Hyper-realistic technology creates avatars from a single photo

The Research Institute on Electronics and Telecommunications (ETRI) has developed a hyper-realistic AI technology that can create an avatar that naturally speaks as a real person using a single portrait photo.

Technology is considered a new generation interface that allows an intuitive interaction between vehicles and humans in preparation for the era of fully autonomous driving, and should propagate in the digital human industry.

While traditional AI assistants focused on speaking in office environments or navigation systems In vehicles, is limited to simply commands, the hyper-realistic IA avatars of Etri have sophisticated facial expressions And the movements of the mouth allow natural and human conversations.

This allows human-machine interaction more centered on humans, such as a driver of the vehicle AI speaking to the driver or interacting with pedestrians.

The heart of this technology is a unique algorithm which, unlike traditional generative AI, selectively learns and synthesizes parts of the face which are directly linked to the statement, such as lips and chin. This approach reduces the learning of useless information and allows more sophisticated facial expressions, including mouth forms, teeth and skin wrinkles.

Etri explained that technology has shown higher performance in terms of synthetic visual quality and lip synchronization precision, as presented in the main international conferences such as CVPR and AAAI.

In addition to autonomous vehicles, this technology can be used in various industries such as kiosks, banking meters, news presentations, advertising models and should stimulate innovation in AI digital human industry.

The ETRI mobility research research section is currently focusing on human-machine interaction technologies (HMI), and also develops pilot interface technologies based on AI that analyze the emotions of the pilot and pedestrians, fatigue, concentration, etc.

Daesub Yoon, director of the research section on the experience of mobility users, said: “As mobility technology becomes more advanced, the elderly and socially disadvantaged could be marginalized. We hope that this AI Avatar technology will help improve digital literacy and make mobility services more accessible to all.”

And the main researcher Daewoong Choi also said: “We plan to further advance our generative AI technology so that AI avatars can naturally speak and move like real people. In the future, we aim for interactions that can replace a little human work for order, consultation, and even more.”

Technology is currently registered on the ETRI technology transfer site as “a frame for the generation of a photorealistic speaking face”. Researchers will also actively continue technology transfer and marketing strategies in various industries.