Textract - How to extract just certain fields

0

Hello, Im starting with Textract.

I just need to extract certain fields of invoices, but I see no option to do it, it just extract everything. How can I say to Textract which fields I need?

Thank you

asked 10 months ago236 views
4 Answers
0

Hi! Textract by default extracts all text (DetectDocumentText API) or relationship between text such as forms and tables (AnalyzeDocument API). From your post, sounds like AnalyzeDocument would suit your use case. You can then post-process Textract response to extract the required information you need. Hope this helps!

Durga_S
answered 10 months ago
0

Hello, for your usecase if you are specifically looking at extracting textract related to financial documents like Invoices and Receipts, AnalyzeExpense API might be the right API for you. Hope this is helpful!

answered 10 months ago
0

Thank you for your response. I'm trying to use API, but I receive "NoRegionError: You must specify a region." I have set it in default file, as it is in my s3 bucket. What could be the problem?

mytestingbuckettry EU (Paris) eu-west-3

[default] region = eu-west-3 output =

Thanks

answered 9 months ago
0

In short, you cannot. What you can do is use AnalyzeDocument API which will give you LINES, WORDS, TABLES, and FORMS in the entire document. You can take the response JSON and use a tool like amazon-textract-textractor to parse the response and get the desired output out of it.

That being said, since you are looking for only specific pieces of information from your document, you might want to try out the new Textract Queries feature. Using Textract queries, you can pass in questions in natural english language and let Textract find the values out of the document. Queries is supported via the AnalyzeDocument API by passing in QueriesConfig as shown below-

"QueriesConfig": { 
      "Queries": [ 
         { 
            "Alias": "string",
            "Pages": [ "string" ],
            "Text": "string"
         }
      ]
   }
profile picture
EXPERT
Anjan
answered 6 months ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions