Png 형식 파일 OCR 추출 질문

안녕하세요 혼자 독학을 하고 있고 따로 UiPath 관련 질문을 해결할 방법이 없어서 이렇게 여쭤드립니다.

현재 png 형식 이미지 파일에서 OCR 기능으로 텍스트 데이터를 추출하려 하고 있습니다.

제가 시도해본 방법은 다음과 같습니다.

첫 번째 방법은 “UiPath Document OCR” 액티비티를 사용해봤는데 계속해서 에러가 발생하고 있습니다. (에러 내용이 너무 길어서 첫 줄 내용만 올려드립니다. “UiPath Document OCR: Request CorrelationId: 11a2448b-9dd2-4bdc-afae-d23408a166f5”)

두 번째 방법은 여기저기 찾아보다가 “Microsoft OCR” 액티비티를 사용하는 샘플을 발견해서 제가 사용하는 프로세스에도 적용하려 했는데 “Microsoft OCR” 액티비티 자체가 없더라구요.
분명히 패키지, 패키지 버전도 똑같이 환경 세팅을 하였는데 왜 저는 “Microsoft OCR” 액티비티가 존재하지 않는지 여쭙고 싶습니다.

혹시 다른 분들께서는 이미지 파일에서 OCR 액티비티로 텍스트 데이터를 추출할때 어떤 방법을 사용하시는지도 궁금해서 이렇게 여쭤드립니다.

긴 글 읽어주셔서 감사합니다.!

Hi @dongbinr56 ,
To get text in image PNG, you can use OCR - Japanese, Chinese, Korean
with korean character,
I think it will correct, I tried it with japan text is ok

image

regards,
LNV

2 Likes

Hi @dongbinr56

=> Use Load image acitvity to load the image and store the output in a variable.
=> Use Tesseract OCR Engine and pass the variable. Stroe the output in text.
=> Use Write line or Message box to print the content.

Hope it helps!!

2 Likes

@dongbinr56
If you find solution for your query please mark it as solution to close the loop.

Regards

1 Like

First of all, thank you so much for telling me how to fix it. But there is another problem.

That’s the “endpoint” part, and I don’t understand what the “endpoint” part does.

I’ve filled in the API key, image, and text settings fields, but the “endpoints” part keeps getting into trouble.

I would like to take this opportunity to ask you because I want to make sure that I know what I don’t know.

Thank you!

Thank you so much for coming up with a solution!

Hi @dongbinr56
OCR- Japanese, Chinese, Korean activity rather than the normal UiPath Document OCR and use this endpoint.
https://du.uipath.com/cjk-ocr

have many ways to read text in image PNG, can you share your image, I will test for you

regards,

1 Like

@dongbinr56
Tesseract OCR doesn’t need an end point. You can directly use that.

Regards

1 Like

You can try
my input
donbinr56
my code
image.xaml (7.7 KB)
I tried 2 ways with Tesseract OCR and KCJ OCR
if your image have korean character, I think KCJ will ok

my output
image
you can try

regards,
LNV

1 Like
  1. Here’s the image file I’m testing right now
    OCR_사용절차
  2. I got the same error again. The error details are as follows
    Uipath Error text.txt (66.7 KB)
    If you try to upload the error content in text format, an error will occur, so we will write it in a notepad and upload it

In here you use document OCR ?
document OCR use generate API key of UiPath
It cannot read correct korean character from image
I tried it
my image
donbinr56
my code


my output
image
it is not ok
you should use KCJ ocr
regards,
LNV

1 Like

my test result with your image

1 Like

I’ve uploaded another error file, :cry:
I’ll upload it again
Uipath Error text.txt (66.8 KB)

error is "UiPath Document OCR: Request CorrelationId: 11a2448b-9dd2-4bdc-afae-d23408a166f5 "?

It’s necessary to use CimputerVision API key from AutomationCloud - Admin - License page as the following image

can you share your file you need read ?

1 Like

Even if I change it to the “CJK-OCR” activity and change the API key again, I still get an error…

The error message that occurred this time is as follows
Uipath Error text.txt (66.8 KB)

can you share a picture of your error?
you send the text file, I need to download it, very inconvenient :slight_smile:

can you share your workflow, your image what you need read, your input, your expect output?

have you tried my .xaml file with your input?
regards,
LNV

1 Like

It seems that it will be difficult to upload documents and files due to security issues :cry:

In the case of error messages, I also tried to upload them in image format at first, but the content of the error message is really too long, so I can’t upload it in image format.

The problem has not been solved, but I am so grateful to you for helping me to do this

my code run in your PC have same error ?

Yes, I tried, but I got the same error again

KCJ have API key
“HESNfqLJMUjrCvAupBOT8e6yocmqY1IWgH2RUANyLzLuKNb7kV23Mdz0KzzKddDq09tQAo/CZ1iYUOIAgnf42Q==”
and end point
https://du-jp.uipath.com/cjk-ocr


I tried with
ex
donbinr56
test
I have correct output with all of them
have you changed CimputerVision API key from AutomationCloud - Admin - License ?
this problem is small in UiPath, you can try again