Advanced Table Locator
This locator uses table extraction results from a separate recognition engine to locate one or more tables on a document. On its own, this locator method does not extract anything. Because of this it is necessary to configure table detection and train your project before using an Advanced Table Locator.
One document may have several tables. You can assign a table model to each unique table on the document, but if you want to extract multiple tables from that document, a separate Advanced Table Locator is required for each table label.
To ensure successful extraction, add 20 to 30 training documents for a specific table model. When that table model is detected and a corresponding Advanced Table Locator exists for a class and is assigned to a field, then that table is included in the extraction results.
Table detection does not support right-to-left languages.
Manage an Advanced Table Locator as follows:
-
Enable table detection for one or more classes.
-
Add training samples to the Table Extraction Set for a specific class.
-
Edit a table training document.
-
Train your project
The Properties of Advanced Table Locator window has the following tabs:
-
Settings tab
-
Test Results tab
Related topics: