New version of Sensory TrulySecure supports 3D cameras and cloud-based biometric authentication
Sensory has released the fourth generation of its TrulySecure FIDO Certified multimodal biometric authentication technology, featuring a variety of new AI features for smartphones, mobile apps, and IoT devices.
The TrulySecure 4.0 SDK is smarter, faster, and more secure than previous versions, according to the announcement, and includes support for 3D and IR cameras, authentication processing using a GPU, new liveness challenges, and options for integrating TrulySecure components into cloud-based systems.
The data set used to train TrulySecure’s deep learning AI is an order of magnitude larger than the previous generation, improving 2D facial recognition accuracy by more than 50 percent, the company says. The addition of 3D and IR camera support makes the technology applicable to a broader set of use cases, and makes it more resistant to spoof attacks. The 3D camera capabilities have been developed in collaboration with Germany-based pmd Technologies, which supplies 3D Time-of-Flight CMOS-based digital imaging technology currently used in smartphones, robots, cars, and AR headsets.
TrulySecure now provides the option to split the authentication processing load between the GPU and applications processor, allowing the applications processor to prioritize other tasks, or reduce the amount of processing time needed by half. Isolating and streamlining the intensive components of the algorithm also presents the possibility of porting to other embedded environments in the future.
When used with 3D cameras, Sensory says TrulySecure provides a major leap forward in anti-spoofing and liveness detection. With traditional cameras, a new face liveness challenge with movement prompts which makes spoofing virtually impossible, according to the announcement, and the new version has also added an “eyes open” challenge.
For cloud-based deployments, TrulySecure provides fast on-device enrollment for federated authentication, which creates an enrollment template TrulySecure transmits to the authentication server to be pushed out to other edge devices as needed. Pre-authentication steps such as liveness detection and feature extraction can be performed at the edge to reduce the data loads created by authentication.