Graduates of the "Information Technology" faculty of the Moscow State University of Psychology and Education (MSUPE) Andrey Prokhortsev and Ksenia Shilova have developed a hardware and software complex for biometric access control. Their system uses a hybrid neural network that combines the advantages of different architectures, which significantly increases the speed and accuracy of identification.
The development allows for automatic updating of reference data (embeddings) with each successful passage, which increases recognition accuracy over time. The system supports a modular architecture, including video capture, data processing on a neural network, microcontroller management (e.g., Arduino-based turnstiles), and a web interface for the administrator.
The State Examination Commission noted that the project is ready for implementation, which is rare for student works. The complex can be used in private homes, offices, and enterprises, providing security and flexible configuration.
Read more on the topic:
AI from NtechLab will help Muscovites find items lost in transport or parks