HANOI,
VIETNAM – Media OutReach – 20 April 2020 – On April 20, 2020, VinAI Research, a Vingroup-funded
research laboratory, announced its results of a successful study on facial
recognition technology that can identify people wearing face protective masks with
stable accuracy. VinAI Research has become one of the first units in the world
to successfully study this technology. VinAI is willing to provide the recognition
technology to the community for free to contribute to the prevention of the ongoing
COVID-19 pandemic.
VinAI’s
system uses a new and unique deep learning model developed by experts of the Institute for research and application of optimisation
(VinOptima),
to study a method with stable accuracy to identify people who may have their
faces covered.
This
deep learning model designed to reproduce activities of neural networks in the
human brain, allows training and automatically extracting valuable information
of part of the face to address the recognition problem when people use masks.
The
study shows that The VinAI’s system has significantly better
accuracy than the current technology when identify people wearing face masks.
Advantage of VinAI’s recognition technology is a different algorithm and application
of identification technology.
VinAI’s
recognition system only uses information from normal camera sensors while other
recognition technologies use infrared and depth sensors. Therefore, the cost of
VinAI’s products will not be high and can be easily integrated into existing
camera systems. The products will be widely applied into the employee time
tracking and authentication systems of businesses.
“Masks
are essential item in the COVID-19 prevention rules, but when users wear masks,
the accuracy of current facial recognition technology can be reduced by more
than 50 per cent. Therefore, VinAI has urgently developed this technology to
promptly meet social needs.” – Dr. Bui Hai Hung (Director of VinAI Research)
said.
VinAI
has also found a solution to automatically monitor users wearing masks on
cameras. This solution can actively support the management of spacing among
people in organisations and companies when they return to work after the pandemic.
VinAI
Research is cooperating with VinSmart electronic equipment manufacturing
company to install the new technology on Vsmart phones, allowing
users conveniently unlock phone by facial recognition.
The
applications on Vsmart also makes VinAI become one of the first units in the
world to apply identification technology without removing face masks on
commercial products.
Besides
the commercialisation, VinAI is ready to provide this latest technology to
partners for the purpose of serving public health during the COVID-19 pandemic.
The
achievement not only has practical significance on a global scale, but also
affirms Vingroup’s transformation into a leading technology corporation in
Vietnam which has step-by-step adapted to advanced global and complex
technologies.
References:
VinAI Research was established on April
17, 2019. In just one year, the institute has made two AI technology researches that were announced
at the world’s leading conference on AI in Canada, that made Vietnam only the
second country in the region to have projects published at this conference.
The world’s recognition
technologies: The system of facial recognition is an app that can identify people
from their photos or videos like FaceID from Apple, DeepFace from Facebook,
Rekognition from Amazon. These
technologies currently do not solve the problem of identifying people in masks.
JAKARTA, INDONESIA - Media OutReach Newswire - 22 November 2024 - VinFast Auto has officially…
SYDNEY, AUSTRALIA - Media OutReach Newswire - 22 November 2024 - The global cryptocurrency market…
HANOI, VIETNAM – Media OutReach Newswire - 22 November 2024 - By capitalizing on its…
HANGZHOU, CHINA - Media OutReach Newswire - 22 November 2024 - As the 2024 World…
BEIJING, CHINA - Media OutReach Newswire - 22 November 2024 - The 2024 Beijing Changping…
Tickets Now Available via Urbtix HONG KONG SAR - Media OutReach Newswire - 22 November…