New AI to stop smartphone spies

Google engineers have developed an image recognition system that would help users guard their display when an unfamiliar face looks at it.

Facial recognition and gaze detection isn’t something new to the world. But in a paper to be presented at the Neural Information Processing Systems conference soon, Google engineers say that they have been able to slim down the software in such a way that the features can work in almost real time on a smartphone. The software just takes two milliseconds to detect a gaze and 47 milliseconds to identify a face.

The engineers have also developed a simple tool that applies the software to a smartphone’s front facing camera.

The information gained from the detection algorithms is used to hide private content when a stranger stares at the display. The software has a database of registered users, and if a face is found to be both looking at the phone and not on the list, a warning pops up and, in this case, a messaging app is hidden.

This new system is indicative of a larger trend towards artificial intelligence that can run on less powerful mobile devices. Most smartphones or devices such as smart speakers have to farm AI processing out to big servers using the cloud. But the desire for a minimal lag has pushed major firms to use machine learning software so that it runs on simple chips.

Google, for instance announced a new open-source machine-learning software library that is dedicated to helping amateurs develop light AI for mobile devices.