Form Recognition in OnRamp

January 11, 2022

OnRamp features a form recognition tool, which allows you to automatically receive and scan version customer, vendor, bank, and internal documents. Documents such as:

• Vendor Invoices
• Customer Receipts
• Vendor Packing Slips
• Bills of Lading
• Bank Statements

This can greatly increase your financial team productivity. By automating data entry, an activity that can require added staff, you can free up multiple hours a day for that department, replacing it with a mere log review.

To set up form recognition, you must:

1. First, setup the location where the files will be stored and enter that data in OnRamp.
2. You also have to setup the Form Labelling Tool, or FLTool.
3. Next, you must setup the records for what forms will be scanned and imported.
4. Then, you must train the form tool to recognize the different forms.
5. Finally, you are ready to process the import, or allow the automated OnRamp processes to import them.

Setting up the Form Recognition Location

The first part is simple, have your IT team create a shared folder on your server. Once that address has been provided to you, you just have to set it up in OnRamp.

To setup the form location:

1. Navigate to the Business Info[S1824] screen.
2. In the Searchbar, type in OCR.
3. Select the OCR_NETWORK_LOC
4. On the screen task bar, click Edit.
5. In the Valuefield, enter the IT provided file location.
6. Click OK.

There are few other factor values that can be checked here, like OCR_STORAGE_CONTAINER, OCR_SEC_TOKEN, and OCR_STORAGE_KEY, which are provided by OnRamp. Follow the above steps with a different factor in step 3 to enter the factors, if required.

Setting up the FLTool

Once we have the factors setup in OnRamp, open a web browser (such as Google Chrome, Microsoft Edge, or Mozilla Firefox) to open the FLTool.

To setup the FLTool:

1. Enter the FLTOOL URL: https://fltool.onramp-solutions.com/connections
2. Click the to create a new connection.
3. On the Display name field, enter the OCR_STORAGE_CONTAINER factor value.
4. If desired, enter a Description.
5. On the SAS URI field, enter https://orformrecognition.blob.core.windows.net/ followed by the OCR_STORAGE_CONTAINER and OCR_STORAGE_SAS factor values.
   For example: If your OCR_STORAGE_CONTAINER is abcdocs and your OCR_STORAGE_SAS is ?sv=12302340987dfs, then your SAS URI will be https://orformrecognition.blob.core.windows.net/abcdocs?sv=12302340987dfs
6. Click Save Connection.
7. Click the to open the Application Settings screen.
8. On the Name field, enter the OCR_STORAGE_CONTAINER factor value.
9. On the Key field, enter the OCR_SEC_TOKEN factor value.
10. Click Save Settings.

Now that the one-time setup components are complete, you are ready to move on to the per-document setup bits.

Setting up the Form Recognition Models

For each document and each entity (client, vendor, or internal) you need a form model on the Form Recognition Models [S4246] screen. So, if vendor A sends you a packing slip and an invoice, then you need two separate form models.

Each form model requires at least 3 document keys. One to define the entity, one for the location, and one for the document type (invoice, receipt, etc.)

To setup the form model:

1. Navigate to the Form Recognition Models [S4246] screen.
2. Select the Type: Vendor, Client, or Internal.
3. Select the entity.
4. Select the Document Type.
5. On the screen task bar, click Add.
6. Enter the desired information.
7. Click OK.
8. Click on the Document Keys frame.
9. On the screen task bar, click Add.
10. Enter a description and a key value.
11. Click OK.
12. Repeat steps 9 – 11 to add more form keys.
13. With the keys added, select the entity type, the entity, the form type, and click Upload Modeling Documents. This will launch Windows Explorer.
14. On Windows Explorer, navigate to the location where the PDF files for this entity are stored, and select 5 – 10 files.
15. Click Open.

This will add the form models which you will label and train in the next step. While still on this screen you can also specify certain field formats on the Model Fields tab. Like if a client uses DD/MM/YY format, you can specify that here.

This screen will be required after the form tool is trained, so keep it open.

Labelling and Training the Form Recognition Tool

Here, you train the FLTool on where to view the data on the previously entered model forms. This is done by labelling the areas on the received documents. Like in the FLTool setup, this part of the process is done in a web browser.

1. Enter the URL for the form labelling app: https://fltool.onramp-solutions.com/
2. Click Open Project.
3. Select the cloud connection.
4. Select the desired form model document and click OK.
5. For each file, click the field on the screen and select the correct tag on the right-side menu.
6. Repeat step 5 for all the listed tags for each document
7. Once you have completely labelled all tags on every file, click to train OnRamp.
8. Click Train.

Once the process is complete and error free, copy the Model ID number and paste it into the Trained Model ID field for the entity and form type on Form Recognition Models [S4246].

When the overnight processes run in OnRamp, any documents saved in the shared folder will be scanned and, if the FLTool training is complete and the trained model ID assigned, the record will be imported into OnRamp.

A log record of the import is created when the files are processed. These logs can be reviewed on Form Recognition Log [S4248].

To import the files manually, go to Process Documents [S4250] where you have a button to press which runs the import.