Kneron has announced its latest 3D Binocular Facial Recognition Module, which integrates proprietary edge AI chips, lighting AI algorithms, and infrared liveness detection technologies for facial authentication in smart access control applications.
The module is designed for use in smart homes, enterprise security, and smart city deployments. It combines stereo vision cameras with Near-Infrared (NIR) imaging to generate 3D facial depth information and detect real human skin characteristics. The system defends against spoofing attacks that use photos, replay videos, and high-precision 3D masks. It has passed payment-grade verification as well as ISO 30107 and ISO 19795 anti-spoofing certification tests.
The module is powered by Kneron’s KL520 edge AI architecture, which features a reconfigurable Neural Processing Unit (NPU) for high AI inference efficiency and low power consumption. According to the company, the lightweight AI model architecture achieves over 2x higher energy efficiency (FPS/W) compared with competing solutions.
It employs a multi-model AI framework that integrates facial recognition, liveness detection, facial occlusion analysis, and head pose estimation. Through model optimization and lightweight attention mechanisms, Kneron reduced model latency by 46%, lowered parameter size by 25%, and improved inference performance by approximately 39%.
The module is optimized for battery-powered devices. It supports up to 350 days of operation using AA batteries and approximately six months with lithium-ion battery configurations, making it suitable for smart locks and always-on edge security systems.
Kneron stated that it will continue expanding its edge AI and intelligent sensing portfolio to support deployment of secure, energy-efficient AI authentication technologies in smart security and intelligent infrastructure markets.
