FaceMe® SDK, a cross-platform facial recognition engine, is uniquely positioned to integrate edge-based AI facial recognition into a wide range of IoT and AIoT solutions. FaceMe® SDK is optimized to run on most hardware configurations, from high-end workstations to low-power chipsets typically used in IoT devices. It is compatible with Windows, Linux, Android and iOS systems.
FaceMe® SDK is top ranked in both NIST FRTE 1:1 and 1:N. In 1:1, FaceMe® provides up to 99.83% TAR (with FMR at 1E-6), which is ideal for use cases like eKYC and 2-factor authentication. In 1:N, FaceMe® reached 99.73% TAR in VISA Border investigation mode, ideal for use cases like security and surveillance.
NIST FRTE 1:1
1:1 Face Verification
False Match Rate (FMR)
0.000001
(1 in 1 Million)
True Acceptance Rate (TAR)
99.83
%
NIST FRTE 1:N
1:N Face Search
Database Size
1.6
Million People
True Acceptance Rate (TAR)
99.73
%
NIST FATE PAD
Passive 2D Anti-spoofing
True Rejection Rate (TRR)
100
%
True Acceptance Rate (TAR)
99
%
CyberLink’s accurate and reliable AI facial recognition engine, FaceMe®, can be leveraged and deployed to a variety of IoT devices including high-end workstations, PC’s, mobile devices, digital signage, access control devices, POS systems, and kiosks. It’s versatile and supports major operating systems, AI inference engines, and multiple edge computing platforms. FaceMe® SDK processing speeds can also be accelerated with NVIDIA RTX GPU’s.
Based on the test by NIST's Face Recognition Technology Evaluation (FRTE), FaceMe® achieves a 99.83% True Acceptance Rate (TAR) at an ultra-low one-in-a-million (1E-6) false acceptance rate (FAR). It ranks among the most precise and fastest facial recognition engines in the NIST FRTE.
FaceMe® demonstrated outstanding age estimation accuracy with a Mean Absolute Error (MAE) value of 3.282 in NIST's latest Face Analysis Technology Evaluation (FATE) Age Estimation & Verification.
Multiple anti-spoofing technologies provide highly secure, accurate liveness detection to protect against biometric fraud, e.g. print attacks, replay attacks, and deepfakes. FaceMe® supports anti-spoofing with 3D depth-of-field, smartphone and tablet 2D, and also IR+RGB infrared camera lenses.
FaceMe®'s updated algorithm provides the ability to accurately recognize a person's identity with up to 98.21% accuracy rate even if they are wearing a mask. It enables instantaneous and hygienic touchless biometric recognition.
Detect person attributes such as age, gender, and clothing without face recognition. This allows for the search and tracking of specific individuals, such as lost persons or suspicious individuals.
Facial recognition can enhance security, identify potentially fraudulent activities and help loss prevention. FaceMe®’s cross-platform SDK and flexible architecture enable an easy deployment in any environment. For more information on the latest trends in facial recognition applications.
FaceMe® supports a variety of operating systems and AI inference engines. It is designed to run on mainstream processors and is optimized to take advantage of GPU / VPU hardware acceleration. FaceMe®’s flexible and scalable architecture provides hardware and software developers the tools to create customized facial recognition solutions and experiences.