Getting captcha from an image everytime with OCR

Hello all.

I have a question regarding the bot I am making.
Basically, I want to create a bot to get some reports from a page that uses captcha. The captcha itself can be easily solved with OCR Get Text Activity, but the problem is that I am not managing to get the abstract object, given that when I select the object on the screen, UiPath recognices the same picture, giving me errors when the image changes.
Do you know if I have to select the object in another way? Or I have to select only the area in which the OCR activity is working and how to do it?

Really appreciate your help.

@Maximiliano_Miliozzi, Hi and welcome to the community

I think its best to just point the OCR where capture is, and maybe configure the selector to be dynamic, its possible that it captures the name or index of the captcha image so you might need to replace that with a wild card.

If you have any challenges doing this, please share the screenshot of selector.

Thank you for your quick response!

Honestly, I have no idea about how to make dynamic selectors with a wildcard but I am open to read any documentation or information about it.

Here is the screen of the selector I have:

Please, let me know if you need any other information.
Best regards!

@Maximiliano_Miliozzi, see the open in UI Explore, click that and click the indicate element, point it to the captcha code, and then share with me that screenshot

Got it,

This is what appear when I select the element:

Best regards,

@Maximiliano_Miliozzi, Try this:

select aaname, and parentid…then replace title with a wild card eg title='*'/> and test it.

I have completed what you said but I am having trouble reading the text, it is not now, but when I changed the selection. It seems that the Get OCR Text Activity is not retreiving the text from the image:


This error is not allowing me to try the changes you have suggested.

Hi, welcome to the community!
Captchas are things to stop robots from interacting with a website, we are not meant to try this, most OCR solutions will not be able to do this.