How to deploy the Intelligent Form Extractor On-Premise ?
Make sure the pre-requisites are met before starting the installation:
Set-up AI Center On-Prem and Orchestrator
If air-gapped, make sure you have also followed steps listed here .
Now you are ready to install IFE and use it in a workflow. There are two main steps that a user needs to do to use Intelligent Form Extractor (IFE): Set-up Handwriting Recognition & Set-up Intelligent Form Extractor. Please note that if you want to just use Form Extractor (FE), you can skip the “Set-up Handwriting Recognition part” and just perform the “Set-up Intelligent Form Extractor” (but make sure you perform the steps on Form Extractor).
Set-up Handwriting Recognition
Navigate to Document Understanding ML packages and create a handwriting package.
The package can be created by hitting “Submit” on this page and providing essential information – Package Name – on the next page. The other information is optional and can be provided/left out based on user preference.
Once the handwriting package is created, you need to create a corresponding ML Skill. The ML Skill can be created by hitting “Create new” on the ML Skills tab and supplying the needed information in all the fields. Note that “Skill description” is optional but we recommend providing this information for later convenience. Make sure the “Enable GPU” toggle is turned off.
Once you hit create and the skill gets created, click on the row corresponding to the skill you just created in the ML Skills tab and go to “Modify Current Deployment”. Turn on the toggle and make the ML Skill Public:
Once the ML Skill is updated and made public, click on the row corresponding to the skill you just created in the ML Skills tab and copy the URL from the top section. This is the URL that we will later use for setting up the skill for IFE.
Set-up Intelligent Form Extractor
Navigate to Document Understanding ML packages and create an Intelligent Form Extractor package. Provide all the needed information. You need to select “UiPath OCR” in the “OCR Engine” field and provide the URL that you obtained previously from public handwriting ML Skill in the “OCR URL” fields. Note that even though both these fields are suggested as “Optional”, you still need to provide this information. Not providing this information will result in errors and you will not be able to use IFE.
Next thing is to create an ML Skill corresponding to the IFE package by clicking on “ML Skills”, hitting “Create new” and providing the required info:
Once the ML Skill is ready and the status changes to “Available”, click on that ML Skill and go to “Modify current deployment” and make the ML Skill “Public”:
Once the skill becomes “Available”, click on that Skill and copy the URL. This is the URL that you will use in the Studio workflow:
Now you are ready to use IFE in a workflow:
Airgapped: use the URL obtained above in the “Endpoint” configuration of IFE and leave the API Key field blank or “”:
Non air-gapped: provide the DU API key