- The new development can be used where there are passageways to monitor people and identify those who are not allowed access to the facility.
- Masks created a challenge for the facial recognition.
- The new version of the algorithm allows you to pass up to 30-40 people per minute through the turnstile: there is no need to stop in front of the camera.
The engineers of the Novosibirsk State Technical University (NSTU) have developed a facial recognition system that can distinguish people even with medical masks on them. The new development can be used where there are passageways to monitor people and identify those who are not allowed access to a facility.
Egor Bukhamer, one of the project developers, and an engineer at the computer vision and augmented reality laboratory of NSTU NATI said:
“The data center uses powerful video cards from 24 GB, such as Nvidia Tesla. They are made to work with neural networks, and their cost starts from 300,000 rubles [$4,084.50]. The engineers used regular processors that cost from 14,000 rubles [$189.93]. Everything is optimized according to processor instructions. This greatly expands the prospect of implementing the system. The new version of the algorithm allows you to pass up to 30-40 people per minute through the turnstile: there is no need to stop in front of the camera. The speed of passage will depend only on the capacity of the turnstiles.”
Moreover, to recognize people, the system uses a special algorithm that analyzes key points in the upper part of the face, which is not hidden by a mask, and compares it with biometric data in the system’s database.
Ivan Tomilov, another one of the authors of the project, associate Professor of the Department of automated control systems of the faculty of automation and computer engineering of NSTU NETI said:
“We create a database to determine whether a person has access to the enterprise or not. The database stores not images of faces, but their digitized values: the distances between key points on the face. The database is formed as follows: all employees of the organization take photos from different angles and upload their biometric data. If someone comes whose data is not in the database, the system will show it.”
Currently, the development of the Russian scientists is in the testing phase, following which a decision will be made on the introduction of a new facial recognition system in mass operation at those enterprises that will be interested in this.
The project is designed for two years. Now the development is at the stage of pilot operation. Subsidies from the regional government received in 2019 were allocated for its implementation, according to the University’s website.
Overall, wearing masks created challenges for many facial recognition systems. Even developers of Apple had to tweak the system, since using facial recognition created challenges with the mask wearing due to the Coronavirus.
The Face ID system detects the mask on the user’s face, and in this case, speeds up the transition to unlocking with a password. Previously, in this situation, the user spent at least a few seconds before the system issued an interface for entering a password. Now the transition is very fast.
However, it still doesn’t solve the issue to actually have face recognition immediately unlock on the iPhone. Hence, new Russian technology, if successful could be utilized in the many areas specializing in the security. The new technology can also be utilized by law enforcement and other government entities.