Facebook IT & Web-tech News Research

Facebook’s Image Detection Technology Gets Better, Can Now Differentiate Between An Apple And An Orange

facebook Signal, Facebook
Share on Facebook
Tweet about this on TwitterShare on Google+Share on StumbleUponShare on LinkedInPin on PinterestShare on Reddit

Image Detection is increasingly becoming the next it technology with major corporations vying to get the best of each other. Today, Facebook announced yet another milestone when its Chief Technology Officer, Mike Schroepfe talked about recent improvements in its image detection technology.

Teaching computers to be able to detect and differentiate between objects — to train them to understand what the patterns in the pixels mean — is something the Facebook AI Research (FAIR) team has been working on for the last year. They’ve made huge progress in a short time. FAIR has been able to create more than a 60 percent improvement in its object detection and segmentation technology in a year

Impressive. In fact, fairly speaking FAIR is one of the teams that have made the most substantial progress in the field. The team bagged the second spot in Microsoft Common Objects in Context (COCO) image recognition competition 2015, close on the heels of Microsoft Research itself, which took the first.

Of course, the work is immensely complicated. After all, while picking out an Apple from an orange would be a piece of cake for a human, the computer finds its — hyperfast when it comes to mathematical calculations — processors blunted when faced with the daunting task of putting pixels together to form fruits. I mean, really, how different is one colored collection of dots from another? And what about Apple and cherries? What is to prevent the system from branding the cherry as a small Apple?

923497_10153811271039443_5994720034472359771_n

Despite all these difficulties, considerable progress has been made in the field and Facebook is quite hopeful.

The use of such a technology would ofcourse be, leigon. In fact, i can think up a couple of hundred right now, courtsy Sci-fi movies and novels. However, Facebook is currently developing it with an eye upon subsequent integration with its own apps and systems.

By enabling computers to recognize objects in photos,it will be easier to search for “the picture of the fruit bowl” without you having to explicitly tag eachphoto you upload.

And that is just the beginning. Using this image detection system, people with vision loss would be able to get a better experience with the system detailing each and every picture for them, while also enabling the social networking website to make intelligent decisions about what you want to see on your newsfeed — At least that is whats currently on Facebook’s plans.


 

A bibliophile and a business enthusiast.

[email protected]


Add Comment

Click here to post a comment

Your email address will not be published. Required fields are marked *