I have a form parser processor setup, and I only need the bounding box of the detected page in my image, I don't need it to do the table text extraction as well. Is there anyway I can do this (if yes, will this take lesser latency somehow?). Currently it takes ~5 seconds per page. I'm using the python library.
I already have extraction set up properly using the python client library.
There isn't a way to only perform object detection (such as table detection) using the Document AI Form Parser.
There are a couple of options to get what you're looking for:
In your post processing, you can search the Document object response for the presence of the Document.pages.tables field without getting the data from the table. Then you can access the bounding box information in the Table.layout field.
If latency is a concern and you only need specific fields in the Document object response, you can provide a fieldMask in your processing request which lists the top-level fields that you want returned in the Document object. For example pages.tables would only return the pages.tables field.
