Skip to main content

Automatically Populate Invoice Data via OCR VISION

OCR VISION is an integrated OCR feature embedded within Chrome River INVOICE that employs the latest in machine-learning-based optical character recognition technology to quickly and accurately read and extract data from invoice images. Such data as the invoice number, date, amount, vendor and purchase order number are all extracted by OCR VISION seamlessly, automating the capture and extraction process of invoice data quickly and efficiently.

OCR VISION allows you to significantly decrease manual invoice handling and review to save time and increase accuracy, validate extracted data, and submit images via “drag and drop” or send as an email attachment for OCR processing. Emailed attachments are automatically attached to the resulting invoice as an image.

You may also choose to implement automatic submission of invoices created via your organization's OCR VISION email address.

Types of Data Extracted

  • Invoice Number: If the system is unable to find or extract an Invoice Number, it will generate one in the following format: [Invoice Amount] [Invoice Date]
  • Requester: Extracted from the image and matched to your organization's master data.
  • Vendor Name: Extracted from the image and matched to your organization's master data.
  • Vendor Address: Extracted from the image and matched to your organization's master data.
  • Currency Code
  • Invoice Amount
  • Invoice Date: If the system is unable to find or extract an Invoice Date, it will use the date that the attachment was uploaded.
  • Bill To
  • Ship To
  • VAT information

If the invoice also contains a purchase order number that is in the Chrome River system, the PO will be linked for extraction of the relevant data listed above. Additionally, the following information will be extracted and used to create line items:

  • Amount Spent
  • Item Quantity
  • Item Unit Price
  • Vendor Item ID
  • Item ID
  • Item Description
  • Item UM (unit measurement)
  • Allocations
  • Cost Code
  • Timekeeper ID
  • Units of Measure
  • Line-Item User-Defined (UDA) Data

Supported Languages

  • English
  • Spanish
  • German
  • Dutch
  • French
  • Chinese
  • Japanese
  • Estonia
  • Latvian
  • Lithuanian
  • Italian
  • Portuguese
  • Romanian
  • Swedish

Attachment Specifications

  • Users may upload or email invoice images in JPG, PNG and PDF format.
  • The maximum size for a single attachment is 10 MB.
  • Emails may contain multiple attachments, up to 100 MB.
  • Only the first four pages of the attachment will have data extracted via OCR, but all pages will be attached.
  • Password-protected PDFs cannot be processed.

Upload Attachments In Chrome River

OCR Vision will automatically extract data from an invoice image uploaded via the Invoice Entry screen

1. From the Dashboard, click the +CREATE button in the upper right corner of the Invoices ribbon.

2. Drag the desired invoice image file over and drop it anywhere on the Invoice Entry screen to upload it and begin data extraction.

3. You will be asked to confirm that you want to extract data from the image. If so, click EXTRACT DATA.

  • There may be circumstances under which you wish to skip data extraction, e.g., to prevent overwriting of data copied from a prior invoice or to create an invoice with an attached image in the event that the OCR software is temporarily unavailable.
  • Clicking SKIP allows the attachment to be uploaded without triggering OCR and attached to the invoice without overwriting any data.

You may uncheck the "Ask me before extracting data when uploading to OCR Vision" box to prevent the Extract Data pop-up from appearing every time you add an attachment during this browser session.

  • This box will re-set to checked every time you log in or start a new browser session.


4. You will see an upload confirmation message that informs you the image is being processed.

5. You will see a new draft invoice for the image you uploaded. Click on it to view the extracted data and the attached image. You may continue to edit the invoice even if the image is still being processed.

6. If the attachment contained multiple pages, you will see thumbnails for each page. Data will only be extracted from the first four pages, but all pages will be attached.

7. If you then attempt to attach an additional invoice, you will be asked to choose how the system handles it:

  • Data Extract: Replace the original attachment and overwrite all the extracted data with data from the new attachment
  • Attachment: Attach an image as supporting documentation without extracting its data

  • Images uploaded for data extraction will be labeled "OCR."
  • Images uploaded as simple attachments will be labeled "ATTACHMENTS."

