When to retrain a project for table detection

You need to retrain your project for table extraction when the following circumstances occur:

An existing table changes and is no longer extracted successfully.
A new table is added to a document from an existing class.
You want to process a new document type with one or more tables.

In any of these situations, additional training documents that contain the new or modified tables are needed. For the best results use training documents are typical examples, and not mocked up documents that contain unrealistic data or layouts.

To update your project before retraining for table detection, perform the following:

Add new training documents to the Table Extraction Set for the relevant classes.
Edit the table training document by adding table labels and columns, as well as other settings.
Either add or edit a corresponding Advanced Table Locator or script that makes use of the new table detection results.
Test your table extraction results.
Train your project.