Old photos in a box

Maybe you have a photo with a substantial hole in it that you want to fill in, but the section mission is a complex background. This deep learning tool offers an impressive, albeit not perfect, cleanup patch.

What are the implications? People can more easily restore pictures, and photo manipulation becomes easier.

Photo by Mr Cup / Fabien Barral on Unsplash

You saw the movie when you were nine years old, but you know it showed a young boy alone at sea adrift on a raft. You just cannot remember the name of it for the life of you, and you really want to find it. You’ve mentioned it to several people, but nobody’s heard of it, or maybe they vaguely remember the plot but can’t help you with the name. Now there’s an AI that might help. You can search its video database to find that illusive movie by typing in specific objects.
On the other hand, if you’re a movie maker and you want to have your videos indexed by this company, you can sign up to send your videos through their analyzer for free, at least for now. Then your movies would pop up as people search the site.

Photo by Seb on Unsplash

This augmented reality (AR) company uses deep learning to pull Uma Thurman and John Travolta from Pulp Fiction and project 3D versions of them into a living room (see video below). While we think this tech’s current form has a lot of glitches to work through before it’s flawless, it’s still brilliant, and it gives us a unique view into our possible future: one day you’ll be able to have anybody from a video hanging out in your living room in 3D.

Photo by Jason Briscoe on Unsplash

Your birthday comes around reliably every year, same day. It tells you how “old” you are, but it doesn’t tell you how biologically old you are. There’s a difference in geroscience. Your birthday may numerically bump you up every year, but if you’re, say, really fit, your biological age may be lower than your traditional age. Essentially, your body isn’t deteriorating as quickly as other peoples’ born the same year as you.

There are several different biomarkers that may indicate your biological age, and scientists are using them to develop biological aging clocks that predict this age. Deep learning helped create an aging clock that more accurately predicts this age across multiple populations—which means it created a clock that can be used around the globe, not just locally like other aging clocks.

Photo by Sonja Langford on Unsplash

Say you’re a law enforcement officer who is searching surveillance videos for a suspicious black car. You could watch hundreds of hours of surveillance video, or you could simply use a video search engine called EllaElla employs deep learning to easily detect objects and find you every frame that has a black car in it. The impressive thing here is that you can use normal search language, like you would in Google search, to find important footage quickly.

Photo by Nathaniel dahan on Unsplash

While the ship date for this new AI application is delayed until April 2018 (or longer), Arsenal, a deep-learning photography assistant, will likely revolutionize advanced picture taking, at least for photography newbies. It simplifies the process of choosing control settings, like shutter speed, aperture, and ISO, on a professional camera. Funded on Kickstarter.com last June, it’s hit a snag in getting an Apple certificate, but it hopes to be ready to ship soon.

Photo by Alexander Wang on Unsplash

You feel the room shaking and ask yourself, “is that a truck going by?” But you start to realize it’s an earthquake. Before long, it’s over and you’re safe, but shaken. You keep thinking it would have been really helpful to get a notification before it struck.

What’s coming down the pipeline is an early-warning deep-learning system, created by The National Center for Research on Earthquake Engineering under National Applied Research Laboratories (NARLabs), that aims to send you earlier earthquake warnings. It can predict, within three seconds, the magnitudes of oncoming earthquakes by detecting microseismic P-waves—high-speed waves people can’t feel but machines can detect.

Playing the piano requires complex and dexterous finger movements, controlled by the musician’s brain. After losing his arm, musician Jason Barnes thought he would never play two-handed again—until researchers at Georgia Tech developed a prosthetic arm that allowed him to do so.
It uses a deep neural network trained to sense, through ultrasound, the muscle movements in the arm and predicts which finger the musician wants to move.