3D spherical search could accelerate YouTube queries

Posted by Lydia Leavitt

We’re so used to seeing search results presented in lists of words or pictures. 

But what if there’s a better way to search and find what you’re looking for? 



Well, one company, Telcordia Technologies, thinks the answer lies with 3D spherical search results - kind of like being able to spin query results within the palm of your hand.

By wrapping multiple frames of a particular video into a sphere, the user can quickly skim through and enlarge video frames to ensure that the video is the right one before spending time loading it. 



Ben Falchuk of Telcordia says, "The frames on the sphere are then reselected to correspond to the sub-scene of the video that the user is examining."

Oftentimes when people search for video content on a smartphone it takes a long time to load and once it’s loaded, may be the wrong video. 

This experience is time consuming and frustrating, which is why Telcordia created the sphere search technology.

With sphere search technology, a user can enlarge and view only a certain portion of a video that seems promising, thus saving him or her time and energy by only downloading about 3% of the video rather than the whole file. 

By downloading a small portion of the video, the user can determine whether or not the video is correct without loading it only to realize it’s not the right video.

Of course, the challenge for Telcordia lies in choosing what clips will appear in the 3D sphere.

Falchuk notes that the simplest approach would be to select the first frame of the video, but there’s no guarantee that this will be an accurate portrayal of what’s in the video.

Another approach would be to crowd-source in order to analyze the way users watch a video (which portions they watch, which ones they skip) to produce a "hotspots" map of clips to include in the 3D sphere. 

A digital multimedia researcher at the German Research Center for Artificial Intelligence, Adrian Ulges, adds that the 3D sphere may be beneficial for advertisers that want to make sure their ads appear on the right type of content.

But can the app scale and analyze each new piece of video content that appears on the web? 



Stats that reveal 24 hours of video is uploaded to YouTube every minute may prove this to be a daunting task.

(Via New Scientist)