[Help] How to classify different types of invoices in Document Understanding?

Help Document Understanding
,
1

Hi,

I am processing only invoices, but they are of different types like example type A, Type B, and Type C, each with different extraction fields.

All documents are currently in one dataset, and training has become slow due to a large volume (2500+).

I want to ask:

  • Can I create separate datasets and separate ML models for each invoice type?
  • If yes, how can I classify incoming documents correctly to route them to the right model or extractor?
  • If there is a better or recommended approach to handle this kind of situation, I am open to suggestions.

Thanks for your help!

2

@parth.modi

  1. Yes you can create separate models all together with their dedicated ml models and datasets
  2. Classification can be a separate model as well and results can be passed in ml extractor with multiple extractors and highest is picked
  3. Ideally using single is better as per the consumption as well..else more units might be consumed

it does take more training time but the over all runtime would be better and costing would be better

cheers

3

Thank you so much Anil!
I’ll try it out and update here. Will mark the thread as completed after testing.
Really appreciate your help!

1 Like