Subscribe to EDOM TECH Newsletter
Cyberlink
2021/08/03
CyberLink's FaceMe® Ranked Top 3 in Face Anti-spoofing Challenge at ICCV 2021
With up to 96.8% anti-spoofing accuracy rate, FaceMe® is ranked top 3 globally and no. 1 when excluding Russia and China vendors, proving FaceMe® ideal for highly secured facial recognition and eKYC solutions
CyberLink Corp. (5203.TW), a pioneer in AI and facial recognition technologies, is excited to announce its FaceMe® AI facial recognition has ranked in the Top 3 in the face anti-spoofing challenge held by the ICCV 2021. Sponsored by the IEEE (Institute of Electrical and Electronics Engineers), the ICCV (International Conference on Computer Vision) is the premier international computer vision event held every other year.
Accurate anti-spoofing technology is crucial for highly secured facial recognition deployments, such as face authentication for banking and financial services or eKYC (electronically know your customer) solutions. FaceMe®’s achievement at the ICCV challenge, places it as the most suitable solution for highly-secured facial recognition and eKYC solutions.
The ICCV 2021 anti-spoofing challenge was open to commercial vendors, research teams, and academic institutes. In previous years, the ICCV face spoofing challenge focused on 2D face prints and video-replay attacks.This year, the ICCV 2021 anti-spoofing challenge focuses on 3D high-fidelity mask attacks, which most of the existing 3D mask anti-spoofing benchmarks suffer to perform well in.
Of the 195 teams joining the development stage, only 56 made it to the final stage, and 18 performed better than the minimum criteria listed in the ICCV leaderboard. In the final result, FaceMe® reached a 96.8% anti-spoofing accuracy rate (average error rate 3.215), which is listed in the Top 3 in the ICCV 2021 challenge and is the number 1 algorithm when excluding vendors from Russia and China. Additionally, the difference between the best-ranking algorithm and FaceMe® is a small0.16% difference.
"With the increasing use of facial recognition, the risk for spoofing-attacks rises," said Dr. Jau Huang, CEO of CyberLink. "Making facial recognition more reliable and secure is one of the top priorities for the providers of this technology. CyberLink is honored that FaceMe® has performed well enough to finish on the podium in the ICCV 2021 anti-spoofing challenge. "
CyberLink's FaceMe® SDK is a cross-platform AI facial recognition engine and ranked as one of the most accurate in the world. With the comprehensive support for operating systems, including Windows, Linux (Ubuntu, RedHat, CentOS), JetPack (Jetson), iOS and Android, and the optimization for CPU, GPU, SoC, APU, and VPUs, FaceMe® provide a flexible solution to enable facial recognition across a wide range of IoT/AIoT devices, and provides developers and system integrators highly accurate and flexible facial recognition capabilities, enabling a wide range of solutions applicable across industries.
CyberLink Corp. (5203.TW), a pioneer in AI and facial recognition technologies, is excited to announce its FaceMe® AI facial recognition has ranked in the Top 3 in the face anti-spoofing challenge held by the ICCV 2021. Sponsored by the IEEE (Institute of Electrical and Electronics Engineers), the ICCV (International Conference on Computer Vision) is the premier international computer vision event held every other year.
Accurate anti-spoofing technology is crucial for highly secured facial recognition deployments, such as face authentication for banking and financial services or eKYC (electronically know your customer) solutions. FaceMe®’s achievement at the ICCV challenge, places it as the most suitable solution for highly-secured facial recognition and eKYC solutions.
The ICCV 2021 anti-spoofing challenge was open to commercial vendors, research teams, and academic institutes. In previous years, the ICCV face spoofing challenge focused on 2D face prints and video-replay attacks.This year, the ICCV 2021 anti-spoofing challenge focuses on 3D high-fidelity mask attacks, which most of the existing 3D mask anti-spoofing benchmarks suffer to perform well in.
Of the 195 teams joining the development stage, only 56 made it to the final stage, and 18 performed better than the minimum criteria listed in the ICCV leaderboard. In the final result, FaceMe® reached a 96.8% anti-spoofing accuracy rate (average error rate 3.215), which is listed in the Top 3 in the ICCV 2021 challenge and is the number 1 algorithm when excluding vendors from Russia and China. Additionally, the difference between the best-ranking algorithm and FaceMe® is a small0.16% difference.
"With the increasing use of facial recognition, the risk for spoofing-attacks rises," said Dr. Jau Huang, CEO of CyberLink. "Making facial recognition more reliable and secure is one of the top priorities for the providers of this technology. CyberLink is honored that FaceMe® has performed well enough to finish on the podium in the ICCV 2021 anti-spoofing challenge. "
CyberLink's FaceMe® SDK is a cross-platform AI facial recognition engine and ranked as one of the most accurate in the world. With the comprehensive support for operating systems, including Windows, Linux (Ubuntu, RedHat, CentOS), JetPack (Jetson), iOS and Android, and the optimization for CPU, GPU, SoC, APU, and VPUs, FaceMe® provide a flexible solution to enable facial recognition across a wide range of IoT/AIoT devices, and provides developers and system integrators highly accurate and flexible facial recognition capabilities, enabling a wide range of solutions applicable across industries.
Source: CyberLink Website
