The modern facial recognition software system, which combines two prominent technologies—YOLO and ArcFace—is expected to promote the application of AI in education, security, and smart management.
According to the research team, the system is designed with a two-stage process. First, the YOLO model is used to detect faces in images or videos in real time, offering high processing speed and effective performance under various conditions. YOLO is one of the advanced computer vision models, capable of processing dozens of frames per second, making it highly suitable for real-time surveillance applications.
After detection, the system uses ArcFace to extract facial features and perform identity matching. ArcFace is a deep learning method that represents faces as feature vectors, where images of the same person are positioned closely together, while those of different individuals are clearly separated. Thanks to the “additive angular margin” mechanism, ArcFace significantly improves facial recognition accuracy compared to traditional methods.
In addition, the development team places strong emphasis on security and privacy issues, such as data encryption and limiting the storage of raw images, to ensure user information safety. In the future, the system may be expanded with features like liveness detection and deployment on IoT devices.
The proactive research and development of advanced AI solutions by the AI Center of Tra Vinh University demonstrates the strong potential of Vietnamese educational institutions in mastering technology, contributing to the development of a domestic AI ecosystem, and moving toward practical applications that serve the community.