How to associate the textract output with input for human verification?

Question

After I extracting table using textract, I need some human verification for the output. I use Sagemaker to do this. I hope I can associate the extracted table cell or table row to the input table cell or row, which is in a pdf file.  Could I do that?

Answer

Detected tables are returned as Block objects in the responses from AnalyzeDocument and GetDocumentAnalysis. You can export the tables into CSV files. If you can specify more on how exactly the human verification process will work, I maybe able to suggest something more. I'm imagining something like human comparing the table in the pdf and the output table side by side.
[Amazon Textract capabilities for processing tables in pdfs](https://docs.aws.amazon.com/textract/latest/dg/how-it-works-tables.html)

How to associate the textract output with input for human verification?

Contenuto pertinente