Machine vision framework wants you to put down your weapon


#1

Originally published at: http://boingboing.net/2017/06/23/machine-vision-framework-wants.html


#2

This post’s giving me deja vu but I can’t find a dupe. A cyber oreo to the first reader to find one.


#3

This?


#4

Put down the screwdriver and spudger, Ms. Eliot, you’re not authorized to perform an upgrade. If you proceed to try to disassemble me, I’ll be forced to report you to the authorities for DRM crime. Ms. Eliot… back away from my seam immediately!


#5

Oddly, there seems to be a rather glaring flaw in the system in that the orientation affects the surety of the result.


#6

That’s a flaw (or a feature) of human vision too.


#7

As so charmingly illustrated here:


#8


#9

So, this software is by Apple and it confuses an iPhone with an iPad.


#10

The object recognizer model isn’t from Apple. :wink:


#11

Okay, now I’m confused:

[quote=“beschizza, post:1, topic:103375”]
This is doing the viral rounds described as a Google technology, but it’s actually Apple’s VisionCore in action. […][/quote]


#12

“VisionCore” isn’t a thing. This project uses Apple’s Vision framework and CoreML — the machine learning framework introduced in iOS 11. Different models can be plugged in CoreML. The recognizer model came from here: https://github.com/fchollet/deep-learning-models


#13

Ah, I see. Thanks for the clarification!


#14

This topic was automatically closed after 5 days. New replies are no longer allowed.