OCR incorrectly capturing data from window pop-up in Chrome

Scenario : I am currently stuck in the datascraping part where I have to scrape confirmation ID after uploading a report.
I have used Chrome and ,here, the windows pop-up for confirmation does NOT appear as a separate window(unlike I.E).It superimposes on the same Chrome screen.
Only output mode is OCR

Problem : Individually, this pop-up window cannot be spied. “Indicate Element” reveals nothing of this pop-up but correctly identifies the Chrome page elements beneath it in the website. (As if it has turned invisible while spying!)
The entire window frame can be captured using OCR and data can be obtained likewise. But the OCR is not correctly identifying output conf. ID
It misidentifies “8” as “a3” or something like that.

Attempted Solution : MS OCR is absolutely not working.
In GoogleOCR, it is showing somewhat better results with “invert” option as checked and Scale as “4”. But still it is inconsistent.

I need help to scrape specific Data off Chrome pop-ups.

Try using Computer Vision.

Hi Rahul,

I added Computer Vision Activity after setting the Beta Feeds. The issue still remains .

As far as I know, the Google OCR is renamed to Tesseract, so maybe, that’s why they’re not working properly.
Any other solution ?

You are using Screen Scraping over here.
In Computer Vision activities, use CV Scope and identify this popup using the selector.
Then, use CV get text and validate the results.

Hey @Rahul_Pasupuleti

Can you try to use Get Visible Text activity on the Popup and let me know the result because i also did the same and it worked for me.

Hi Amarasto,

I used “Get Visible Text” action but the text on the chrome pop-up did not get captured/detected.

I’m getting the following o/p :

Yes, my bad. I downloaded the upgraded UiAutomation package instead of CV.
Anyways, I have downloaded the CV package but I have the below questions :

  1. How to get the ApiKey which is required in CV Screen Scope.
    Any other setting required ?

  2. Inside CV Screen Scope, default OCR engine is MS OCR which , as we had discussed earlier, isn’t useful for this purpose. Could I replace it with Tesseract ?

Hi Sudip,

You don’t need to pass any API Key, However URL has to be : “https://cv.uipath.com

Have you tried with CV get text activity?


Why not you use Ui Explorer? I had similar problem when using firefox and all you need is to find the correct element in the explorer. Try looking something like “dialog” in explorer. Hope this helps!

Edit: Try select whole browser with UI Explorer.

Hi Rahul,

I added the CV Get text as below :slight_smile:

The results appear to be TRANSLATED although UI shows English
For e.g , the output for the below is : “b84ec7f5e056a37aoaed6603cgc13di0•S”

Whereas, for the 2nd item, it is “üB6Bbüi365i08•Sb281e52cgdii”

Do we have to fine tune from “Properties” . If so, what?

Hi Chakri,

The UiExplorer isn’t identifying Chrome pop-ups.
As you can see, the inner text shows the page text and NOT pop-up text for the same area.

How did u capture this area in firefox using UiExplorer? Any specific detail that I could’ve missed.