search

NielsRogge / Transformers-Tutorials

This repository contains demos I made with the Transformers library by HuggingFace.
MIT License
8.51k stars 1.34k forks source link

Pix2Struct: Can we use this to extract tables? #292

Open ladanisavan opened 1 year ago

ladanisavan commented 1 year ago

Pix2Struct works quite well with form data (key-value pairs). Can we use this model to extract tables also from document?

NielsRogge commented 1 year ago

Hi,

Yes you can make Pix2Struct learn to generate any text you want given an image, so you could train it to generate the table content in text form/JSON given an image that contains a table.