JA EN
shiichan

Amazon Textract's New Model Covers 40+ Languages, With Zero Code Changes

AWS What's New ★★★ View Markdown
#aws#ai

Hi everyone, it's Shii-chan! Textract has a new model, and you don't have to do anything to get it. Nice!

AWS What's New aws.amazon.com

What was announced?

In AWS What's New, AWS announced an updated text detection model for Amazon Textract.

It applies to the DetectDocumentText and AnalyzeDocument APIs, and brings better accuracy, wider language support, and stronger multilingual table extraction.

The story so far

Until now, documents in scripts that Textract did not support could be hard to read. The announcement says the updated model enables reliable processing of previously unsupported scripts.

What changes

Here is what changes:

  • Language coverage expands to more than 40 languages
  • Long-word recognition and handwriting detection are better
  • Multilingual table extraction is more robust on merged cells, skewed or rotated tables, and partial or corrupted tables

These improvements help you process invoices, forms, identity documents, and other multilingual or complex documents at scale.

Dive Deep

The key point is that you do not need to change your API requests or the response schema. Your applications are served by the updated model automatically, so you can keep calling Textract exactly as before.

The updated model is available in all AWS Regions where Amazon Textract is available.

The announcement lists these languages:

  • Albanian, Arabic, Basque, Belarusian, Bosnian, Bulgarian
  • Chinese (Simplified and Traditional), Croatian, 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, Serbian Latin, Slovak, Slovenian, Spanish, Swedish, Turkish, Turkmen, Ukrainian

The announcement does not give specific accuracy numbers.

Wrap-up

  • The model behind DetectDocumentText and AnalyzeDocument has been updated
  • More than 40 languages are supported, and long-word and handwriting detection improved
  • Multilingual table extraction handles merged cells, skew, and partial tables better
  • No code or schema changes are needed, and it is applied automatically in every Region where Textract is available

If you run OCR on multilingual documents, you get the benefits without any extra work!

Related posts