
+    - KeepKeys:
+        keep_keys:
+        - image
+        - label_ctc
+        - label_gtc
+        - length
+        - valid_ratio
+  sampler:
+    name: MultiScaleSampler
+    scales: [[320, 32], [320, 48], [320, 64]]
+    first_bs: &bs 128
+    fix_bs: false
+    divided_factor: [8, 16] # w, h
+    is_training: True
+  loader:
+    shuffle: true
+    batch_size_per_card: *bs
+    drop_last: true
+    num_workers: 8
+Eval:
+  dataset:
+    name: SimpleDataSet
+    data_dir: ./train_data/
+    label_file_list:
+    - ./train_data/val_list.txt
+    transforms:
+    - DecodeImage:
+        img_mode: BGR
+        channel_first: false
+    - MultiLabelEncode:
+        gtc_encode: NRTRLabelEncode
+    - RecResizeImg:
+        image_shape: [3, 48, 320]
+    - KeepKeys:
+        keep_keys:
+        - image
+        - label_ctc
+        - label_gtc
+        - length
+        - valid_ratio
+  loader:
+    shuffle: true
+    drop_last: false
+    batch_size_per_card: 128
+    num_workers: 4

paddlex/repo_apis/PaddleOCR_api/text_rec/register.py

@@ -241,3 +241,30 @@ register_model_info(
         "supported_apis": ["train", "evaluate", "predict", "export"],
     }
 )
+
+register_model_info(
+    {
+        "model_name": "en_PP-OCRv5_mobile_rec",
+        "suite": "TextRec",
+        "config_path": osp.join(PDX_CONFIG_DIR, "en_PP-OCRv5_mobile_rec.yaml"),
+        "supported_apis": ["train", "evaluate", "predict", "export"],
+    }
+)
+
+register_model_info(
+    {
+        "model_name": "th_PP-OCRv5_mobile_rec",
+        "suite": "TextRec",
+        "config_path": osp.join(PDX_CONFIG_DIR, "th_PP-OCRv5_mobile_rec.yaml"),
+        "supported_apis": ["train", "evaluate", "predict", "export"],
+    }
+)
+
+register_model_info(
+    {
+        "model_name": "el_PP-OCRv5_mobile_rec",
+        "suite": "TextRec",
+        "config_path": osp.join(PDX_CONFIG_DIR, "el_PP-OCRv5_mobile_rec.yaml"),
+        "supported_apis": ["train", "evaluate", "predict", "export"],
+    }
+)

paddlex/utils/fonts.py

@@ -108,6 +108,8 @@ if Path(str(LOCAL_FONT_FILE_PATH)).is_file():
 PINGFANG_FONT = Font(font_name="PingFang-SC-Regular.ttf")
 SIMFANG_FONT = Font(font_name="simfang.ttf")
 LATIN_FONT = Font(font_name="latin.ttf")
+TH_FONT = Font(font_name="th.ttf")
+EL_FONT = Font(font_name="el.ttf")
 KOREAN_FONT = Font(font_name="korean.ttf")
 ARABIC_FONT = Font(font_name="arabic.ttf")
 CYRILLIC_FONT = Font(font_name="cyrillic.ttf")