Facebook Developes An Advance Face Recognition Tool

Tagging photographs on Facebook will be easier than ever with a new technology that identifies faces just like humans.

 

If you're human there's a 97.53 % chance that you'd recognise the same person in two unfamiliar photographs regardless of whether the person was looking into the camera or not. On the other hand, computers haven't been anywhere as efficient as humans in recognising faces. That's until now.

 

Social media giant Facebook through its DeepFace project has achieved a 97.25 % accuracy in recognising faces. Through some smart 3D technology, the facial features of a person in one photo are mapped to be compared with another image.

 

For testing the technology, Facebook tapped into a pool of 4.4 million labelled faces from 4,030 people on its network. The system works even if the person in the photograph is not directly looking into the camera. that's a huge leap in efficiency from the facial recognition tool.

 

Instead of immediately rolling out project for use on the social media, Facebook will be presenting it at the IEEE Conference on Computer Vision and Pattern Recognition this June. The future of the project will partly depend on the feedback from the research community.

 

While the technology might have infinite uses in numerous fields, Facebook will primarily for simplifying photo tagging.


TAGS: Facebook, facial recognition, DeepFace, Artificial Intelligence