invoice2data icon indicating copy to clipboard operation
invoice2data copied to clipboard

Published 20 hours ago • verified publisher icon invoice-x

Is there any way to auto generate templates?

Open kiranbudati opened this issue 6 years ago • 5 comments

kiranbudati avatar Jun 20 '19 07:06 kiranbudati

With enough sample data, one could train a neural net to do it. If you are looking to do research in this area or can provide the training data, I'd be interested in getting involved.

m3nu avatar Jun 21 '19 04:06 m3nu

An interesting article expanding on this. With this, we can partially automate template creation and we could e.g. find invoice numbers with a certain format or dates.

https://nanonets.com/blog/ocr-with-tesseract/

m3nu avatar Dec 20 '19 15:12 m3nu

hi, I am looking to do research in this area, can you provide the training data? I'd be interested in getting involved.

johnsmithm avatar Feb 17 '20 09:02 johnsmithm

I found these guys doing automatic templates generation quite successfully https://scandocflow.com

ocr-avenger avatar Aug 11 '20 20:08 ocr-avenger

I found this project. It is using ML for field detection. Could this one be added as one of the parsers? https://github.com/naiveHobo/InvoiceNet

bosd avatar Jan 30 '22 15:01 bosd

We don't have enough manpower to implement such solution as part of this project. I also think it's out of scope of the invoice2data. See also https://github.com/invoice-x/invoice2data/issues/361

rmilecki avatar Jan 22 '23 16:01 rmilecki