Can i extract text from an image and populate the extract data inside sharepoint

[johnjohn123]

Hi All,

I have the following scenario, but not sure How/IF i can implement it using Power Automate and SharePoint list:-

1) We have a club registration cards which include VIP member and Regular member.

2) the 2 cards types have different info and different layout.

3) we want to scan the documents using scanner >> upload the scanned image to SharePoint document library  >> then run a Power automate workflow which will check the image type (VIP card or Regular member card)

4) then based on this to populate the following data into a SharePoint custom list:-

- Content type (VIP or Regular)

- First Name

- Middle Name

- Last Name

- Registration number

- Role (Only for VIP cards)

- Member image

- Expiry date

- Issue date

- Department (Only for VIP cards)

I heard that Power automate provides an OCR capabilities, so how i can benefit from it? on the other side there is SharePoint Syntex which allow us to use AI capabilities to extract data.. so not sure from where i need to start? and which approach i can follow? Power App OCR or SharePoint Syntex and how i should utilize the correct approach?

Thanks

Here is a general sample of a Regular Member card:-

[PrasadAthalye]

You can use following AI Builder ready template and alter it to save the output to SharePoint List.

Go to AI Builder Menu - Build and you can select Extract text from images and then select Using Flow.

[johnjohn123]

@PrasadAthalye  thanks for the reply, but which ready template i need to use? is it the form processing? can you provide a screen shot?

Thanks

[PrasadAthalye]

Use this one ...

[JoeF-MSFT · Accepted solution]

Hi @johnjohn123,

One way to OCR those registrations cards is by using AI Builder Text Recognizer as suggested by @PrasadAthalye. Just note that Text Recognizer will give you all the text present in the card, so you will need to a way to parse and extract the right information. 

An easier approach can be to use AI Builder Form Processing, where you can train a custom AI model to extract the fields you want. The following tutorial will walk you how to achieve this: Get started with Form processing in AI Builder - Learn | Microsoft Docs since you have two card layouts, you will need to define two collections in your model.

Hope this helps. Let us know how it goes!

[johnjohn123]

@JoeF-MSFTyes i think AI Builder Form Processing is the way to go.

Thanks

[PrasadAthalye]

Form Processing works better with PDFs. I see following note so you should ensure that Images are high quality.

"You can use images or PDFs of your documents to train your model, but text-embedded PDFs will provide the best results. Images should be as high-quality as possible to avoid any issues in detecting text."

I would still try Text Recognizer...

[Siddharth2]

CAN ANYONE TELL HOW TO ACHEIVE THIS PROCESS WITHOUT USING PREMIUM ACTIONS LIKE TEXT RECOGNISER , PLEAS PLEASE HELP , PLEASE GIVE WHOLE PROCESS WITH IMAGES FOR BETTER UNDERTANDING