Sep 19, 2017 | By Tess

A team of researchers from the University of Nottingham and Kingston University in the UK have developed a new algorithm that can transform a 2D photo of a face into a pretty accurate 3D image.

Typically, to generate a 3D model of a person’s face, one would have to use some sort of 3D scanning or photogrammetry technology to capture images of the subject from various angles. This technology, as we see on a regular basis, is quite accessible and is offered through a range of smartphone apps as well as standalone devices.

Still, being able to turn a single two-dimensional image into a 3D model is something that has remained difficult for developers and challenged researchers in many fields. As the Nottingham and Kingston researchers explain: “3D face reconstruction is a fundamental Computer Vision problem of extraordinary difficulty.”

That is why, despite how rudimentary the graphics may seem, this new artificial intelligence (AI) algorithm is actually pretty exciting.

Basically, the researchers managed to build their algorithm by “training” a Convolutional Neural Network (CNN) with data sets of 2D facial images and 3D scans of the same faces. This enabled the program to effectively learn how to construct a 3D model from a flat image.

Now, the software can read any 2D face image (provided it’s forward-facing) and use its knowledge to generate a protruding 3D model of the face.

Those interested in the software can create their own 3D face models here. More than just a fun tool to make 3D selfies with, however, the algorithm could have important applications in virtual reality, as it could be used to create realistic avatars, or in the cosmetic industry to virtually test makeup, for instance.

Other research projects dedicated to transforming 2D images into 3D models are currently underway at a number of universities. A team from the Berkeley Artificial Intelligence Research lab, for example, is developing a way to generate 3D objects from simplistic 2D color images.

Their software, which has been geared towards shapes rather than detailed faces, uses a technique called “hierarchical surface prediction” to map out a shape’s boundaries and possible dimensions.

For now though, I have to say I am content to keep plugging in various people’s faces into the 3D face reconstruction tool developed by the joint UK team. After all, who doesn’t want to see their own face transformed into a strangely unsettling 3D pop-out?

 

 

Posted in 3D Software

 

 

Maybe you also like:


   


opal wrote at 9/20/2017 11:10:27 AM:

Great words are written in this article. Thanks a lot for Sharing



Leave a comment:

Your Name:

 


Subscribe us to

3ders.org Feeds 3ders.org twitter 3ders.org facebook   

About 3Ders.org

3Ders.org provides the latest news about 3D printing technology and 3D printers. We are now seven years old and have around 1.5 million unique visitors per month.

News Archive