Text not recognized as "HANDWRITING" in Textract

0

Hi everyone,

I have an issue with a code that was working fine in production using Textract and which does not work anymore. When using Textract, it does not detect anymore that a line or a word is handwritten and not printed. In the output of textract.analyze_document(Document={'Bytes': file_content},FeatureTypes=["FORMS"]) (Python), when there is a "TextType" field, there is always a "PRINTED" value even though the word is obviously handwritten and that for any document i used as an input.

Does anyone have the same issue ? Is it a change with boto3 or Textract itself ?

Thanks in advance, Quentin

  • Hi everyone,

    I have the same issue. Using textractClient.startDocumentAnalysis(startDocumentAnalysisRequest) all words are set to PRINTED.

    But using StartExpenseAnalysisRequest the textType is set correctly. { "BlockType": "WORD", "Confidence": 35.711299896240234, "Text": "Untaborist", "TextType": "HANDWRITING", ... }

    Best regards

gefragt vor einem Jahr236 Aufrufe
1 Antwort
0

Thank you for using Textract. We would like to help with the issue you are facing. Can you create a case through AWS Support Center in your AWS account so that you can share some sample documents in which you are seeing issue if nature documents is not sensitive.

AWS
beantwortet vor einem Jahr
  • As my company only have the Basic Support plan, we cannot contact the AWS Technical Support about this issue.

    Best regards, Quentin

  • Can you share image or document here?

Du bist nicht angemeldet. Anmelden um eine Antwort zu veröffentlichen.

Eine gute Antwort beantwortet die Frage klar, gibt konstruktives Feedback und fördert die berufliche Weiterentwicklung des Fragenstellers.

Richtlinien für die Beantwortung von Fragen