No episode of Law & Order (and its many offshoots) is complete without a scene where a tech-savvy detective uses the trusty "enhance" button to unmask this week's perp.
Now Google is giving you superhero powers (kind of), in the form of its own Enhance function.
Look sharp
The entity behind the breakthrough is none other than Google Brain, Google's deep learning project (and part-time composer of songs).
Google Brain's engineers have devised a way to clean up the resolution in very low-res photos, making it easier to identify people and objects that were previously unrecognisable.
It employs an algorithm that pulls info from two neural networks (ResNet and PixelCNN, in case you were wondering), to essentially take educated guesses around what information should be supplied to convert grainy pics into something easier on the eyes.
Keeping it real
While the process is still being tweaked, Google Brain's manufactured images of celebrities managed to fool human subjects 10% of the time. Images of a bedroom hoodwinked humans 28%.
While there's obvious benefits for law enforcement officials (and armchair detectives), it's an interesting look at how pre-digital and obscure content might be resurrected when the tech gets smart enough.
RECOMMENDED FOR YOU
Google Once Again Delays Ad Cookie Cull
Google has announced another postponement in the removal of…
Google has announced another postponement in the removal of…
Meta Bets Big on Meta AI
Meta has recently amplified its role in the generative…
Meta has recently amplified its role in the generative…
LinkedIn Trials Premium Company Page Business Subscription
LinkedIn is looking to follow Meta’s lead as it…
LinkedIn is looking to follow Meta’s lead as it…