Skip to content

Latest commit

 

History

History
146 lines (101 loc) · 8.56 KB

README.md

File metadata and controls

146 lines (101 loc) · 8.56 KB
Raw

Document recognition with Python Azure Form Recognizer

In this lab, you will learn how to quickly store information you collect from your documents and receipts and prepare them for further analysis.

You will learn to:

  • Deploy new resources in Azure
  • Write your own code using new Python packages for Azure
  • Analyze the information returned from the service

Prerequisites

  • Azure Subscription (e.g. Free or Student)
  • macOS, Windows, or Linux
  • Python 2.7, or 3.5 or later is required
  • Your favorite editor or Visual Studio Code

Create an Azure Form Recognizer resource

Azure Form Recognizer is an Azure Cognitive Service focused on using machine learning to identify and extract key-value pairs and table data from scanned paper documents. Applications for Form Recognizer service can extend beyond just assisting with data entry. It could also be used in integrated solutions for optimizing the auditing needs of users, letting them make informed business decisions by learning from their expense trends or matching documents with digital records.

  1. Open Azure Portal and login using your account with an existing Azure subscription.

  2. In the top search bar search for Form Recognizer. You will find it in the Marketplace section. Or go directly to the resource creation page

  3. A series of open fields will appear. Enter the following values:

    Field Value
    Name ReceiptAnalyzer. You can choose your own name, just make sure to change it in your code later
    Subscription Select Active Azure subscription
    Location Select location that works best for you
    Pricing tier F0
    Resource group Click on Create New and put receipt-analyzer as the new name
    I confirm I have read and understood the notice below Check the box

  4. Click create and wait for the deployment to finish. You should be able to see Go to resource button once your resource is ready.

  5. In the left-hand menu click Keys and Endpoint.

Accessing these values will allow you to get the values necessary for authentication for you to be able to interact with your resource. You will need these values soon so leave this tab open for now

Start your own application

Open the directory where you will be working and start a new .py file. Open it in your editor

Install the package

Install the Azure Form Recognizer client library for Python with pip:

pip install azure-ai-formrecognizer

Authenticate the client

There is a couple of different ways in which you can authenticate, but for now we are going to use the values that you have collected from Azure Portal. If you want to learn more about other ways of authenticating you can read about it here

In order to authenticate you will need to collect your credentials from the Azure Portal tab that you opened before. Make sure to replace <endpoint> with the ENDPOINT you can see in Azure Portal, and <api_key> with the value found in KEY 1 in Azure Portal.

Note: Normally, you wouldn't want to store the credentials in your code but rather as environment variables. If you want more information access this guide for other authentication methods.

endpoint = "<endpoint>"
credential = AzureKeyCredential("<api_key>")

FormRecognizerClient will be your primary way of interacting with your Form Recognizer resource. You will be able to access all of the possible options by just typing form_recognizer_client. and choosing the method you want.

For now, we want to focus on recognizing the information in a receipt. In this tutorial we will be focusing on the receipts that are uploaded and you have an url for, but if you want to try analyzing receipts you have on your computer you can find samples for using local files here

Write the logic of your application

Replace all of the code that you have in your .py file with the following sample. Make sure to change the two lines at the beginning with the information you retrieved the values for in the previous step.

from azure.ai.formrecognizer import FormRecognizerClient
from azure.core.credentials import AzureKeyCredential

class RecognizeReceiptsFromURLSample(object):

    def recognize_receipts_from_url(self):
        # This is your client - a one entry point to working with the resource
        form_recognizer_client = FormRecognizerClient(
            endpoint = "<endpoint>",
            credential=AzureKeyCredential("<api_key>")
        )

        # This is the receipt we will be recognizing - you can swap it with any other receipt url
        url = "https://raw.githubusercontent.com/Azure/azure-sdk-for-python/master/sdk/formrecognizer/azure-ai-formrecognizer/tests/sample_forms/receipt/contoso-receipt.png"

        # Here we start the recognition and poll to see if we have results available
        poller = form_recognizer_client.begin_recognize_receipts_from_url(receipt_url=url)
        receipts = poller.result()

        # We know we sent only one receipt so we get the first (and only) item in the list of returned results
        recognized_receipt = receipts[0].fields

        # Let's see what we recognized
        print("Receipt Type: {}".format(recognized_receipt.get("ReceiptType").value))
        print("Merchant Name: {}".format(recognized_receipt.get("MerchantName").value))
        print("Transaction Date: {}".format(recognized_receipt.get("TransactionDate").value))
        print("Receipt items:")
        for idx, item in enumerate(recognized_receipt.get("Items").value):
            print("...Item #{}".format(idx+1))
            item_name = item.value.get("Name")
            if item_name:
                print("......Item Name: {} has confidence: {}".format(item_name.value, item_name.confidence))
            item_quantity = item.value.get("Quantity")
            if item_quantity:
                print("......Item Quantity: {} has confidence: {}".format(item_quantity.value, item_quantity.confidence))
            item_price = item.value.get("Price")
            if item_price:
                print("......Individual Item Price: {} has confidence: {}".format(item_price.value, item_price.confidence))
            item_total_price = item.value.get("TotalPrice")
            if item_total_price:
                print("......Total Item Price: {} has confidence: {}".format(item_total_price.value, item_total_price.confidence))
        print("Subtotal: {}".format(recognized_receipt.get("Subtotal").value))
        print("Tax: {}".format(recognized_receipt.get("Tax").value))
        print("Total: {}".format(recognized_receipt.get("Total").value))
        print("--------------------------------------")

if __name__ == '__main__':
    sample = RecognizeReceiptsFromURLSample()
    sample.recognize_receipts_from_url()

Run the app

If you are using the Python extension in VS Code simply click Ctrl-Alt-N (Windows or Linux) or Cmd-Option-N (macOS). Otherwise just run

python <name-of-your-file>.py

Explore the results

Congratulations! You have successfully analyzed the receipt information. You should see the values from your receipt printed right after running your application. You will be able to directly access all of the information and analyze the information. For example you could calculate things like average cost of your transactions etc.

Next steps

If you are interested in learning more about different things you can do with your Form Recognizer resource, such as writing and training your own models you can find more documentation here

For more samples visit our GitHub page

If you have any feedback for us feel free to fill this form