Accelerated Document Processing - Amazon Textract

Manual data entry is a major challenge companies face every day when trying to extract useful data from scanned documents, such as PDFs and images, because it is error prone, expensive, and time-consuming. Simple optical character recognition (OCR) software and APIs for invoice processing and customer-form extraction, for example, requires manual configuration, such as picking up a sample document, configuring fields, testing for accuracy, and validating results for each type of document. 

Amazon Textract is a machine learning (ML) service that automatically extracts text, handwriting, and data from scanned documents. It goes beyond simple OCR to identify, understand, and extract data from forms and tables.

MothersonSumi Infotech & Designs Ltd. (MIND) offers a solution for an accelerated pipeline using Amazon Textract, which helps customers to reduce cycle time, i.e., setup time with additional functionality, like human loop for manual intervention and Amazon Augmented AI (Amazon A2I) enablement, which improves the overall quality of data extraction. Customers can use any scanned documents, including invoices, financial, legal, or medical documents, for easy analysis and review. Pre-processing of these documents is done in AWS Lambda using a function present for image processing in Pillow/Python Imaging Library (PIL). Additional post-processing is done by using validation checks, i.e., comparing summation of individual values extracted vs. total extracted, and implementing custom business rules.

Amazon Textract is a HIPAA-eligible service, which means it can be used to process protected health information (PHI) extracted from images to power healthcare applications. Also, custom redaction algorithms are used to mask personally identifiable information (PII) information in customer documents, which further enhances the security and privacy.