Invoice handling best practices

  • Topic Is Sticky
  • 2.1K Views
  • Last Post 23 October 2013
thomasf1 posted this 16 October 2013

Is there a starting point for invoice handling? I´ve seen the sample images to contain a invoice, there surely are some best practices on how to structure the "XML Parameters of Field Recognition" file do best deal with invoices.

Is there any chance that there will be specific method like the processMRZ Method for assisting with handling that in the the future?

Order By: Standard | Newest | Votes
Anastasia Galimova posted this 23 October 2013

The scenario for the invoice recognition depends on your aim.

If you need to extract some handprinted text, it is necessary to specify its coordinates directly. You can use the processTextField or the processFields method for this task. Several examples of XML files with settings for the processFields method can be found in the article XML Parameters of Field Recognition.

If you recognize the printed text only, you can either use the same methods to recognize some text with specified coordinates, or recognize all the text using the processImage method with the textExtraction profile. If the text contains some garbage, we recommend to perform export to searchable PDF or XML export format.

All this settings could be easily tested with the ConsoleTest application from our С# sample.

thomasf1 posted this 23 October 2013

Thanks for the Answer... There´s more to Invoice detection than the basics in the SDK though... Detecting address fields, Invoice items, summaries... I´ve seen ABBYY has quite some solutions to offer and there are a few Open Souce solutions as well (for tesseract).

The question is if there´s a way to use the SDK to do something as complex as that or if the current ABBYY solutions intelligence will come to the SDK. The ABBYY software specialized in it seems to rely currently on a desktop app...

Anastasia Galimova posted this 06 November 2013

Our offline solutions for invoice recognition (ABBYY FlexiCapture, ABBYY FlexiCapture Engine) are rather complex and need to be configured according to specific project.

We are thinking about how to implement more simple functionality for invoice recognition to Cloud OCR SDK. It should be realized in the future, but we cannot provide exact dates now.

Close