search

graphefruit / Beanconqueror

An open source project for coffee enthusiasts.
https://beanconqueror.com/
GNU General Public License v3.0
471 stars 66 forks source link

Feature request: text analysis from image of coffee bag #192

Open tokenstar opened 3 years ago

tokenstar commented 3 years ago

This is a bit far out but would be a neat user experience

Imagine taking pic of coffee bag and the fields such as roaster, bean type, process, etc are filled in automatically

graphefruit commented 3 years ago

Hi @tokenstar,

The app will go into different solution. I link my instagram post to this: https://www.instagram.com/p/CL7fVnXh1Zg/

We're going into cooperations with roasters which will have the possibility to add qr-codes on their coffee bags, and the coffee-lover can scan this one.

The solution from you would be "non plus ultra", the problems and concerns I see (also in germany), is that all roasters have different bag designs, different information on bags, and when I'm looking out from german/english speaking region I wouldn't even know where to start.

I'll let this request open but link it to the other issue: #182

MaluNoPeleke commented 1 year ago

Maybe you can take a look at the "Coffeely" app which offers this already https://play.google.com/store/apps/details?id=io.appery.project320490

graphefruit commented 1 year ago

Maybe you can take a look at the "Coffeely" app which offers this already https://play.google.com/store/apps/details?id=io.appery.project320490

I've tried out the app yesterday. The app is doing - in my eyes - just a picture of an bag, and tries to search if a user already uploaded one, its not extracting any information out of the bag which was not put in by hand.

A discussion on discord about using the text-analysis feature from iOS or OCR was a bigger thing, and the conclusion was that its to much work for to less output at the moment.

I know for the end user it would be very handy, but when it comes to different languages, different countries, every bag is different styled, labled etc.

This won't be done any time soon

MaluNoPeleke commented 1 year ago

Maybe you can use the Web Detection API from Google for a simple mapping? https://cloud.google.com/vision/docs/detecting-web

graphefruit commented 1 year ago

Hi @MaluNoPeleke, thanks for the suggestion. I had a small look, tbh it was hard for me to get what needs to be done and not. Sadly I don't have the time to have a deeper look at this topic. There are a lot of things on the road, and for me alone developing this project, this is a too huge topic with too less outcome at this moment. If someone is willing to do a PR for this, I'd happy accept this.