8. Now, continue filling out the rest of the fields as you would with any new invoice. See Create an Invoice for additional information.

Add Attachments Via Email

Chrome River users and third parties may also forward one or more invoices via email to have the data extracted automatically and create a new invoice for each attachment.

1. Forward an email with the invoice(s) attached to the email address provided by your Chrome River administrator. If you are an administrator, you may generate the email via the Customer Preferences Admin Screen.

2. Attachments must comply with the image specifications listed above.

Your organization may wish to use the OCR Email Minimum File Size threshold customer preference to filter out larger non-invoice images like email signatures to prevent them from being treated as invoices. Conversely, you may set the image-size threshold lower to accommodate smaller non-PDF invoices. See "Customize Invoice OCR Email Minimum File Size" in the Customer Preferences Admin Screen's Invoice OCR Vision section for complete details.

3. Chrome River will create one invoice for each valid attachment. Data will only be extracted from the first four pages of each attachment, but all pages will be attached.

4. Each email sent in to create an invoice will be converted into a PDF and attached to the corresponding invoice when it is automatically generated in Chrome River. The resulting invoice will then have two attachments: the invoice image and a PDF of the email.

5. The Notes section of the new invoice will contain the following data from the email:

  • Email address of sender

  • Email received date and timestamp in MM/DD/YYYY HH:MM format

6. The system will set the creator of the new invoice based on the email ID of the sender.

  • If the system cannot find the email ID, the default user assigned via your organization's business rules will be set as the creator of the invoice.
  • If the email was sent by a vendor, the same default user will be assigned as creator of the invoice.

7. The creator or the default user will receive an email notification confirming receipt of successfully processed attachments and/or offering suggestions as to why the submission may have failed. If the attachment was submitted by an external user, such as a vendor, he or she will be copied on the email notification.

OCR Success/Cancelation/Failure

  • If the upload fails because the image quality was too low, you will receive a notification stating "We were unable to extract data from the invoice image. Please upload a higher quality image or remove the invoice attachment before manually editing."
  • If the upload is canceled because it took too long to upload, you will receive a notification stating: "Invoice OCR was canceled because it was taking too long. Please try attaching Invoice again or remove the invoice attachment before manually editing."

In either case, you will be unable to edit or submit the invoice until you either remove or replace the attachment.

Confidence Levels

Chrome River customers who use OCR VISION may set a confidence level threshold for invoices that have been created via OCR in order to trigger routing rules. The software calculates a confidence level to measure the likely accuracy of the OCR data extracted.

For customers who set a preferred Confidence Level Threshold via the Invoice OCR Vision tab on the Customer Preferences Admin Screen, invoices that do not meet the selected threshold will be highlighted with an “I” icon in the Draft list.

Additionally, your organization may choose to tie confidence levels to rules that will route invoices to a specific user or role if the OCR accuracy drops below the specified threshold.

If you wish to take advantage of the confidence level indicator, simply set your organization’s threshold above 0 on the Customer Preferences Admin Screen. If you wish to tie confidence levels to your organization’s business rules, contact us via the Chrome River Help Desk.

Delete Attachments

Deleting any attachment on an invoice will delete ALL attachments on the invoice, regardless of whether they are labeled OCR or ATTACHMENT.

You will be asked to confirm before you may delete the attachment(s).

Auto-Submit OCR VISION Invoices Created by Email

Your organization may choose to enable automatic submission of invoices created via your OCR VISION email address.

When a user or vendor sends an invoice to that email address, it can be created in Chrome River and entered into approval routing with a status of Pending, eliminating the need for a user to open and submit the automatically generated invoice. Any compliance warnings or violations triggered by the invoice will display for users when they open the invoice for editing or attempt to approve it.

Your organization may choose to add an identifier to the submission email address and apply business logic to specify whether automatically created invoices should be routed to an individual or to a group for approval. For more details, see the “Invoice – OCR Vision” section of Customer Preferences Admin Screen.

If you are interested in enabling automatic submission of invoices, contact us via the Chrome River Help Desk.

Was this article helpful?