Automate Report Extraction with Microsoft Azure Cognitive Service: Form Recognizer – Part 1

March 21, 2022 | Comments(3) |

Introduction to Microsoft Azure Form Recognizer

Traditionally, organizations orchestrated numerous AI talents, integrated business logic, and constructed a user interface to move from development to deployment – all of this took time, expertise, and resources.

With Azure Applied AI Services, developers can upgrade business processes in days rather than months. Using a combination of Azure Cognitive Services, task-specific AI, and business logic, these services enable you to accelerate time to value for specific business scenarios.

In this growing industrial era, when the organization wants to process or extract the required information from invoices, receipts, and documents for some business requirements, it is time-consuming and prone to errors when there is human intervention. Companies can save time and money by automating this procedure. However, collecting information from photos and documents programmatically is challenging and necessitates complex machine learning models. Azure can help with this.

Let us understand Azure Form Recognizer in detail:

What is Azure Form Recognizer?

Microsoft Form Recognizer is a cloud-based Azure Applied AI Service that extracts key-value pairs, text, and tables from your documents using AI-powered data solutions. Form Recognizer examines your documents and forms, extracting text and data, mapping field relationships as key-value pairs, and returning a structured JSON output. Form Recognizer is used to automate the data processing in applications and workflows, enhance data-driven strategies, and enrich document search capabilities.

The following document data is quickly identified, extracted, and analyzed by the Form Recognizer:

  • The structure and content of the table.
  • Field values and form elements
  • Alphanumeric text, both typed and handwritten
  • The interrelationships between the elements.
  • Key/value pairs
  • Bounding box coordinates for each element.

Form Recognizer features supported by v2.1:

  • Layout API – Extraction and analysis of text, selection marks, and table structures, along with their bounding box coordinates, from forms and documents.
  • Custom model – Extraction and analysis of data from forms and documents specific to distinct business data and use cases.
  • Invoice model – Automated data processing and extracting key information from sales invoices.
  • Receipt model – Automated data processing and extracting key information from sales receipts.
  • ID document model – Automated data processing and extraction of key information from US driver’s licenses and international passports.
  • Business card model – Automated data processing and extracting key information from business cards.

Form Recognizer features supported by v3.0:

  • Read – Extract text lines, words, detected languages, and handwritten style if detected.
  • General document model – Extract text, tables, structure, key-value pairs and, named entities.
  • Layout model – Layout API has been updated to a prebuilt model.
  • Custom model API v3.0 supports signature detection for custom template (custom form) models. In addition, it offers a new model type Custom Neural or custom document, to analyze unstructured documents.
  • Invoice model – Automated data processing and extracting essential information from sales invoices.
  • Receipt model – Receipt model v3.0 supports the processing of single-page hotel receipts.
  • ID document model – Prebuilt ID document API supports the extraction of endorsements, restrictions, and vehicle classifications from US driver’s licenses.
  • Business card model – Automated data processing and extracting key information from business cards.

Data Privacy and Security for Form Recognizer

Compliance, privacy, and security are all priorities for Form Recognizer. But, on the other hand, you oversee how this technology is used and implemented.

Let us see how form recognizer Microsoft process data:

  • Authenticate (with subscription or API keys)

Using the customer’s single-service (Form Recognizer) or multi-service (Azure Cognitive Services) API key is the most popular approach to authenticate access to Form Recognizer. An authentication header must accompany every request to the service URL. This header includes an API key (or token, if appropriate), which is used to verify your membership to a service or set of services.

  • Secure data in transit (for scanning)

For encrypting data during transit, all Cognitive Services endpoints, including the Form Recognizer API URLs, use HTTPS URLs.

  • Encrypts input data for processing

The incoming data is processed in the same Azure region as the Cognitive Services Azure resource. So, for example, when you submit a document to a Form Recognizer operation, it begins analyzing it to extract all text and identify the document’s structure and critical values. Your data and results are then encrypted and stored in Azure Storage temporarily.

  • Retrieving the results

The “Get Analyze Results” operation is authenticated using the same API key used to run the “Analyze” activity to ensure that no other customer may access your data. The procedure also returns the extracted results in JSON format when the status is completed.

Data stored by Form Recognizer

  • For all analysis, the data and extracted results are temporarily stored in Azure Storage in the same region to permit asynchronous processing, check the completion status, and return the extracted results to the client upon completion. All clients share the temporary storage in the same region. With their Azure membership and API credentials, the customer’s data is logically segregated from those of other customers.
  • For customer trained models: Customers can use the Custom model functionality to create custom models using training data stored in Azure blob storage locations. After analysis and labeling, the interim results are kept in the exact location. The trained custom models are logically segregated with their Azure subscription and API credentials and saved in Azure storage in the same region.
  • Deletes data: The input data and results for all features are deleted after 24 hours and are not used for any other purpose. Customers can delete their models and associated metadata at any time using the API for customer-trained models.

Conclusion:

Azure Recognizer’s deep-learning-based universal models support many languages that can extract multi-lingual text from images and documents, including text lines with mixed languages. Furthermore, it uses Natural Language Processing (NLP) to detect and extract information from forms and documents supported by AI to provide more structure and report text extraction. Hence, it helps us quickly get accurate results tailored to the specific content without excessive manual intervention or extensive data science expertise.

About CloudThat:

CloudThat is the official Microsoft Gold Partner, AWS Advanced Consulting Partner, and Training partner helping people develop knowledge on cloud and help their businesses aim for higher goals using best in industry cloud computing practices and expertise. We are on a mission to build a robust cloud computing ecosystem by disseminating knowledge on technological intricacies within the cloud space. Our blogs, webinars, case studies, and white papers enable all the stakeholders in the cloud computing sphere.

Feel free to drop a comment or any queries that you have regarding Azure Form Recognizer, Azure Cognitive service, Azure form recognizer API, we will get back to you quickly. To get started, go through our Expert Advisory page and Managed Services Package that is CloudThat’s offerings.


3 Responses to “Automate Report Extraction with Microsoft Azure Cognitive Service: Form Recognizer – Part 1”

Leave a Reply