FB pixel

Researchers detail faster 3D “dense” face alignment for biometric recognition, tracking and image restoration

Researchers detail faster 3D “dense” face alignment for biometric recognition, tracking and image restoration

Researchers from some of China’s top institutions for science and academics have published new research showing the way (and the code) for more accurate facial biometrics and tracking based on a novel application of algorithms in the 3D modeling process. The effort is said to result in a faster, accurate, and stable 3D dense face alignment.

Face alignment refers to a process for recognizing the structure of human faces in digital images. The researchers were aiming to perform operations to “CPU real-time.” Increasing accuracy and speed in computer vision systems have typically been mutually exclusive. To get faster results, the algorithms either discard some data points or use fewer steps to achieve a match. Improving biometric and object recognition accuracy typically comes at the expense of speed.

In order to identify the face, a computer is turning the face into a 3D object with tens of thousands of data points—a process called 3D face reconstruction. Creating 3D models from video streams makes the process even harder when the faces are in motion across multiple frames.

“Our promising results pave the way for real-time 3D dense face alignment in practical use and the proposed methods may improve the environment by reducing the amount of carbon dioxide released by the huge amounts of energy consumed by GPUs,” according to the researchers, who hailed from Chinese Academy of Sciences, Beihang University, and Westlake University in Hangzhou, China. The research paper describes their approach further.

Among other applications, better 3D dense face alignment could result in more accurate facial recognition of people using video cameras even when the subjects are wearing face masks. The topic is of particular urgency in China where the government has been working to keep any new outbreaks of COVID-19 at bay.

On the other hand, China has over 600 million surveillance cameras and citizens have been growing wary of abuse of these systems. According to a report in foreign policy journal “The Diplomat,” the government has made tentative steps to protect biometric data through the introduction of the Personal Information Security Specifications.

Related Posts

Article Topics

 |   |   |   |   |   |   | 

Latest Biometrics News

Comments

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Most Read This Week

Featured Company

Biometrics Research

Biometrics White Papers

Biometrics Events

Explaining Biometrics