So, how's it done?
As the expression goes, "inquiring minds would like to know."
At a guess, the easiest technique would be to OCR the label on the device then use an algorithm to parse the data and then search on the name. The technology to recognize textual data from an image already exists in Monterey (macOS 12). The rest would be a reasonably straight-forward Python (the language, not the snake
) scrip
Alternatively, If there is no label or text, the option would be a variation of facial recognition plus an extensive database of shapes. (Not too difficult but it would be a lot easier with 3D images.) The size of the shapes database alone would put something like that well beyond most personal computers, but easily within the capability of Google's or Amazon's servers.
To be useful, probably a combination of the two would be the best choice.
Having said that, I have spent a lot of time on Amazon recently, and I have never seen any sign of what Rita is talking about. However, Amazon's "similar products" is getting very, very good at identifying not just what I search for but often what I am really looking for based on my history. To the point, I often prefer an item from the "suggested" list over the one I searched for. Whatever you think of Amazon, that is darn good marketing. (It also means they know me well, perhaps too well.)
