Invoice parsing in greek

[Logo]

I am trying to find a solution to extract key-value pairs from invoices, however the google doc ai, fails at understanding greek, even at the ocr level (document.text returns nonsense), while it works perfectly with english. Weirdly, the google ocr tool from google vision understands greek, is there any tool/pipeline or trick to get it to work with greek as well?

[MJane]

Hi @Logo,

Welcome to Google Cloud Community!

While Google Document AI is a powerful tool, currently Greek is not included in the lists of supported languages for some of its processors such as Form Parser and Invoice Parser. 

As possible alternative, you can explore other Google Services below that might work for your scenario in extracting key value pairs for invoices:

• Google Cloud Vision OCR - Use OCR capabilities to extract text and then apply custom parsing to identify key-value pairs. It also includes Greek to its supported languages.
• Google Cloud Translation API - Extract text using Cloud Vision OCR, then use the Cloud Translation API to translate text if necessary. Cloud Translation API includes Greek in its  supported languages.
• Custom Model with AutoML - This allows you to build custom machine learning models that can be trained on specific document types and languages.

Here are some key considerations when using Document AI - OCR Level: 

• Quality of Document - Using low-quality images can result in incorrect text extraction. Make sure the document you are processing has a good resolution and is clear. 

• Document Formatting - Simplify the document format if possible. The parser's ability to extract text may be impacted by unusual formatting, intricate layouts, or non-standard fonts.

• API Configuration - Ensure that you are using the API correctly and all required parameters. 

I hope the above information is helpful.