Hello, I have built a custom model with the AI Builder, which should recognize information of a PDF from individual fields and a table.
The recognition of the individual fields is 99 for each field.
The recognition for the fields of the table, however, is only 30-40.
I am currently using 3 collections, as the documents differ slightly in their structure.
I have also tagged additional fields in the table area, which are next to the important fields, to improve the recognition.
I have attached a picture, the upper area inside the black box are the individual fields, as I said everything is fine here.
The lower red area marks the table and the individual fields in blue.
The PDF documents are 99% of the time several pages long (usually 5-10 pages).
I have tagged the table, as well as each field, on every page of every document.
Unfortunately, I am not able to increase the accuracy for the table.
Is there anything else I need to pay attention to?
Thank you for your help
Hi @luca_schulz ,
You can improve your model by providing more examples in your training data set.
Improve the performance of your document processing model - AI Builder | Microsoft Learn
Regards,
Philippe
WarrenBelz
791
Most Valuable Professional
MS.Ragavendar
410
Super User 2025 Season 2
mmbr1606
275
Super User 2025 Season 2