Facebook has finally improved its AI mechanism and unveiled its personal digital assistant that enhances your video watching experience tremendously.
According to Facebook’s Artificial Intelligence Research (FAIR pdf) computer vision team head Rob Fergus, an image recognition program for videos uploaded on Facebook is under development.
This program won’t just monitor but explain whatever is happening in a video along with automatically tagging people on photos.
[must url=”https://www.hackread.com/facebook-dislike-button-scam/”]Scammers Targeting Facebook Users with Dislike Button Scam[/must]
Facebook had been facing a problem with videos lately since most of the posted visual content was lost in the shuffle due to the absence of metadata, which means descriptive text, accompanying the images.
However, with this new AI enabled program videos would get tagged automatically and the program will also narrate the happenings of the video. The prototype of this new AI program is being tested currently, reports Popular Science.
The video uploaded below explains how this new AI system will watch and tag short scenes in the video.
Here’s how it’s done:
[must url=”https://www.hackread.com/facebook-friends-psychologically-harmed/”]Facebook Users With Lots of Friends Could Be Psychologically Harmed, Study[/must]
The benefits of such a program are various and diverse since it will prevent users from uploading offensive content such as porn, and uploading someone else’s content as their own.
It is also expected that this program will help in tracking viral news events and in identifying numerous video genres such as animals or sports.
Getting an AI program to watch and recognize videos is undoubtedly a major feat as far as social media networks are concerned.
This new function would certainly make finding similar videos on a variety of platforms such as Instagram easier.
In June 2015, Facebook’s AI/artificial intelligence lab also introduced an algorithm at the Computer Vision and Pattern Recognition conference earlier that can locate you in a picture with 83% accuracy.
