Power Community

Power Community

Extract text from Image and PDF – Using Power Automate and AI Builder

Introduction  

 

In today’s data-driven world, we often need to extract valuable information from various sources, including images and PDFs. Power Automate and AI Builder offer a powerful solution for extracting text from these documents, enabling businesses to streamline processes and gain insights from previously unstructured data. 

 

Power Automate and AI Builder 

Power Automate is a cloud-based automation platform that allows users to create automated workflows, or flows, to connect various services and applications.  

AI Builder, a component of Power Automate, provides pre-built AI models that can be integrated into flows to perform tasks such as text recognition, sentiment analysis, and object detection. 

 

 

Setting Up Text Extraction Automation 

Before diving into automation, ensure that all necessary components are in place. This includes access to Power Automate, AI Builder, and any additional tools or services required for the specific workflow. 

Go to the Power Automation dashboard and click on “Instant Cloud flow”. And select “Manually Triggered flow”. Give the name of flow whatever suits you, I given name “Extract text from images or PDF documents using AI Builder and Power Automate”. 

 

sandeepstw_0-1701184386862.png

 

 

 

Then click on add action to add a new action in flow, search for “AI builder” in the “Add new action dialog box”. And click on see all in right side to see all action available in that component.  

sandeepstw_1-1701184386864.png

 

 

 

In the expanded list you can see all available actions in the AI connector. We need to find the action “Recognize text in an Image or in a PDF document”. 

 

sandeepstw_2-1701184386865.png

 

 

“Recognize text in an image or in a PDF document” contains a file input parameter. The rest of things will be done by this action itself you do not need to worry about it.  

 

sandeepstw_3-1701184386867.png

 

 

 

We need to add a variable to appending text to a string so we can use it further in our flow. I added an action “Initialize a variable” action. I named it “Extract Text.” 

 

sandeepstw_4-1701184386867.png

 

 

 

We will add one more variable to add a new line in String. I added a variable “NewLine” variable.  

sandeepstw_5-1701184386868.png

 

 

Next step to add “Apply to each” control and pass response from action “Recognize text in an image or in a PDF document” ai build action.  

 

sandeepstw_6-1701184386869.png

 

 

Again, we will add “Apply to each” control to fetch each line from the AI response.  

sandeepstw_7-1701184386870.png

 

 

 

Add new action “Append text to string” under variable control. And then from the name dropdown pick the name “Extract Text” created before in this flow. And in value, input adds values ‘line’ coming from ai response as a new line and “NewLine” variable to add a new line in the string.  

sandeepstw_8-1701184386871.png

 

Add new action to fetch profile information from office database ”GetProfile V2”. 

 

sandeepstw_9-1701184386873.png

 

 

At the end of this flow, we will send an email with extracted data to see the output and send the uploaded file as an attachment. For this, we will add “Send Email V2” from Outlook Connector.  

 

sandeepstw_10-1701184386875.png

 

 

In Parameter, we are using – 

 

To – Mail (From Get profile v2 action) 

 

"emailMessage/To": "@outputs('Get_my_profile_(V2)')?['body/mail']", 

 

 

Subject – “Text Recognition result” 

 

"emailMessage/Subject": "Text Recognition results", 

 

 

Body –  

 

Congratulations @{outputs('Get_my_profile_(V2)')?['body/givenName']}! Your Power Automate flow ran successfully. Here is all the text extracted from the attached file:nn@{variables('Extracted text')}nnLearn more about AI Builder Text Recognition and the information it can extract: https://go.microsoft.com/fwlink/?linkid=2161167

Attachment –  

"emailMessage/Attachments": [ 
        { 
          "Name": "@triggerBody()?['file']?['name']", 
          "ContentBytes": "@triggerBody()?['file']?['contentBytes']" 
        } 
      ] 

After adding all these values make sure you saved flow.  

 

 

Testing – 

 

Run flow using the flow detail page. It will ask for an image upload. Upload an image. Once flow runs successfully. You will get a mail with data and upload an image. Like this. 

 

Email body  

 

Congratulations Sandeep! Your Power Automate flow ran successfully. Here is all the text extracted from the attached file: 
 
Home Marketing Sales v Operations v Productivity v Business Analysis v HR Admin v Master v Settings v Edit Public group & 22 members … S T W STW365 + New v 6] Promote Translation Page details J Analytics Published 8/27/2023 Share v Swati Chavan is editing this page 6 Edit LEADS DASHBOARD LEADS DASHBOARD T W FY FY 2023-2024 V Industry Source City Sales Person Status All V All V All V All V All V Disqualified Open Qualified 7 123 13 STATUS RECORD SOURCE TYPE 
 
Learn more about AI Builder Text Recognition and the information it can extract: https://go.microsoft.com/fwlink/?linkid=2161167 

 

sandeepstw_0-1701184685847.png

According to the uploaded image we will same data in email, we got from power automate flow.  

 

 

Hope this will help you to extract data from ID card, pdfs or any scanned document to automate your process in your business using Power Automate and AI builder.  

 

Happy learning and process automation.  

 

Thanks!! 

This post was originally published on this site

- Advertisement -spot_img

LEAVE A REPLY

Please enter your comment!
Please enter your name here

- Advertisement - Advertisement

Latest News

Filter data with single date slicer when multiple dates in fact table fall in range without creating relationship in Power BI

Hi Folks, After a while, I am back with another interesting way to solve this type of problem in Power...

More Articles Like This

- Advertisement -spot_img