Announcing an updated text detection model for Amazon Textract

Amazon Textract now offers an updated text detection model that enhances accuracy for long-word recognition and handwriting detection. The improvements apply to the DetectDocumentText and AnalyzeDocument APIs, which automatically serve the new model without requiring changes to existing API requests or response schemas. This ensures continuity for applications already using Amazon Textract while benefiting from the upgraded performance. The update maintains backward compatibility, allowing users to continue processing documents exactly as before.

The model expands language coverage to more than 40 languages, including Albanian, Arabic, Basque, Belarusian, Bosnian, Bulgarian, Chinese (Simplified and Traditional), and Croatian. Additional supported languages are Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Hebrew, Hindi, Hungarian, Icelandic, Italian, Japanese, Korean, Latvian, Lithuanian, Luxembourgish, Norwegian, Polish, Portuguese, Romanian, Russian, Serbian (Cyrillic and Latin), Slovak, Slovenian, Spanish, Swedish, Turkish, Turkmen, and Ukrainian. This broadens the ability to process documents in previously unsupported scripts reliably.

Accuracy improvements extend to multilingual table extraction, making it more robust for handling merged cells, skewed or rotated tables, and partial or corrupted tables. These enhancements enable more reliable processing of complex documents such as invoices, forms, and identity documents at scale. The updated model supports a wider range of document types and structures, improving overall extraction consistency.

The updated Amazon Textract model is available in all AWS Regions where Amazon Textract is currently offered. Users can access documentation and additional details through the Amazon Textract product page and the Amazon Textract Developer Guide to explore the new capabilities and integration options.