Optimized for Intel Core Ultra Series 3 processors, FaceMe delivers up to 20 times the processing speed with iGPU acceleration in internal tests

TAIPEI, TAIWAN – [November 10, 2026] – CyberLink Corp. (5203.TW), a global leader in AI facial recognition technology, today announced that its FaceMe® facial recognition technology has been optimized for Intel Core Ultra Series 3 processors (codenamed Panther Lake). Leveraging Intel integrated graphics (iGPU) acceleration with OpenVINO™, FaceMe® SDK significantly increases facial recognition performance on edge devices. This enables businesses to deploy high-performance facial recognition on compact hardware without requiring a discrete graphics card. 

In CyberLink’s internal tests, FaceMe SDK running on an Intel Core Ultra X7 358H processor with iGPU acceleration delivered approximately 20 times the processing performance of CPU-only operation on the same processor and 37 times that of a 12th Gen Intel® Core™ i5-1235U processor. In a separate product-level test, FaceMe Security, CyberLink’s all-in-one security solution powered by the optimized engine, simultaneously supported blocklist alert monitoring across up to 59 camera channels on a single mini PC.

FaceMe SDK Delivers Faster Facial Recognition on Intel Platforms 

FaceMe SDK provides developers with a high-performance facial recognition engine and development tools for building real-time recognition applications. Its accuracy has been independently evaluated by the National Institute of Standards and Technology (NIST), where FaceMe achieved a 99.83% True Accept Rate (TAR) at a False Accept Rate (FAR) of one in a million (1E-6) in the Face Recognition Technology Evaluation (FRTE) 1:1. Building on this performance, CyberLink continues to optimize FaceMe for edge computing and expand hardware compatibility, giving developers more flexibility to run facial recognition directly on edge devices. 

In CyberLink’s internal performance tests, the Intel-optimized FaceMe SDK, using the Ultra High Accuracy (UH10) model, processed 1080p video at 217.9 frames per second (FPS) using iGPU acceleration on the Intel Core Ultra X7 358H, compared with 11.1 FPS in CPU-only mode, delivering approximately 20 times the processing performance. Compared with the 12th-generation Intel Core i5-1235U, which achieved 5.9 FPS in CPU-only mode, performance was approximately 37 times higher. 

This increase in processing capacity allows systems to analyze more video frames and handle greater workloads in the same amount of time, while helping developers and system integrators reduce the time required for performance tuning and hardware integration. For many edge deployments, Intel integrated graphics can provide sufficient processing power without the need for a discrete GPU, simplifying system design, on-site installation, and ongoing maintenance while helping lower overall deployment costs. 

FaceMe Security Supports Up to 59 Camera Channels on a Single Mini PC  

FaceMe Security is an all-in-one solution for video surveillance and access control. By integrating the Intel-optimized FaceMe SDK, it gives system integrators and enterprises greater flexibility in deploying and managing multi-camera surveillance and access control applications. In CyberLink’s tests with FaceMe Security 8.9.2, an ASUS NUC 16 Pro mini PC equipped with an Intel Core Ultra X7 358H processor and Intel Arc B390 integrated graphics supported simultaneous video processing from up to 59 camera channels in a blocklist alert scenario and up to 49 camera channels in an access control scenario. 

These results demonstrate how high-performance facial recognition can run on compact hardware, making it easier to deploy in office buildings, factories, retail stores, and multi-site facilities while reducing hardware complexity and equipment space requirements. FaceMe Security also supports employee and visitor identity verification and access control, allowing organizations to select hardware configurations based on application requirements, visitor traffic, and deployment scale.  

“As businesses expand their use of facial recognition, they must balance performance, deployment space, and operating costs,” said Mei Guu, President of CyberLink Business Unit 2. “By optimizing FaceMe for Intel platforms, we enable organizations to run facial recognition locally on more compact hardware, reduce reliance on cloud computing, and build flexible system architectures that meet their needs. We will continue expanding FaceMe’s support for edge AI hardware platforms, helping developers and system integrators bring high-performance facial recognition to more commercial environments.” 

・For more information about FaceMe optimization and deployment on Intel platforms, please visit: https://builders.intel.com/solutionslibrary/edge-ai-facial-recognition-for-trusted-identity-verification

・For more information about FaceMe SDK, please visit: https://www.cyberlink.com/faceme/sdk/overview 

・For more information about FaceMe Security, please visit: https://www.cyberlink.com/faceme/solution/security/overview

About CyberLink

Founded in 1996, CyberLink Corp. (5203.TW) is a leading provider of digital creative software and AI solutions. CyberLink's portfolio spans consumer and commercial applications for video, photo, and content creation, as well as AI-powered facial recognition technology. Its award-winning products include PowerDirector, PhotoDirector, Promeo, and FaceMe®. CyberLink is committed to making advanced technology accessible through intuitive, powerful solutions that help creators, businesses, and organizations work more efficiently and creatively.

For more on Director Suite, visit: https://www.cyberlink.com/products/director-suite/overview_en_US.html 

Press Contact

press@cyberlink.com