Facebook’s Facial Recognition System Has Reached Near Human Accuracy

Facebook’s facial recognition technology which allows the social network to identify and guess people in photos has actually come very close to human levels of recognition. The AI project dubbed DeepFace can recognize the same face from different photographs and get it correct 97.25% of the times. This is eerily close to the 97.53% accuracy score registered by humans!

The new tech uses Deep Learning, a name given to the process of using networks of simulated neurons to find and recognize patterns in huge volumes of data. This is principally similar to how our brain works but is undoubtedly rudimentary when compared to the actual working of the brain.

DeepFace searches an image for a face. It then changes the angle of the face to make it forward facing. Then the simulated neurons study the front face and assign a numerical value to the arrangement of various facial features. If DeepFace finds the same numerical value in a completely unrelated photo then it recognizes the face as being the same from the previous photo, thus giving us a more accurate way to identify faces.

Facebook beta tested the software by letting it access 4 million photos belonging to 4000 users in its database. Even though this is meant to be a research project, it does have a lot of data which it can use to become smarter at facial recognition. In 2011, the company already deployed facial recognition in its services, by guessing a person in an untagged photo. This has been met with quite a bit of resistance from users due to perceived privacy invasion from the company. Facebook has also been criticized for not accurately informing its users about facial recognition.

The software has gotten much better now and if integrated with Facebook’s platform, could give the social network a much bigger advantage in identifying people in its ever increasing media database. How will users react to this news?

Technology which approaches human levels of performance is inevitable as we keep progressing in time, but there are ramifications involved with such features being deployed on the world’s largest social network. Allegations of government’s spying on online data makes it even more difficult to see such developments in a positive light.

What are your thoughts on this? Share them with us in the comments.

Link | DeepFace

Source | MIT Technology Review

Looking For A Social Media Agency?? – Contact WATConsult – India’s Leading Social Media Agency