RoboScan - Automatic Receipt Scanning, Recognition, OCR and Parsing

Note: Currently this app will recognize german receipts only. See our Github repo for scans of german receipts that you can use for testing.

Automatic Receipt Detection, Recognition and Parsing with a Smartphone Camera - Softmatic RoboScan

Softmatic RoboScan uses the smartphones's camera to locate and scan receipts in real time. Once the document has been photographed the app will automatically start recognizing text and parse the contents of the receipt. Text recognition (OCR) and document analysis run entirely on device with low latency and immediate results; the RoboScan app employs the latest image processing and machine learning technologies with Google's ML Kit and OpenCV on Android and Apple's Vision framework on iOS. An online or "Cloud" connection is not required.

Video

With recent smartphones, the OCR and subsequent parsing happen in near real time (see video which features an iPhone SE 2nd gen. with A13 CPU) and will yield store's name, date of purchase, amount due, payment method and list of items. The receipt's content can then be exported to all common spreadsheets (Excel, Zoho, Numbers, Google Docs etc.).

Based on the analysis of thousands of receipts, the RoboScan app will locate, identify, and parse receipts from all leading grocery and hardware stores, gas stations, drug stores etc.:

  • Food: Aldi, Edeka, Lidl, Marktkauf, Netto, Penny, Real, Rewe etc.
  • Restaurants: Burger King, McDonald's etc.
  • Drug stores: dm, Rossmann etc.
  • Furniture and hardware stores: Ikea, Bauhaus, Hornbach, Toom etc.
  • Elektronics: MediaMarkt, Saturn etc.
  • Gas stations: Aral, Shell, Jet, Esso, Total etc.

Workflow of receipt scanning and recognition

Main screen of the app shows the list of scanned and parsed documents; receipts can be filtered by date or searched for:

RoboScan - List of scanned and parsed receipts

(Screenshot shows iOS version)

Adding a new receipt:

RoboScan - Scanning a new receipt

(Screenshot shows iOS version)

Tapping the "Camera" icon opens a live camera view with receipt recognition in real time. A properly oriented document is indicated by a green frame (hints and best practices for photographing and scanning receipts):

RoboScan - Receipt detection and cropping in real time

(Screenshot shows iOS version)

Releasing the camera shutter will capture the receipt and trigger the OCR and subsequent document analysis.

Major data points at a glance:

  • Date of purchase: 12.05.2020
  • Merchant / store: Lidl
  • Number of items: 9
  • Amount due: EUR 23,51
  • Payment: Debit card
Details of the scanned receipt filled in with the photo of the properly isolated and cropped document:

RoboScan - Scanned receipt OCR and cropped

(Screenshot shows iOS version)

For privacy reasons, all photos taken with the app stay on the device by default but can be exported or shared if so desired.

Detailed list of items with price info:

RoboScan - Scanned receipt item list

(Screenshot shows iOS version)

The list of articles is fully editable to correct misreadings:

RoboScan - Scanned receipt edit item list

(Screenshot shows iOS version)

The app's image processing can handle all common printing technologies and paper types, including the new and environmentally friendly blue-gray FSC thermal paper (unedited scan result of the receipt on the left).

Exporting scanned receipts to Excel and other spreadsheets

Receipts can be exported and saved for later use or printed out to any Air Print-compatible printer (printing is currently available in iOS version only).

Bon Beleg exportieren für Spreadsheet Tabellenkalkulation

(Screenshot shows iOS version)

The app exports the receipt as an industry-standard XLSX document which can be opened in all common spreadsheets: Microsoft Office Excel (mobile and desktop), Apple Numbers (mobile and desktop), Google Docs, Zoho Sheets:

Receipt scanning with OCR export to Excel

RoboScan for automatic receipt scanning and parsing is available for Apple iPhone and for smartphones with Android 9 or higher.


 

©2020 Softmatic GmbH. All rights reserved. Trademarks are the property of their respective owners.