Hi,
I’m looking for a way to take information from a document scanned by a user and input various information from that scan to multiple places on the app. For example, on the uploaded document I need to take the Producer name (which has been blacked out) and add that to a table within the app. Most of the information on the document will need to go on a list or in a table for my app to work. Does anyone know how I can achieve this? Thanks
@redvivi any plugins that work with this?
2 Likes
Thanks @tylerboodman .
I have already answered to the OP on a different channel.
Unfortunately no “simple” OCR model provides this out of the box.
The solution is a combination of preprocessing via OCR and post processing to get a reliable and repeatable result over multiple documents, under the right format.
Any other solution won’t be reliable enough.
1 Like
How about using GPT-4 Vision for this? We have created an application for a venture startup at Zeroic using this approach, where it extracts the text from an image and returns it as a JSON, that we process for further calculations
In case it’s a regular file, you can use the vector store with OpenAI Assistants along with Structured Outputs
Feel free to try it in the OpenAI Playground and then deploy in Bubble
Thanks for taking the time to answer my question.
What would the post processing look like?
Thanks again!
This is an interesting solution. I’ll give it a try later.
Thank you for your time!
1 Like
Hi, is it the Formula bot that you are suggesting?
Thanks!