Skip to main content

MIT algorithm can predict the (immediate) future from still images

Creating Videos of the Future
Humans still can’t predict elections but we’re pretty good at predicting the immediate future. Baby drops glass cup, cup falls and shatters, and baby starts to cry. We’re so good at these short-term forecasts that we can often even describe what events will happen next in an image.

But what’s second nature for us can prove complicated for computers. Will the glass break or bounce? Will the baby laugh or cry?

A team of researchers from the Massachusetts Institute of Technology (MIT) Computer Science and Artificial Intelligence Laboratory (CSAIL) have developed a system that can predict the following events in images and generate videos to depict them. The system needs work — its current productions are simple, short, and unassuming — but it stands out for its unique approach and accuracy.

“Instead of building up scenes frame by frame, we focus on processing the entire scene at once,” Carl Vondrick, PhD at MIT CSAIL and lead author of the paper, told Digital Trends.
video-examples-with-input-and-output

Alternative computer vision models that attempt the same task use recurrent networks to generate predictive videos on a frame-by-frame basis. The system developed by Vondrick and his team uses “convolutional networks” to generate all 32 frames simultaneously.

“The existing approach of going frame by frame has a certain logic,” Vondrick said, “but it also creates a massive margin for error. It’s sort of like a big game of ‘Telephone,” which means that the message most likely will fall apart by the time you go around the whole room.

“In contrast, our approach is the ‘Telephone’ equivalent of speaking to everyone in the room at once,” he added.

The researchers trained the system on a year of footage packed into two million videos and — in order to generate all frames at once — taught it distinguish foregrounds from backgrounds, and mobile objects from stationary ones. They then showed the system still images and had it generate short clips of subsequent events.

Once the system could generate video clips, Vondrick and his team set out to refine it through a method called adversarial learning.

“The idea behind adversarial learning is to have two neural networks compete against each other,” Vondrick said. “One network tries to decide what is real versus fake, and another tries to generate something that fools the first network.”

Through this computer competition the generative algorithm improved the accuracy of its video clips until it was able to fool human subjects 20 percent more often than a baseline model, according to a paper that will be presented next week at the Neural Information Processing Systems conference in Barcelona.

But with accuracy comes complexity and with complexity comes obstacles.

The current system’s videos are short — a mere one and a half seconds long. If the clips were much longer than that, they’d risk their consistency. “The key challenge is being able to reliably track the relationships between all of the objects in a scene … to make sure that the video that’s being generated still makes sense five or ten seconds later,” Vondrick said. To develop accurate and long videos, the system may need human input to help it grasp context and connection between seemingly unrelated actions, such as jogging and showering.

Vondrick’s ambitious end goal is to develop an algorithm that can create believable feature-length films, though he admits that is still some years off. In the near term though he thinks this system could refine AI systems by helping them adapt to unpredictable environments.

Editors' Recommendations

Dyllan Furness
Dyllan Furness is a freelance writer from Florida. He covers strange science and emerging tech for Digital Trends, focusing…
WASD and arrow keys swapped? Here’s how to fix it
Fingers on WASD on a gaming keyboard.

PC gamers are more than familiar with the WASD keyboard layout. The WASD configuration will typically route a game’s directional commands (up, down, left, and right) to the W, A, S, and D keys on a QWERTY keyboard. This arrangement allows the right hand to focus on mouse movement, while the left does all the navigating.

Once in a blue moon though, your keyboard’s WASD and arrow keys can get swapped. Fortunately, the process of switching things back to normal is easy as can be. Let’s take a look at how to fix this little layout problem, as well as what causes the swap to happen in the first place.

Read more
How to save a webpage as a PDF on desktop and mobile
A person sitting down and using a laptop.

You'll probably eventually come across a webpage that you'd like to save for later, and if you save a webpage as a PDF it will stay the same as it is today when you access it again. As a result, converting a webpage to PDF format is an excellent way to record web data as it is for our records, research, and other applications. Here's how to go about it.

Read more
How to set up an out-of-office reply in Outlook on any device
Automatic Replies in Outlook on the web on a Mac.

If you’re going to be away from work on vacation, a leave of absence, or something similar, you may want to set up out-of-office in Microsoft Outlook. This is an automatic reply that lets others know that you’re away and won’t be responding for a particular time period.

The nice thing about automatic replies like this in Outlook is that you can schedule them for the timeframe you need. If you prefer, you can also turn the feature on and off manually.

Read more