Scribd
Upload
61 views21 pages

Begin Extracting Information From Receipts The Smart Way With Azure Form Recognizer

This document discusses how Azure Form Recognizer can be used to extract information from images and documents like receipts and invoices using pre-trained machine learning models. It provides step-by-step instructions on setting up a Form Recognizer resource in Azure, obtaining the API key and endpoint, and using a sample receipt to demonstrate how Form Recognizer can recognize fields in the receipt including the tip amount.

Uploaded by

Purvang Joshi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
61 views21 pages

Begin Extracting Information From Receipts The Smart Way With Azure Form Recognizer

This document discusses how Azure Form Recognizer can be used to extract information from images and documents like receipts and invoices using pre-trained machine learning models. It provides step-by-step instructions on setting up a Form Recognizer resource in Azure, obtaining the API key and endpoint, and using a sample receipt to demonstrate how Form Recognizer can recognize fields in the receipt including the tip amount.

Uploaded by

Purvang Joshi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 21

Begin extracting information

from receipts the smart way


with Azure Form Recognizer
• It is quite difficult for anyone to programmatically extract information
from images and documents.
• It also requires some complex machine learning models.
• This is when Azure Form Recognizer comes into picture.
• It comes with pre-trained machine learning models that can recognize
information on various documents, receipts and invoices.
• Let’s go ahead and see how we can do that in this step by step demo.
Search for Form Recognizer in the Azure Portal and
create it.
Put it in a resource group. Choose the location and
choose free tier which will be enough for us now. Click
review + create and then click create.
This is what the form recognizer service would
look like once it gets created.
Now go to keys and endpoint. For the demo purpose, we are
going to need key 1 and the endpoint. Copy those and keep
them aside for a while.
For the demo purposes, we are going to use this sample
Contoso receipt. You can easily find this by a simple google
search.
We are going to run the demo in the form recognizer sample
tool. That is a demo website that is going to help us test the
Form Recognizer without building an application of our own.
Form Recognizer can use a pre-built machine learning
model.
You can use custom training as well and you can use layout to
get text, tables and selection marks.
• We are going to go ahead with the pre built models.
• So, from the options, choose the prebuilt model.
First of all, put in the key and the endpoint in the
respective fields.
For the source, choose URL, put in the image URL and click
Fetch. It will fetch the image for you.
This tool then creates an API request that it will
send to the Form Recognizer service.
Click Run Analysis.
You will see that it will recognize the details in the
image.
And you will find the results down here. You can also find
confidence which you can use in your apps. However, if you
noticed, it hasn’t recognized the tip here.
That is because we had chosen the form type as invoice earlier.
Change it to receipt and run the analysis again.
Now you will find the tip as well.
You can download the prediction results in both JSON and
CSV format and then use it further in your applications.
• So, you just now saw how Azure Form Recognizer removes the
complexity and helps you automatically analyse and recognise
documents like receipts and invoices.
• You can use this one as a REST API and with SDKs for several
languages.
• What are you waiting for? Go ahead and try this yourself.

You might also like