Azure AI Document Intelligence
An Azure service that turns documents into usable data. Previously known as Azure Form Recognizer.
1,358 questions
After analyzing a document, are there any plans to retain the recognized tables when a document is added to the dataset?
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(19.2, 65%, 11%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ED%3C/text%3E%3C/svg%3E)
Dev
31
Reputation points
We are using custom models and labeling documents using single field and table tags.
When using the "Analyze" functionality in the FOTT 2.1, it's able to recognize all the fields and tables based on the custom model.
However, whenever I want to add the analyzed documents to the project/training dataset, the labeled tables are not being retained, only the single fields.
Are there any plans to retain the recognized tables when a document is added to the dataset?
This has been our biggest bottleneck and it would be extremely helpful if the recognized tables are retained to save us costs, time and effort considering we need to label couple of tables for each document.
Looking forward to your response.
Thanks
{count} votes
-
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(22.400000000000002, 11%, 12%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ERM%3C/text%3E%3C/svg%3E)
Ramr-msft 17,611 Reputation points2022-05-23T16:28:08.703+00:00 @Dev Thanks for the question. Please use the new Form Recognizer v3.0 Studio . The v3.0 Studio Supports training models with any v2.1 labeled data. Please refer to the migration guide.
In V3.0 Tables are currently only supported for custom template models. When training a custom neural model, labeled tables are ignored.
-
Dev 31 Reputation points
2022-05-23T23:13:52.42+00:00 Thanks for the response @Ramr-msft .
Does it mean, when you analyze a document using form recognizer v3 and add it to the training dataset, it will retain all the recognized fields/tables?
If yes, that would be good. However, migration would take some time given we have a number of projects and there are breaking changes.
I also noticed that form recognizer v3 studio's table tagging is not working all the time compared to the FOTT 2.1 which is a showstopper.
Sign in to comment