AI Screen shot of table to Coda Table

Hi Coda! Thanks for an amazing product! I’m not sure if other people would find this useful, but I know this would be AMAZINGLY useful for my company: A way of extracting data from a screen shot of a table and converting it into a coda table (or simply a string of text).

1 Like

I’m kinda doing that here -

I’ll experiment with some data grid snapshots.

1 Like

Thanks Bill, a way of incorporating data grid screen shots would be amazing!

What you have already in your post seems like it would already be amazingly useful. Is it available for me to use?

Juan

Not yet. My work has me slammed this month but I hope to share something soon.

Hi Bill,

How’s work going? Hope you’re doing well. Have you by any chance been able to do anything regarding this? Let me know if there’s anything I can do to help!

I have not, but I did read a little concerning the OCR extraction of tabular data from images and PDFs and the only viable approach is to create a Pack that uses an integrated library such as Tesseract. This would require development using the Pack CLI instead of Pack Studio, so it’s a little more involved.

The Pack I’ve experimented with that I referenced works great for general text extraction but not for tabular data. Tesseract is apparently able to recognize tables and extract the data in a manner that would be simple work to inject into a Coda schema. Unknown what level of effort would be required, but it appears to be possible.

Thanks for looking into this more. So I might have misspoken when I mentioned translating a table. In my particular case the data looks more like this:

Not quite a table (there are numbers in each row of the blue rectangle). Any chance the pack you’ve already worked on would work on this (with numbers for each row obviously)?

I’ll give this a try independently and let you know how it goes.

I’ll give it a test asap.