How to get Text from this Image?

image

#1

Hi Angels(found many angels in this Community),

I am trying to get Text (P@0020CD) from image attached.
Google OCR and Microsoft OCR have been tested but the most accurate result I could get is “P@OO2OCD”. (Correct one is number “0” instead of character “O”).

Can I make it with Uipath?
It would be really appciated if another angel can teach me with details

Many thanks,


#2

have you tried regular scraping, without using OCR?


#3

Hi @yangbo,
Try ABBYY OCR Engine

https://forum.uipath.com/t/abbyy-ocr-engine/2237/10?u=akila93

Else this is an application you can click on to the particular position and send copy the content and use it.


#4

You can scrape data by fine tuning some properties like “Profile” -> Scan and “Scale”-> 3.
Please find attached xaml.
Main.xaml (6.1 KB)


#5

image


#6

Thanks for the comment Tiberiu!
do you mean the basic or the desktop?
If so, i tried but it did not work well.


#7

Try experimenting with the scale of the OCR a bit. Also, since the background is darker than the text to be scraped, invert the color using google OCR option and then scrape. I tried scraping the code from this image directly from this page and my best result was: P50020CD. It’s missing the @. I’ll save the image and try again to see if something can be done.

You can use ABBYY OCR as well. However, you need a licensed version of ABBYY first. The results are way better.


#8

Hi akila, thanks for the details.
I succeeded with google OCR but still thanks for the help, guess info provided by you will do great help from now on.
Thanks again!!!


#9

So many angles,hahahha

Thanks Prashant, I tried the same way as provided and finnally succeeded!!!:heart_eyes::heart_eyes:
However it seems sometimes we can while sometimes not.
I guess it depends on the scraping.

Many thanks!!!


#10

You know siddharth,
the funny thing is I totally experienced the same with yours,hahahhahhaha

I also edit scale but it did not work well at first,
finnally made it by indicating the image again.
Guess you are correct since the background is dark so its not easy to recognize the image.

Have a good day!!!


#11

Thanks @yangbo