Granted by the United States Patent and Trademark Office (USPTO), the utility patent covers the invention titled “Method of Masked Face Recognition by Artificial Intelligence Technology.” The latest patent further underscores Viettel AI’s growing research capabilities and innovation in core AI technologies.

Facial recognition technology is widely used in applications such as attendance management, smart access control and device authentication. However, the widespread use of face masks in workplaces and public spaces has remained a major challenge for conventional facial recognition systems.

According to the US National Institute of Standards and Technology (NIST), face masks can increase recognition error rates by 20–50% on some facial recognition platforms. In many cases, users are required to remove their masks for identity verification, interrupting the user experience and raising hygiene and security concerns.

To address this challenge, Viettel AI developed a new AI-powered method capable of significantly improving facial recognition performance when faces are partially occluded by masks. The company said the innovation was a key factor in successfully passing the USPTO's rigorous patent examination, as masked face recognition remained a complex technical challenge when the application was filed.

Rather than relying on conventional facial recognition techniques, Viettel AI built a large-scale dataset of partially occluded faces to better train its AI models in face detection, identification and feature extraction.

The research team also enhanced its facial feature extraction model and developed a high-speed similarity search algorithm, enabling the technology to deliver reliable performance in real-world environments.

The solution allows facial recognition systems to operate effectively in applications such as attendance management, workforce timekeeping and access control without requiring users to remove their masks.

For public security and surveillance systems, the technology can also improve the identification and tracking of individuals on watchlists or security alert lists, even when a significant portion of the face is covered.

The US patent marks another milestone in Viettel AI's efforts to develop proprietary core technologies and expand its intellectual property portfolio in artificial intelligence.

In February 2026, Viettel AI was granted a utility patent by the United States Patent and Trademark Office, for its invention “Deep-learning method for multi-object tracking from video”.

According to the company, it will continue investing in research and development to further enhance the technology, broaden its commercial applications and accelerate deployment in international markets.

FPT’s Akabot AI Agentic Automation solution receives patent in Taiwan (China)

VOV.VN - The Akabot Agentic Automation plaform, developed by Vietnam’s telecom giant FPT, has recently been granted a patent by the Taiwan Intellectual Property Office for the automated solution that separates and classifies input documents using machine learning.