Scientists from the National Research University of Technology MISIS, together with experts from the National Research University Higher School of Economics and the AIRI Artificial Intelligence Institute, have developed a system that automatically selects the most effective neural networks for facial recognition services (Face ID). Home that speeds up the operation of Face ID on smartphones, tablets and smart devices. The press service of NUST MISIS reported this to socialbites.ca.
Face ID services usually consist of several layers of neural networks that are responsible for analyzing various image parameters. The NUST MISIS system is software that accelerates and automates the selection of Face ID components to ultimately achieve the most effective recognition service. When choosing Face ID components, Russian development is based on the technical characteristics of each device on which it is launched. Face ID systems selected in this way work within 5-10 milliseconds even on the weakest devices.
The main potential benefit of the NUST MISIS program is that it can accelerate the development and deployment of facial recognition systems in organizations whose technology park consists of devices with different computing powers.
“Imagine you have a community of a million tablets on which you need to install a module for facial recognition. They have certain microchips and certain technical capabilities. “You can unbox a device and install our application, which will determine the most suitable model specifically for this tablet model and recognize faces in 5, 10 or 20 milliseconds,” he said. deputy director of the center socialbites.ca artificial intelligence NUST MISIS Andrey Savchenko.
NUST MISIS also added that manually selecting the neural network components required for Face ID is time-consuming and does not guarantee correct operation of the service on all devices.
The program of Russian scientists was developed based on open source code and can be downloaded from the GitHub portal. NUST MISIS technology is described in more detail in an article published in the scientific journal IEEE Access.
Lavrentiev, who was previously an expert on neuromorphic technologies at Kaspersky Lab said “socialbites.ca” about artificial intelligence of the future.