I’m a big fan of Apple’s on-device machine learning approach and it’s associated privacy benefits:
However, due to Apple’s strong commitment to user privacy, we couldn’t use iCloud servers for computer vision computations. Every photo and video sent to iCloud Photo Library is encrypted on the device before it is sent to cloud storage, and can only be decrypted by devices that are registered with the iCloud account. Therefore, to bring deep learning based computer vision solutions to our customers, we had to address directly the challenges of getting deep learning algorithms running on iPhone.
I wish other vendors would use a similar strategy, but other business models make for other priorities.
Back to the subject at hand, though: like the previous article I posted from Apple’s Machine Learning Journal, this is a little light on details. Good look creating a reimplementation of this based on the provided information. It’s still a reasonably interesting read, though. I’m hoping that there will be a similar post about Face ID at some point in the not too distant future, since I can only imagine that this involves actually training a neural network on the device.