Amazon Receipt Extraction

The ​Amazon Receipt Extraction Service makes use of pre-trained Invoice models to extract Key-Value Pairs of invoice-related data making use of Microsoft Cognitive Services. The processing of invoices returns typical information found on invoices, such as Subtotal, Vat, Total, and Date. Microsoft Receipts allows users to categorize and visualize their expenses. Below is an example of a receipt.

The service extracts basic fields that can typically be found on a Receipt, such as the Receipt #, Receipt Date, Vendor Name, Shipping Address Recipient, and Purchase Orders, to name a few. In addition to this, the line items are also extracted.

Supported Content Types

  • Pdf
  • jpeg
  • png

If you need to use a different type of content, you can use the AIForged Pdf Converter.

Possible use cases

  • Extract data from receipts
  • Visualize your receipts data

Service Setup

  1. Open the Project Detail View of the project you would like to add the service to.
  2. Click on the Add Service button in the command bar.
  3. Select Amazon Receipt Extraction Service from the available Service Types.

  4. A new Service Configuration Wizard will open:
    (When navigating the Wizard, please make sure to use the Next Step button in the command bar to save any changes made).

    • Step 1 - Allows configuration of various service settings, including the name and description. The default settings are sufficient for most use cases.
    • Step 2 - Verification of any documents that have been processed.

Service Configuration Settings

The Amazon Receipt Extraction Service can be configured by the user as a flexible solution. The following Settings are available:

SettingTypeRequired TypeDescription
ArchivingStrategyOptionalDays before documents get deleted.
AccessKeyOptionalOverride the Access Key to the configured AWS cloud service.
BaseURLOptionalOverride the URL to the configured AWS cloud service.
BatchSizeHiddenProcessing batch size.
CheckElectronicOptionalCheck if the document is electronic and extract text from there.
DocumentProcessedStatusOptionalDocument status used to denote that a document has been processed.
EnabledHiddenEnable or disable the service.
ExecuteBeforeProcessWhen set up as a child service, specify whether this service should be executed before the parent service gets executed.
ExecuteAfterProcessWhen set up as a child service, specify whether this service should be executed after the parent service gets executed.
MinSizeOptionalMinimum size of the document to OCR.
PasswordOptionalUsed for service authentication. Custom Code can be used to set the password. Can be set per document.
RemoveCommentsOptionalRemove human comments from a document.

Add and Process Documents

  1. In the Amazon Receipt Extraction Service click on Inbox in the App Bar.
  2. Select the Status you want to upload and use Status None or Received for new documents that have not been processed yet.
  3. Select an optional category if you know the category for the document, if you don’t want to select one just click on “No selection”.
  4. Find the files on your Local machine and upload them. The demo’s test files can be found at the following link: Click here
  5. After all the documents have been uploaded you can check the documents to be processed, click on Processed Checked to process the documents.

It is recommended to only process a few documents at a time, especially if it is a new service to properly test if you receive the results you want before processing everything.

View Processed Documents

  1. In the Amazon Receipt Extraction Service click on Outbox button.
  2. You can view the Processing results by opening a processed doc for verification.