Ref: https://learn.cantrill.io/courses/1820301/lectures/42176882

• Summary:
• Contents:

Amazon Textract - Key Concepts

• 🔧 Detect, extract and analyze text contained in input documents
  • Common documents: receipts, invoices, medical records, identity documents…
  • Outputs extracted text with structure and analysis
• Functionality
  • Detection of text
    • e.g. scanned invoice or receipt → detects prices, products, dates…
    • Text can be computer generated or handwritten
  • Detects relationships in the text
    • e.g. associated tax in the price
  • Metadata generation
    • e.g. where exactly in the receipt is price shown
• Features
  • Several supported input formats: JPEG, PNG, PDF, TIFF
  • For most documents synchronous operation (real-time)
    • Large documents (e.g. big PDFs) → asynchronous
  • Pay-per-use (with custom pricing available for large volume)
• Use cases
  • Document analysis (names, address, birth date…)
  • Receipt analysis (prices, vendor, line items, dates…)
  • Identity documents (abstract fields e.g. PassportID…)