AI service
Amazon Textract
Extracts printed text, handwriting, forms, tables, and document structure.
Key points
- Extracts text and structured data from scanned documents and images.
- Can identify forms, key-value pairs, tables, selection elements, and document layout.
- Supports printed text and handwriting in supported scenarios.
- Returns machine-readable output that applications can process downstream.
- Targets document understanding rather than general image recognition.
When to use it
- Choose Textract for invoices, receipts, tax forms, loan packets, and scanned documents.
- Use it when the output must preserve forms, tables, or key-value relationships.
- Use it to reduce manual data entry from document workflows.
Exam tips
- Textract is not just OCR; it understands document structure such as tables and forms.
- Rekognition can detect text in images, but Textract is the better answer for documents and forms.
- Comprehend can analyze extracted text after Textract has converted the document to text.
- A2I can add human review when document extraction confidence is low or policy requires review.