S u m m a r y :
You can now have 3D selfies, thanks to a new technology developed by researchers from the University of Nottingham and Kingston University. The findings will be presented at the International Conference on Computer Vision (ICCV) 2017 in Venice in October.
Selfies: From 2D To 3D
The world of technology has witnessed tremendous headway in the last few years, and now, a recent endeavour by a team of computer scientists promises to further revolutionise vision and graphics research. They have put together a web app that can turn a single 2D image into a 3D selfie in the form of 3D facial reconstruction. You only have to upload a colour image of your face to generate its 3D model. Good news, you can also do it!
Where to Get a 3D Selfie
The researchers have put the technology at the disposal of the public on their website. Until now, over 400,000 users have already tried it. Just take a selfie and upload it here.
The 3D model allows you to see the shape of the face in so many angles. What are you waiting for?! Check this out to view one of the images already available on the website.
3D Selfies With Double Chins!
The brains behind this unprecedented opportunity are PhD students Aaron Jackson and Adrian Bulat. The pair used a form of artificial neural network known as Convolutional Neural Network (CNN) which is based on machine learning that allows computers to learn without explicit programming. Their CNN was fed with data comprising 2D pictures and 3D facial models, and it used this information to reconstruct 3D facial geometry from 2D images. Furthermore, the CNN could also produce parts of the face that were not conspicuous. So, yes, you can check out your double chin!
“The method can be used to reconstruct the whole 3D facial geometry including the non-visible parts of the face,” says Bulat.
Variety of Facial Expressions
What makes this research special is that unlike other systems that attempt to do the same, it needs only one 2D image. As for previous ones, they not only require several images of the face but they also have to deal with difficulties dealing with expressions and non-uniform illumination in terms of ascertaining which parts of the face correspond with which parts.
“Our CNN uses just a single 2D facial image, and works for arbitrary facial poses (e.g. front or profile images) and facial expressions (e.g. smiling),” explains Jackson.
Applications of the 3D Selfie
The technology can be used as face and emotion recognition. It can help personalise video games by bringing a sense of reality to them. It can also allow people to try on accessories like glasses online.
Moreover, it is hoped that it could have medical applications as well, such as providing a deeper understanding of conditions like depression and autism, and predicting the outcomes of plastic surgery.