Page 1 of 1

Using Google Cloud Vision API for OCR process

Posted: Tue Oct 12, 2021 9:46 pm
by teamitdenver
Does anybody have any thoughts or experience with this?

GCV is probably the most accurate OCR for both printed and handwritten text we have used.

Re: Using Google Cloud Vision API for OCR process

Posted: Thu Oct 14, 2021 6:53 am
by RobertVib
We have an integration using Google Cloud Vision API to categorize documents. Our solution adds a custom file metadata driver that talks to Google. After the file metadata driver finishes, we use the event as a workflow trigger to do other document maintenance like change the document type, set default metadata and such.

The Google Cloud Vision API was more complicated that we thought. We hired Mayan's team to perform the initial integration and they did so in record time. After that they handed over the code and documentation. Since then we've been updating it easily for each new version of Mayan.

My recommendation is don't fork Mayan, instead work with the hooks and placeholder it provides for expansion. With a fork, you'll need to merge and handle conflicts every time you upgrade. Finally, don't go about doing this all by yourself, buy some hours from the Mayan team, you won't regret it.