Invoice recognition and text extraction
From scanned or uploaded invoice images the system records invoices or matches them to existing ones; it extracts the text of PDFs and makes it searchable.
The system recognises uploaded invoice images: it reads the invoice data and, if everything needed is there, records the invoice or links the image to an existing invoice. It extracts the text of PDFs so that document contents can be searched.
Recognising one document
- In the Documents list open the row menu of the document.
- Click Recognise as invoice.
- The result tells what happened: a new invoice was created, the image was linked to an existing invoice, or why it could not be processed (e.g. pro forma, not an invoice, incomplete data).
If an invoice already belongs to the document, no new one is created; at most the existing one is completed. A pro forma does not become an invoice.
Several documents in the background
- Select the invoice images (up to 2000) and click Recognise selected invoice images in the background in the header.
- The system queues them per company and processes them in the background. You get a notification at the end.
- Check the result in the list with the Invoice recognition filter.
Text content
- When a PDF is uploaded, the system extracts its text in the background on its own.
- Extract text content in the row menu starts it by hand; for several documents use the header button of the same name.
- When text exists, a green badge appears in the row; click it to read the text. The Search in the text filter searches the content.
Example
Sample Ltd. scans and uploads 40 supplier invoices received by post. It selects them all and starts background recognition. Next morning the Invoice recognition filter shows 37 invoices recorded, and for 3 documents the system reports that manual entry is needed because of missing data.