The use of face masks is certainly not new, but as a result of the COVID-19 pandemic, many more people around the world are using them for the first time. This has created some interesting issues in the technology world. People covering half their faces with masks comes at a time when many phones are switching to facial recognition for security. Sensory is one company working on a solution to this problem.
Sensory is a Silicon Valley-based company that creates AI products, some of which are used by Android smartphone OEMs. We’ve talked about Sensory’s TrulySecure facial recognition in the past and the company also works on hotword detection and natural language processing. Sensory announced that its TrulySecure face and voice platform has been adapted to meet the demands of the “new normal.”
TrulySecure facial recognition can now recognize users while wearing masks and it can also detect coughs and sneezes. It does this by utilizing biometric fusion and multiple enrollments to avoid the limitations of typical facial recognition. Biometric Fusions combines both face and voice into one SDK, so when faces are partially obstructed, the voice biometric can provide added confidence and vice versa.
Multiple Enrollments allow users to enroll alternate appearances. Users can enroll their typical mask-free face and then add a second enrollment with a face mask on. Many software solutions don’t allow users to enroll their face if it’s partially blocked. Allowing users to enroll both mask on and off allows the software to apply the appropriate template.
Sensory is not the only company working on a solution to this problem. We previously wrote about Huami and its development of N95 face masks that don’t interfere with face unlock. Wearing face maks is likely to be common practice even after the COVID-19 pandemic peters out, so this is something we’ll likely continue to see tech companies work on.
The post Sensory’s new biometric tech aims to make face unlock work with face masks appeared first on xda-developers.
from xda-developers https://ift.tt/3eLtIU7
via IFTTT
Aucun commentaire:
Enregistrer un commentaire