Heh, a few years ago I thought of this same thing. I didn't know anything about ML so I thought it'd be possible by doing the following (for food served in a restaurant/fast-food)
1. Tag the location (if you go to McDonalds and take a picture of a Big Mac you'll see that you're at McDonalds and you have a picture).
2. Then, to get your "nutrition" info you have to manually specify what you're eating.
3. What you're eating would then be matched to a database that would provide the nutrition information. The picture basically would be there just to show you what you ate.
That’s another feature that we’d like to add to the app. We have a way to match up similar looking meals so if we knew that a user is near shake shack and had examples of their burgers we could predict the exact item.
1. Tag the location (if you go to McDonalds and take a picture of a Big Mac you'll see that you're at McDonalds and you have a picture).
2. Then, to get your "nutrition" info you have to manually specify what you're eating.
3. What you're eating would then be matched to a database that would provide the nutrition information. The picture basically would be there just to show you what you ate.
---
This looks WAY better than that.