The augmented reality specialist Faceunity Technology has released an updated version of its AR SDK. The enhanced SDK comes with improved algorithms that are better able to track someone’s face and body movements, which in turn allows the AR platform to map special effects onto the subject of a video in a more seamless fashion.
In that regard, the Faceunity SDK is able to recognize as many as 241 facial landmarks, and is especially sensitive to landmarks around the eyes and mouth. It can similarly recognize 56 different kinds of facial expressions, and 25 key landmarks on the human body. It also supports gesture recognition, which allows users to interact with the platform and trigger certain effects with basic motions.
Other features of the SDK include portrait head and hair segmentation, and 3D face model reconstruction. Those technologies allow Faceunity to track the position of someone’s face and body in real time, and ensures that any effects that get superimposed on that body are mapped accurately with well-defined borders. On that front, Faceunity’s portfolio includes a wide range of beautification and reshaping features, as well as props, stickers, image matting, and background segmentation utilities.
Businesses that want a branded look can use the open-source FU Creator to create their own custom filters, stickers, and effects. The AR SDK is compatible with the iOS, Android, PC, Mac, and Unity platforms, and can be integrated directly into various applications.
It is worth nothing that privacy advocates have raised concerns about AR technologies in recent years, largely because people who do not want their images captured can easily get caught in the background when people use AR cameras in public. Law enforcement agencies and military organizations have also expressed interest in the technology, though Faceunity’s portfolio is more geared toward the public (and social media and gaming uses cases in particular).
(Originally posted on FindBiometrics)