Facia AI focuses on deepfake and liveness detection, aiming to provide more secure identity verification and authentication capabilities for enterprises and government agencies. Its core products are built around three main directions: liveness detection, deepfake detection, and facial recognition.
In terms of use cases, Facia's solutions can be used for liveness detection during customer digital onboarding, deepfake detection in identity verification processes, real-time deepfake identification during video calls, step-up authentication for high-risk operations, biometric enrollment and re-verification, and surveillance and access control security detection scenarios.
In terms of key capabilities, liveness detection supports active and passive modes, can defend against more than 56 spoofing attacks, still works under conditions involving hats, glasses, and low light, has a response time of less than 1 second, and has obtained iBeta Level 2 certification (0% FAR @ <1% FRR). Deepfake detection can identify AI-generated images, perform cross-platform deepfake video detection, and provide real-time media intelligence, with industry-leading accuracy. Facial recognition supports 1:1 face comparison (matching photos with ID documents), 1:N face search, image-to-video and video-to-video face matching, complies with NIST standards, and takes demographic and racial bias into account. Its latest algorithm update reduced FAR to 0.06% and FRR to 0.3%.
In terms of usage, Facia provides highly customizable integration methods, including SDKs compatible with iOS and Android (lightweight, about 5MB), cross-platform support, robust RESTful APIs, and no specific hardware requirements.
Applicable scenarios include: customer onboarding KYC processes, identity verification, video calls and confidential meetings, high-risk account operations, biometric enrollment and recovery processes, surveillance and access control, and deepfake identification in automated recruitment, among others.
In terms of limitations, the materials do not provide specific information such as pricing, free trials, deployment methods, or language support, and none is currently available.