A company called Blippar is in the computer vision business, which allows your phone to take the matrices of pixels it gets from its camera and identify and understand what it’s ‘seeing.’ Now, Blippar’s app can ‘see’ cars, and can identify them with 97.7% accuracy. That’s incredibly impressive, but there’s a catch.
The catch is that the Blippar app only identifies US market cars made after the year 2000. What? That’s like the last market and era of cars I’d want or need help identifying. This thing would have been great to help identify the stuff that really stumps you, the interesting stuff, like 1910s Franklins or an ‘80s Ebro van or some other obscure thing from decades ago.
I mean, if we’re going to employ advanced artificial intelligence and image identification, it should be for really tricky problems, not identifying a 2006 Nissan Altima or a 2014 Lexus SUV.
Once the app does ID the car, it gives you a little augmented reality experience:
If you blipp any US car (built after the year 2000) — in a magazine, on the streets or even in a video — it will recognize the make, model and year and then unlock an augmented reality experience. This AR experience shows average customer rating, price, a 360 degree view of the car’s interior and more. Effectively transforming the physical world into a showroom.
And, yes, it looks like they want you to use the verb ‘blipp’ to describe this process. I wonder if they want you to add an extra little aspirated plosive at the end so everyone knows there’s that other ‘p’ there and confirming that you’re something of a dipshit.
So, sure, this is impressive, as is any thing where a computer can identify what it’s looking at, but for me, and, I suspect, a great number of our readers, something like this is going to need to be trained with a far larger dataset that actually has a chance of including some of the things we can’t just identify, anyway.
So, Blippar, call me when you can tell a Mathis from a Minerva, and we’ll talk.