Which Language RapidOCR it Support?

[ssavi-ict]

Hello RapidOCR,
I have read the documentation in English. As most of the docs are Chinese (even the linked documentation also), I was not able to locate in which language RapidOCR gives its support.
Can anyone please let me know about it?

Just to mention, I am working on PaddleOCR for a while now. I am opting out to train my own model in PaddleOCR. However, the guideline seems a bit less. I found this repository since RapidOCR also uses PaddleOCR along with the other two engines.

Thanks.

[SWHL]

• RapidOCR can recognize languages that are consistent with PaddleOCR. For details, please refer to here.
• If you want to use RapidOCR inference, you need to convert the model to ONNX format by PaddleOCRModelConverter

[ssavi-ict]

OK. So it supports those languages that PaddleOCR supports. Am I right?
Can you please tell me how can train my own language in PaddleOCR or is it possible to use RapidOCR to train my own language model and use it into PaddleOCR?

[SWHL]

• You are right. But there is a misunderstand.
• RapidOCR is an inference package and cannot train the model. PaddleOCR can train own model.
• The relationship between the two is shown in the following figure：

  flowchart LR
      subgraph Step
      direction TB
      C(Text Det) --> D(Text Cls) --> E(Text Rec)
      end
  
      A[/OurSelf Dataset/] --> B(PaddleOCR) --Train--> Step --> F(PaddleOCRModelConverter)
      F --ONNX--> G{RapidOCR Deploy\n<b>Python/C++/Java/C#</b>}
      G --> H(Windows x86/x64) & I(Linux) & J(Android) & K(Web) & L(Raspberry Pi)
  
      click B "https://github.com/PaddlePaddle/PaddleOCR" _blank
      click F "https://github.com/RapidAI/PaddleOCRModelConverter" _blank
  

  Loading
• You can try you own model with PaddleOCR, please refer this link for the detail.

[ssavi-ict]

Ahh.. OK. Thank you for the information..

[MadMenHitBooker]

Hi, for the conversion i tried with the model en_PP-OCRv4_rec_infer with the dict latin_dict.txt
The inference worked but there was obviously error with the dict (latin_dict was the wrong one):

['13;>?G<5475=5>DC', 'GU«iag]T«]Y[U«fa«fXQ`[«Q]]«fXU«TUeYW`Ude«Q`T', 'Sa`f`YRgfade«iXa«XQhU«RUU`«Y`ha]hUT«Y`«fXU', 'bdaTgSfYa`«aV«fXYe«Raa[+fXUYd«Sa`fdYRgfYa`e', 

But then i used the en_dict, which should be the right one

>>> paddle_result, elapse = rapid_ocr('./img_12.jpg')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.8/dist-packages/rapidocr_onnxruntime/rapid_ocr_api.py", line 91, in __call__
    rec_res, rec_elapse = self.text_recognizer(img_crop_list)
  File "/usr/local/lib/python3.8/dist-packages/rapidocr_onnxruntime/ch_ppocr_v3_rec/text_recognize.py", line 71, in __call__
    rec_result = self.postprocess_op(preds)
  File "/usr/local/lib/python3.8/dist-packages/rapidocr_onnxruntime/ch_ppocr_v3_rec/utils.py", line 36, in __call__
    text = self.decode(preds_idx, preds_prob, is_remove_duplicate=True)
  File "/usr/local/lib/python3.8/dist-packages/rapidocr_onnxruntime/ch_ppocr_v3_rec/utils.py", line 68, in decode
    char_list.append(self.character[int(text_index[batch_idx][idx])])

Do you know what could be the error ?

[SWHL]

You can try to get the converted onnx model by PaddleOCRModelConvert tool, which can automatically include the dict.