Google Cloud OCR Error "scrape returned empty text"

ocr
google

#1

HI,

When using Google OCR in scrape relative, text results is ok. But when i change OCR engine to Google Cloud OCR, then error “scrape returned empty text” appears. Nothing else is not changed than OCR engine. Tryed to increase scaling but it doesn’t help.

I’am reding from PDF what is opened in Acrobat Reader. Text what is read is numbers like “-10.05” and “-3.06”, Error has appeared only when there is one number at left side of comma, like “-3.06”.

-mikko


#2

Hi Mikko,

If is working with the Google OCR why do you want to change with the Google Cloud OCR? Normally the Google Cloud OCR Engine returns better results, but it looks like in your case is working better with Google OCR. If the Google Cloud OCR does not return you anything even if you already play with the scale, then means that for this process (with this type of PDF file) you can not use Google Cloud OCR. All of this are Google OCR Engine limitation and we can not do nothing directly in UiPath Studio.

Thanks,
Ninett


#3

Thanks for reply Ninett,

Google Cloud OCR is giving better results than Google OCR. When it’s working, results are “100%” accurate. With Google OCR minus sign before numbers somehow “distracts” OCR engine i think. In results are more than one minus sign, spaces between numbers, wrong numbers, letters instead numbers and so on. With Google Cloud OCR none of these appears, results are perfect or it gives error #scrape returned empty text".

-mikko


#4

Just tested Google Cloud OCR again. Today it works fine!

Run it on more than 10 different PDF’s, and didn’t get any errors. Changed OCR engine to Google OCR and back to OCR Cloud and it worked fine now. Maybe UiPath just needed restarting?

Google Cloud OCR Rocks! have readed with it about 400 different values with 100% accuracy. With Google OCR almost every number needed cleanup with regex.

-mikko