Optical Character Recognition (OCR) in Power Platform and RPA

[@AlexDestov]

Благодарю за данное видео, оно было очень интересным и виртуозным в части операций с циклами и их наполнением!!!😊

[@TomaszPoszytek]

Thank you 😊

[@thesingularity7622]

Wow, I am just getting too use this tool and have no background in coding but i am heavily interested in RPA. I would have never thought one could go so deep in its usage. Great video. Keep them coming.

[@TomaszPoszytek]

Thank you ☺️

[@franciscogonzales6490]

great video. You show many options. I guess Power Automate AI Builder is the simplest option but most expensive. we are starting to evaluate options to get data from scanned invoices and enter this data into our ERP. I'm not programmer but you gave me a nice general idea. Greetings from Peru.

[@TomaszPoszytek]

Thank you. AI Builder is indeed the simplest approach and recently it received a significant reduce in its license prices.

[@Abhisheknanavare]

Thank You Tomasz, I have a query. I want to extract a specific table from pdf into excel file. I have header name. which approach is advisable here. using RPA.

[@TomaszPoszytek]

With PDF it is quite difficult to get a valid table with RPA. At least with PAD. The tool can only extract plain text from PDF so it will be difficult to read it. Maybe then you can use some regex parsing. However I'd recommend you take a look at AI Builder form processing models. If tables in PDFs have the same format, just different data, you can train the model to get data from tables and it will be stored in a JSON object.

[@vladsamoilov9905]

I think there is much easier way - you can use "Get and transform" in Excel itself - Get Data> From File> From pdf and it recognizes tables you have in pdf, you just choose what you need.

[@TomaszPoszytek]

@@vladsamoilov9905 great idea. I wasn't aware of it 😀

[@Abhisheknanavare]

@@vladsamoilov9905 How to do it using Microsoft Powerautomate Desktop version. I am not able to find any such options.

[@zollyboldizsar9114]

Thanks Tomasz for the great content. You think would be possible to analyze an invoice located on the local machine? Thanks

[@TomaszPoszytek]

The built-in OCR allows to read through contents, but it returns text in a rather unstructured way. I'd rather send the file to cloud flow and use AI Builder or Azure OCR to do the job.

[@Lennings82]

the ocr dont work so well in PAD. i have to use a 3 party snip to ocr.. is it a way to make the PAD ocr better?

[@TomaszPoszytek]

There is actually no OCR in PAD natively. It is using local machine services. The only thing which comes to my mind is to engage Azure Form Recogniser: azure.microsoft.com/en-us/products/ai-services/ai-document-intelligence

[@sathya2870]

hii Tomasz can you help me to read the captures with text in application forms

[@TomaszPoszytek]

Hi, what do you mean exactly?

[@ajayganeshie1857]

Thank You Tomasz, this was really helpfull. I am trying to scan a bulk of documents and then auto parse them in directory with there respective dates. Is something lile this possible? Thank, Ajay

[@TomaszPoszytek]

Hi, you could try to run PAD on a folder, for each stored file. However that sounds more like a job for a OCR solution. Those I know works this way, that they monitor specific network localisations and for every file added they do OCR and whatever else is configured :)

[@BRVR_]

Hello. Could you say? I'd like to OCR-recognize all uploaded jpg.files in blob automatically - what should i write in "Invoke web service - Request body"? instead of {"source": "1111108271985.blob.core.windows.net/input/April.pdf"} direct link to my file? Thanks

[@TomaszPoszytek]

Try and let me know if it worked.