Selfies - are you fond of them? Chances are you are going to like it even more after knowing about this latest development. Scientists have recently developed an artificial intelligence (AI) system, which is capable of turning a simple 2D selfie of anyone's face into a three-dimensional (3D) model within seconds, reported DNA.
The researchers used a Convolutional Neural Network (CNN) to develop the system. CNN is an arm of AI, which makes use of machine learning to make computers learn without being explicitly programmed.
Researchers at the University of Nottingham and in the UK, while developing this system, fed a CNN with a huge pool of dataset that comprised of thousands of 2D pictures and 3D facial models. After being trained with all the information, the CNN was able to reconstruct 3D facial geometry from a single 2D image. Other than that, the system is also good with guessing the appearance of the non-visible parts of a face. It works on a single 2D image and creates more arbitrary facial poses and expressions.
"We came up with the idea of training a big neural network on 80,000 faces to directly learn to output the 3D facial geometry from a single 2D image," said Georgios Tzimiropoulos, Assistant Professor at the University of Nottingham, according to the media.
This CNN can have various applications. Apart from being used to recognize faces and emotions, it can be worked upon to personalise computer games and improve augmented reality. As per the researchers, this system can have a rather interesting use - people can use it to try on accessories, such as glasses, online.
Medical applications of the CNN can be quite significant too. It can help doctors understand certain medical conditions such as autism and depression. Doctors can also make use of the technology to replicate the results of plastic surgery, added the publication.