Document Understanding
2023.4
false
  • Release Notes
    • 2023.4
      • 2023.4.0
      • 2023.4.1
      • 2023.4.2
      • 2023.4.3
      • 2023.4.5
      • 2023.4.6
Banner background image
Document Understanding Release Notes
Last updated Apr 19, 2024

2023.4.0

26 April 2023| LTS Release

Stay up to date with all the latest news regarding ML Packages by going through the next list of changes that have occurred since the last LTS release until now.

What's New

A list of seven new Out-of-the-box pre-trained ML Packages is now available for general usage. Here's the list of the mentioned seven new models:

  • Certificate of incorporation/Good Standing
  • Certif of Origin
  • Children Product Certificate
  • CMS1500
  • EU Declaration of Conformity
  • Invoices Shipping
  • Pay slips
Overall score for all pipelines is now an Accuracy. Previously it was an F1 score. The evaluation artefacts in AI Center still contain both accuracy and F1 score, for backwards comparability.

A new version of the Out-of-the-box Pre-trained ML Packages (23.1.0) and their public endpoints has been released, now using cutting edge LayoutLM Transformers based architecture, which is more powerful and increases accuracy overall, especially on column fields (tables).

We added new extracted fields to the Invoices model that now has Shipping Date, Vendor email address, Bank name, Bank account number, IBAN, SWIFT Code, Bank Address, Bank Routing number, and Tax rate.

Main score displayed by Train/Evaluation/Full pipelines in AI Center is no longer F1 score but Accuracy. This is defined as the percentage correct predictions. In general, the numeric value of Accuracy is higher than F1, but it is easier to understand and interpret. Also, you now have detailed scores for each individual column field, while on older versions only a single score for all column fields taken together was available. F1 scores are still available in the artefacts/eval_metrics folder of each pipeline for continuity with previous releases.

Improvements

The Schedule (Preview) Export feature has been updated with a new minimum recurrence of seven days. All existing scheduled exports have been updated to reflect the new minimum recurrence of seven days.

The UiPath Document OCR public endpoint has been updated and now provides handwriting language support for German and French, and print language support for Danish, Finnish, Norwegian, and Swedish.

We have increased the Exported data accuracy by changing the requirement that data be labelled in 10 places (that could have been on the same page) to 10 different pages.

For all situations where latency is critical (e.g.: attended scenarios) we recommend deploying the models as ML Skills using a GPU.

We removed the import page limitation for Document Manager type imports but there is a size limitation of 4000 MiB per import.

Known Issues

The project import from AI Center is currently disabled. We are actively working on this and expect to have it reenabled soon.

Deprecation Timeline

We recommend that you regularly check the deprecation timeline for any updates regarding features that will be deprecated and removed.

Erratum 8 May 2023

Bug Fixes

We have fixed a bug that was causing a Fatal Python error: Segmentation fault when a Full or Training Pipeline was run. Now everything works as expected.
  • 26 April 2023| LTS Release
  • What's New
  • Improvements
  • Known Issues
  • Deprecation Timeline
  • Erratum 8 May 2023

Was this page helpful?

Get The Help You Need
Learning RPA - Automation Courses
UiPath Community Forum
Uipath Logo White
Trust and Security
© 2005-2024 UiPath. All rights reserved.